Nanite technology in Lamberyard engine

Question

Today came the demo of the new engine Unreal Engine 5 in it we were shown the use of high-poly cinematic assets with billions of polygons in the scene. I would like to be surprised and say that this is a breakthrough in the gaming industry! I don’t know how cool this is implemented, but I’ll agree on what to do low-poly models, bake the normal card, etc. it already seems to be a thing of the past. In this regard, I have a question. Can we wait for the introduction of this technology in our amazing engine in the near future, in a year or two? 
You must add this to our engine!
and I would also like to say that something needs to be done with global lighting ... now it is of little use
That would be great

Answer

But we ultimately need a COMMUNITY DRIVEN PROJECT (and not an Amazon led project).

At the time (6 years ago), CryEngine 3.8.1 was a "beast" (for its era/time), but I'm afraid that those days are gone (dwindling off into the sunset), and it's an extremely old/outdated/archaic mess that was NEVER meant for online games (and multiplayer games) and it's netcode is a complete nightmare, and it was an "afterthought" and not something that was developed (from the start) when CryEngine was envisioned.

Unreal Engine is much better (in that sense), and the code base is cleaner (in my opinion) and it's much easier to read, much cleaner, and much easier to get started in (and to use).  The installer is great, the build process is extremely fast and easy.  The Epic Games Launcher is a model that should be followed by ALL game engines!  Really great!

Lumberyard / CryEngine are still a big huge mess.  Crytek has been making some progress (with CryEngine 5) and tutorials / content has always been much better than Lumberyard, and even the game samples (as well as Crysis Remastered game assets) in CryEngine 5.7 really "show off" the capabilities of the engine and are much better than what Lumberyard has/showcases.

But Unreal Engine definitely "takes the cake" when it comes to trailers, and game samples.

Just look at the Raytracing (and experimental lighting and rendering techniques) that were developed in Unreal Engine 4 for this demo here:
https://www.youtube.com/watch?v=NdE7msb0xSw

I love how Epic Games is working extremely hard to "blur the lines" between cinematics, ArchViz, and game development.

I truly believe that this is the direction that Lumberyard should be focused on!

ONE ENGINE that can be used for cinematics and movies (creating photorealistic movies, similar to Unreal Engine) and photorealistic ArchViz (that looks movie quality), and also game development (using cinematic quality game assets like Quixel megascans, etc.).

Epic Games is working on it (in break neck speeds) and Unreal Engine 5 will answer most developers prayers.  But unfortunately I don't see Amazon LISTENING, and I blame the "Failed leadership" at Amazon for NOT LISTENING, and NOT UNDERSTANDING the market, and NOT UNDERSTANDING what game developers, instructors, teachers, universities, students, and indie developers need and want!

It's NOT just games!  It's about ArchViz, Cinematics/movies, and cinematic-quality games!

In the near future, we will have "interactive movies" (cinematic quality movies, with interactive plots, where you can choose which direction the movie/game/interactive movie takes).

But Amazon is NOT a "forward thinking" company, and their FAILED LEADERSHIP is a complete mess, and Amazon is not focused on the "future" and can't seem to even get Lumberyard straightened out, and help us (as a community) so we can have a stable feature-rich engine (under a nice permissive MIT-license) that we can ALL use to develop content.

Amazon will benefit from it's AWS services, and Twitch tie-ins.  As well as a possible Amazon Games Launcher (similar to Epic Games) and also a nice Amazon Marketplace (similar to the existing Amazon marketplace) where you can purchase Apps/games (for Amazon FireOS / FireHD tablets, etc.) and hopefully in the future games will be available directly on Amazon for purchase, and possibly a handful of games included (for free) as part of an Amazon Prime membership.

[quote="JohnVanDerBurg, post:11, topic:8609"]
I think we have some good changes coming up that haven’t been announced yet though. This should help LY improve in the future.
[/quote]

I hope so.  It's just extremely frustrating to be waiting 6 years, and still using an extremely old outdated engine, and we still have NOTHING "production ready".

It's similar to saying "Hey, let's use valve GoldSrc engine to develop a modern AAA-quality game".  That is laughable at best (in today's modern multi-core and multi-threaded and multi-gpu era).

NOBODY would do such a thing, and nobody wants 1996 or 1998 graphics, or a slow garbage engine (not optimized for modern hardware).

It was great (for it's era) but it's old, archaic and just not suitable for a modern AAA-quality game.

Lumberyard is really in bad shape right now.  Crytek took the money (from Chris Roberts / CIG, as well as the money from Amazon for the Lumberyard license) and used that money to invest heavily into the development of CryEngine 5.

CryEngine 5 has everything that CryEngine 3.8.1 doesn't have.

Crytek worked on adding many of the features that we had been demanding (for years).  But when you follow the two different forks (what Crytek did from CryEngine 3.8.1 to CryEngine 5.7) versus what Amazon did (with CryEngine 3.8.1 fork to what Lumberyard is currently, which is a big mess).

You can see two very different companies, ONE company has an "expertise" with game engine development (Crytek) and another that doesn't (Amazon).

You can compare the two, and see that CryEngine 5.7 definitely "wins" when it comes to "cleaning up the engine" and game samples, documentation, features, and feature-parity, as well as tools.

Lumberyard looks as though it sat "idle" for 6+ years, and still remains very much a CryEngine 3.8.1 engine.

CIG's engine (based on Lumberyard) is like a "blend" between a modern CryEngine 5 (which it's not) but it is far bettter than Lumberyard, and has many of the new features, optimizations, and things that CryEngine 3.8.1 needed (as well as optimizations for multithreading, 64-bit precision, etc.) and it's a complete shame that none of this was added back into Lumberyard core.

It would still make Lumberyard core a complete mess, but at least you could sit and look at everything (as a developer) and then work on "refactoring" everything into a nice new clean engine, with clean source code and make everything into a new highly optimized clean "Lumberyard 2.0" that is 100% new codebase that is free from Crytek and doesn't share any of the same codebase.

Then we can license Lumberyard 2.0 under a permissive MIT license, and move forward as a community!  Lumberyard has just been collecting dust for 6+ years, and we still don't have a working Production-ready engine!

Chris Roberts' CIG/Star Citizen has been flushing $4 Million per month down the drain (for 8+ years now?) and what do we have to show for it?  An early alpha?  An extremely buggy unplayable early alpha?  The most expensive alpha in world history/existence?

My god, who dumps $300+MIL on an early alpha, and we still are no where near close to a production quality game engine (that is stable and bug free, and production ready).  8 years in, and five offices, with well over 500+ employees (I believe it might even exceed 1,000 employees if you include former Foundry 42, etc.) I'm not even sure at this point, but it's a LOT LOT LOT OF MONEY, and to see that amount of money getting flushed down the toilet (each month), and it's an extremely "ambitious" goal and I have been an original backer (back when Chris announced his original kickstarter) and I was excited when he chose CryEngine (back in the day) but nowadays (in hindsight) and 8+ years later (in today's modern era), I just don't know if I would have made that same exact choice again.

Seeing at how terrible Crytek is/was (as a company) and they didn't support their engine (at the time) and they were nearly bankrupt and unable to play their employees, and Crytek was collapsing (as a company) and had it not been for Chris Roberts' money (and Amazon's money) there probably would be no Crytek today.

Instead of being appreciative (and thankful), Crytek is a greedy company that turned around and sued Chris Roberts (a licensee/developer) and that left a "bad taste" in EVERYONE's mouth, and most Star Citizen fans were EXTREMELY UPSET (and very angry at Crytek) and most developers "dropped Lumberyard" because there was so much legal "uncertainty" for several years as this litigation dragged out in the courts, and everyone has been hoping for a "CIG WIN" (to beat Crytek in court and set a precedence in court).

But due to that legal uncertainty, everyone has "shied away" from Lumberyard, and it's like the old UNIX lawsuits back in the day (which is why Linus Torvolds started Linux).

Most developers have completely left and abandoned CryEngine and Lumberyard and moved towards Unreal Engine and Unity.

Unity is the number one engine (For mobile) and Unreal Engine is the number one engine for cinematics, ArchViz, and cinematic quality games.

The ONLY game that is cinematic quality (that isn't using a proprietary engine) would be Star Citizen (which is using a heavily modified Lumberyard/CryEngine 3.8.1), but nobody has $300+ Million (and 8+ years) to dump into engine mods/development.

We had hoped that something "better" would come of this partnership between CIG and Amazon (hoping that Chris Roberts would "contribute" much of the modifications back to Lumberyard core), and Amazon would maintain a Lumberyard-SC fork (for Star Citizen / Star Engine) and that way it would share the same codebase as current-gen Lumberyard, but would have all the modifications necessary for Star Citizen.  (and make this codebase public on GitHub)  That way ALL Developers can see it, learn from it, and use it (if they want) and have a Lumberyard build (Lumberyard-SC build) that is identical to the exact same core/engine that CIG is using.

This way more and more developers can learn the engine, the code base, and tools/editors, and we can create game mods, and also teach students (on how to use the engine) and hopefully make Lumberyard more popular.

[quote="JohnVanDerBurg, post:11, topic:8609"]
We are on your side and many of us know very well that these issues exist and are fighting to get those resolved. It is cool to see you post all this information and we appreciate it a lot.
[/quote]

No problem.  It's been a very long 30+ years, and I've been following game development and game engine development for a very long time (back to early Commodore Vic-20, and even TI-16 days).

But we have come a long way, and having worked for the government/military (for many years) and using high-end hardware (SGI Octane / SGI Onyx) back in the day, and then looking at garbage civilian hardware (during that era) and now 20 to 30+ years later, I believe we are finally reaching a point where real-time ray tracing is possible, and real-time rendering (cinematic quality) is possible, and civilian hardware is finally reaching the point where cinematic-quality games can (and will) finally become a reality fairly soon.

But the biggest problem is the game engines!  It's NOT hardware, hardware can already do it, but it's the current state of game engines.

Unreal Engine still needs quite a bit of work, and Lumberyard (which I was extremely hopeful for 6+ years ago, has fallen to the bottom of the list, and seems to be almost beyond repair).  Unity and Unreal Engine are both making great strides, Frostbite is decent (albeit proprietary) but as far as general purpose game engines, I believe Unreal Engine 5 will probably hold the crown for a few more years (for ArchViz, Cinematics, and general purpose games/AAA-production quality Indie developer games and small studio games).  Unity will continue to hold the crown for low-end development/mobile games, simply due to it's smaller build size and Unity Ads (and wide variety of supported mobile platforms and consoles, especially for 2D games).

But Lumberyard just hasn't found it's place, and 6 years ago CryEngine was better than Unreal Engine (at the time) and Unity was nothing but crap (for low-end mobile) during that era.  But as game engines have evolved, Unreal Engine has been the leader in Cinematic-quality games, ArchViz, and general purpose (for PC and console).

Unity is still the leader when it comes to lower-end games (2D and lower-end 3D) for mobile and console.

But Lumberyard which was based on CryEngine 3.8.1 should have been for top-quality AAA games (i.e. Crysis Remastered, Far Cry 5, etc.) but unfortunately nothing has come of it (other than Star Citizen, which is still an early alpha, and nobody is even sure if it will EVER come out of Alpha, or ever reach the stages of a production-ready game).  Star Citizen looks fairly decent, but the backend servers are a mess, and constant server crashes, and it's impossible to play online (with lots of lag, and servers slowing down or crashing, or trying to talk without voice/chat going down, and being forced to use Discord for voice chat, because CIG's servers never seem to work property and chat seems to go down all the time).  Plus lots of bugs and crashes, and it might be another decade before Star Citizen is ever "stable enough" to actually play.  (It's been 8+ years already, and we're still no where near reaching a stable production ready playable game).

[quote="JohnVanDerBurg, post:11, topic:8609"]
We are on your side and many of us know very well that these issues exist and are fighting to get those resolved. It is cool to see you post all this information and we appreciate it a lot. Please just keep in mind that things aren’t as simple as saying we don’t know anything :slight_smile: I also used to think that way until seeing how it works at large organizations that have to work with a lot of constraints.
[/quote]

Unfortunately, that's the problem with "large organizations".  I've seen very small "lean" teams get far more done (much faster) than very large organizations (that are "slow moving" and waste the majority of their time with "too many project managers" and "too many chiefs" and "not enough indians" to get the job done, or too much time wasted in daily/weekly meetings, and devs are just too bogged down, and unable to get their work done.

I believe in "less overhead" and smaller/leaner teams that can get more done.  Large organizations are problematic and very "slow moving" and this really slows down the process.

It often takes weeks or months for "leadership" to make a decision (often a WRONG ONE) and you just have the "wrong people" working at the top, and it causes a whole project to move slowly (and ultimately fail).

Spending decades in the military and working  for the government, I understand "big government" and "large projects" and how "mismanaged projects" will ALWAYS FAIL (no matter how much money and time you waste and throw at it).

Ultimately, it's better to look at a small team, that is very lean, that is making things happen.  Big organizations have lots of overhead, and when you have "too many chiefs" (at the upper echelons) it just makes the projects a nightmare, and often times the leadership takes the project in the wrong direction and it is NOT what consumers (or the community) wants or needs, but leadership is NOT LISTENING, and then ultimately the company/organization FAILS.  (As so often is the case).

I personally think Amazon's mess is a big mess.  I don't blame the developers, I blame the leaders (at the top).

As for Lumberyard, I really truly hope that you guys can do something with this engine (extremely soon) and release a nice stable clean re-write (Lumberyard 2.0) that has a nice MIT-license (and is free from all old legacy crytek code) and that we (in the community) could finally help take over this project as a Community-driven project (albeit financial help/sponsorship from larger organizations like Amazon, Google, AMD, Intel, NVIDIA, etc.) that are willing to sponsor developers (full time).

But I believe this is the only way to ultimately move forward, as developing the "best" engine is going to take resources, and a strong community (very similar to the Linux community).

But I believe the MIT-license (just like GoDot is using) is the absolute best way to go!

If you look at how many organizations have jumped on the Go Dot Engine, it's a fast growing community.  (Personally, I'm not a fan... not yet at least).  But I am hoping that Amazon can "see the light" and hopefully release a Lumberyard 2.0 release (under a MIT license) that we can ALL use, and that we can ALL develop and work on.  (Not just Amazon employees).

It took several years, just to get Amazon to finally use GitHub (and publicly share the source) but even so, we still have NO ROADMAP and in 6+ years have no real sense of "direction" as to where Lumberyard is going (or headed) and it's beginning to seem extremely "hopeless".

[quote="JohnVanDerBurg, post:11, topic:8609"]
That said though, I think LY is making some good moves now, which you should see later on. We can’t change the past, but I’m more optimistic about the future.
[/quote]

Well, I hope so.  I have heard this many times before (in the past) and 6+ years later, we're still in the same exact situation that we are currently in.

I'm just afraid that Amazon might be "too big" of an organization (too fat, too slow moving) and unwilling to make this a Community Project (with a public roadmap) and these are major problems, and only causes the development to move at a snails pace (which is what Amazon has been doing for 6+ years) and just like you said, with "limited resources" and moving a "snails pace" the project is completely doomed, and will ULTIMATELY FAIL.  (It's inevitable).

The only person that doesn't see it, is Jeff Bezos'.  Because if he saw what I see, he would have "pulled the plug" a LONG LONG LONG TIME AGO.

I want to see Lumberyard succeed, but that is going to take a community effort.  Linus Torvolds didn't develop Linux alone.  Sure Linus worked on the core (and still works on the core kernel) but it takes thousands and thousands of developers (many many years) and it's going to take an "Army" to do what needs to be done, and unfortunately Amazon doesn't understand "leading an Army" and it's best to just refactor the engine, release it as a Lumberyard 2.0 (as quickly as possible) and release it to the community, and then continue to sponsor Amazon Developers (just like Google sponsors google-employee developers that work on open source projects like Chromium or Android, etc.) and Amazon can be a "good corporate citizen" and release Lumberyard to the Community as Lumberyard 2.0 (under an MIT license) and then we (as a community) can help drive the adoption and development of the engine (without Amazon leadership getting in the way or slowing things down, or making it impossible to do).

6+ years is a very long time, and I have a hard time believing that Lumberyard is still in the shape that it is in, in 6+ years and if Amazon doesn't do something FAST... then Lumberyard will be a "lost cause" (similar to developers using or creating mods using GoldSrc!).

https://en.wikipedia.org/wiki/GoldSrc

It's "dead end".

That old CryEngine 3.8.1 bones/skeleton probably doesn't have much life left in it.

Ubisoft has abandoned CryEngine (after their heavily modified CryEngine version for Dunia and Dunia 2).  I don't know of anyone else even messing with CryEngine 3.8.1 (other than Lumberyard and Chris Roberts' CIG), and I'm just afraid that Chris is dumping a LOT LOT LOT of money into something that is extremely old, and is a complete mess, and he probably could have had Ubisoft Montreal create a completely new game engine (from scratch) in LESS TIME, and FAR CHEAPER than what he has spent/wasted on CryEngine (and time that he has dumped into Lumberyard modifications).

If that was shared with the community, then it would at least be beneficial to the community, and we could possibly sift through all that current "mess" and work on refactoring everything, and create a completely new engine (from scratch) that is clean, highly optimized, multi-threaded, and something that can compile/build quickly, and something that can beat/whip Unreal Engine.

But unfortunately, it seems like Amazon is holding their cards tight to their chest, and CIG seems to be doing the same, and nobody is working together (as a community) and inevitably nothing is really happening (over the past 6-8+ years) and it's turning into a big huge mess.

We need OPENNESS and COMMUNITY COLLABORATION.  Which is the exact opposite of what Amazon and CIG are currently doing.  I'm hoping that this can change.  I really do.

I want to see Lumberyard succeed.  I want to see small studios like CIG succeed.  I want a great engine that anyone and everyone can use to create AAA-quality games (like Star Citizen) but being able to do it with a small 11-20 person development studio (just using assets freely available/shared in the marketplace).

If you look at the old Crytek CrySDK it was a "great start" for learning how to use CryEngine, and also a nice source of AAA-quality assets in a AAA-quality game/SDK that anyone could use to modify and make their own game.  The only issue with CryEngine was the source code at the time (not being available, and then years later Crytek finally made it available, but only after Unreal Engine released their complete source code) and now we are finally seeing some openness, but Lumberyard is still a mess because it's still based on that old legacy nightmare of a CryEngine 3.8.1 codebase.

I liked what Amazon was doing (and had great hopes for the Lumberyard project) but Amazon is just so slow moving that by the time anything happens (or comes of this project) it will be decades later, and by then, you'll be using something called "history".  An old archaic engine that resembles Valve's GoldSrc engine!

If Amazon is going to "make something happen" they NEED TO DO IT NOW.  If not, then just throw in the towel and just shut down Lumberyard (which I would hate to see happen, especially to the employees and developers).  But I believe the best (and ONLY) way to save Lumberyard, is to refactor this engine (as quickly as humanly possible) and release a Lumberyard 2.0 (completely refactored and completely free of any old legacy crytek code) and release Lumberyard 2.0 under a MIT license, to the public/community, and then let us (the community) take the reigns and let us work together with NVIDIA, AMD, Intel, Google, Apple, Microsoft and Amazon and together we can draw in some talent (and corporate sponsorship) and really start making progress on the game engine development.

Amazon is NOT EPIC GAMES.

Until Amazon can release a couple of AAA-Quality "Fortnite" or "PUBG" games (under their belt) and bringing in hundreds of millions of dollars in annual profits (from those games) and can sustain their development (and business model) and just go out and buy companies (like Quixel) and give everything away for free (like Epic games does).  I just don't see Amazon doing that (or being able to sustain it), and I'm just afraid that Lumberyard is going to ultimately fail (unless they agree to turn Lumberyard over to the community).

It's just "too big" for Amazon, and we need a grassroots effort to make this happen (and that is something that Amazon has been against and fighting against for years).

Amazon wants to hold everything close to their chest, and doesn't want a public roadmap and doesn't want to "let go" of Lumberyard, and let the community lead the development of Lumberyard.

Which I firmly believe is going to ultimately lead to the "death of Lumberyard".

Mark my words, you'll look back at this post in 5 years, and I'm fairly certain that is what is going to happen, because of the bad decisions being made at the "Leadership level" at Amazon.

This is NOT a project that Amazon can lead.  I think it was a very ambitious effort, I was extremely excited about it (as was most of the community) and it was a great concept/idea (in principle) but it's most likely going to fail a miserable death (in the very short term) because of 6+ years of BAD MANAGEMENT (and bad leadership) at Amazon.

I'm not blaming the developers, I'm sure you have some wonderfully talented individuals that have spent 6+ years learning the engine (at Jeff Bezos' expense) but I just don't see enough happening (in those 6 years) to justify the "expense" and I don't see the "fruit" coming from the tree.

A tree that doesn't bare any fruit, is fairly useless (and we call it "fire wood").  I'm hoping that we can start to see something come of this project (extremely soon) and I'm hoping that Amazon leadership can "wake up" and "see the light" and truly turn this project around (extremely fast) and we need to start seeing "results" (of all the money that Jeff Bezos' wasted on this project), but I do want to THANK Jeff Bezos' (and his deep pockets and generosity for even attempting something like this), but as much as I love Jeff Bezos', he just isn't the "right man" for this job.  He has the deep pockets, but the leaders that Jeff has chosen (to lead this project), are just not steering this project in the right direction (and doing what needs to be done) and doing it as quickly as it needs to be done, and doing it efficiently.

It would have been FAR CHEAPER to just sub-contract the work to Ubisoft Montreal, and have a production ready game engine whipped out in 2 years at Ubisoft Montreal (Ubisoft Montreal has a LOT LOT LOT of experience with older CryEngine versions, and heavy modifications) and Ubisoft is a known LEADER when it comes to AAA-quality games, online game play, and releasing on schedule.

Plus with the money that Jeff Bezos' has already wasted, he could have already had 2-3 AAA-quality games developed (and published) over at Ubisoft (and already released by now) and would at least have his own game engine AND 2-3 AAA quality games (bringing in revenue) and also he could give those game assets away (as he would own them) as part of Lumberyard (similar to how Epic Games gives away $100+ Million of AAA-quality game assets, including [Paragon ($17M) assets](https://www.unrealengine.com/en-US/paragon),  and [Infinity Blade ($4M) assets](https://www.unrealengine.com/en-US/blog/epic-releases-4-million-of-infinity-blade-assets-and-renews-featured-free-unreal-engine-marketplace-content) and Featured Free Marketplace content (every month) as well as the complete Megascans / Quixel AAA-quality (actually Cinematic quality) assets and textures.

See here:
https://help.quixel.com/hc/en-us/sections/360000977797-Unlimited-Access-for-Unreal-Engine

See here:
https://quixel.com/blog/2019/11/12/quixel-joins-forces-with-epic-games

With these big bold moves by Epic Games, and see what is coming from Epic Games (later this year), it really makes Amazon look like little kids splashing in a puddle and unable to even keep their heads above the water.

The ONLY thing that could possibly save this whole Amazon Lumberyard project, is the COMMUNITY.

Amazon has to make a choice whether they want to keep holding their cards (close to their chest) and ride this whole project into the ground (like they have been doing for the past 6+ years) into a BIG COLOSSAL DOOMED FAILURE.

I would hate to see all that money, time, and energy wasted... but that seems to be the direction that Amazon Leadership is headed (and has been headed for the past 6+ years).

It's very short-sighted, and a big waste of money.  It would be sad to watch all of this just go up in smoke (for no reason) only because Amazon is too "bull headed" to let go of the reigns, and turn this project over to the community.

Even Google, Canonical/Ubuntu, Red Hat, Apache, have ALL learned the strengths of community development.

Look at Linux as a whole, and look at how many millions of lines of code have been written (predominately by the community).  Linus Torvolds has only written a very very very very small tiny portion of that codebase, but that is the "strength in numbers" (that Linux has leveraged) and this is something that Amazon just doesn't seem to understand.  We need a COMMUNITY LED and COMMUNITY DRIVEN PROJECT (for Lumberyard).  Lumberyard should be a COMMUNITY LED project (not a corporate led project).  That (in my opinion) is the BIGGEST PROBLEM with Lumberyard!

Answer

Mark,

I feel your frustration, and I fully understand it.
I can assure you though we are up to date about all this. I've seen most of the videos you posted already for example, and I am not even a graphics engineer but an animation engineer. Just please keep in mind that we also have limited resources (yes even at Amazon). It is not exactly fair to compare us to Unreal for example.

Again, I completely understand your frustrations and fully agree with you personally on some of the things, they are spot on, but please keep in mind things aren't as easy as it seems. The devs we have are highly skilled and are up to date, I can assure you that. But we don't always get to work on what we wish to work on as there are other things going on, and I know many of the issues you mention also frustrate us. For example the installer that you mentioned as well as the build process.

Recently an effort has been made to fix a lot of issues with the current installer. This should be released soon. It is a step in the right direction. We are getting rid of WAF as well. This all took way too long for very sure. I personally think Amazon took on too much work, like supporting internal game teams, other select teams, and a community, all with limited resources, with most focus on the internal projects, which made it feel like the community isn't being listened to as much as they should. I think we can all agree on this, but I think some good changes are being made to address this now though.

Lumberyard has a long way to go, and we all know that. I think a lot of mistakes were made in the past, we all agree on that. In my opinion those were mostly not mistakes by the devs though. I think we have some good changes coming up that haven't been announced yet though. This should help LY improve in the future.

We are on your side and many of us know very well that these issues exist and are fighting to get those resolved. It is cool to see you post all this information and we appreciate it a lot. Please just keep in mind that things aren't as simple as saying we don't know anything :) I also used to think that way until seeing how it works at large organizations that have to work with a lot of constraints.

That said though, I think LY is making some good moves now, which you should see later on. We can't change the past, but I'm more optimistic about the future.

Cheers,
John

Answer

[quote="JohnVanDerBurg, post:5, topic:8609"]
There has been some work in this area before. I think the main engineer of Nanite also mentioned this in some post he did.
[/quote]

Epic Games, NVIDIA and Microsoft have been saying for several years now (since 2015-2016) that Mesh Shading is a key technique in the future of graphics processing.  Unfortunately, it seems that Amazon Lumberyard Dev Team just hasn't been paying attention (or listening) for 5+ years now.

You guys should really attend a SIGGRAPH Conference!  LOL!  (Clearly you haven't been paying attention for the past 2003-2019 SIGGRAPH Conferences)

Full Presentation by Manuel Kraemer & Rahul Sathe can be found here:
https://on-demand.gputechconf.com/siggraph/2019/video/sig944-applications-of-mesh-shading-with-dx12/

This is what "Nanite" is based on (something extremely similar/identical), dynamic geometry and dynamic LODs.  This is the "how" of Nanite.  You can reach out to Rahul Sathe and Manuel Kraemer, if you need more information, and possibly to work together on a Amazon Lumberyard Tech Demo (implementing this same technology into Lumberyard) and please make the demo project (and game engine source code) available to everyone, and hopefully this can be added to Lumberyard fairly quickly.  NVIDIA did this modification in 8 weeks (for Unreal Engine).  Hopefully we can see this (as an 'experimental' feature) in Lumberyard very soon, so we can learn from it, and begin developing games to use/support it in Lumberyard (especially for next-gen consoles).

It would also be great if Amazon could license [Oodle Kraken](http://www.radgametools.com/oodlekraken.htm) and [Oodle Leviathan](http://www.radgametools.com/oodleleviathan.htm), and [Oodle Network Packet Compressor](http://www.radgametools.com/oodlenetwork.htm) and include this (just like SpeedTree or Audiokinetic Wwise Audio) with Lumberyard, so it will be part of Lumberyard (and we will all have licenses to use it) when creating games with Lumberyard.

You can learn more about Oodle Compressors (by Rad Game Tools) here:
http://www.radgametools.com/oodlecompressors.htm
- But please include licenses and support for [Leviathan](http://www.radgametools.com/oodleleviathan.htm), [Kraken](http://www.radgametools.com/oodlekraken.htm), and [Network Compression](http://www.radgametools.com/oodlenetwork.htm) with and inside Lumberyard!
 
Oodle is easy to use, cross platform, and super fast!

Especially for those long co-op Survival Missions and MMOs so data sent at 20% of its actual size (with 80% reduction in file size with Oodle compression), between the file compression and network compression!

It's huge for large MMOs, on servers with a large number of connections (like MMOs), and results in a huge memory savings, and increases the number of players (per server).  Please make this happen in Lumberyard!  (and please include the licenses).

Please get this done (as a top priority) in Lumberyard!  Please add the "Nanite" (Mesh Shading with dynamic geometry and dynamic LODs) and please add the [Oodle Leviathan](http://www.radgametools.com/oodleleviathan.htm), [Oodle Kraken](http://www.radgametools.com/oodlekraken.htm), and [Oodle Network Compression](http://www.radgametools.com/oodlenetwork.htm) to Lumberyard! (Especially for multiplayer games).  Also please increase the map size (to unlimited size maps) using streaming.  Also please implement (and refactor) much of the same technology that CIG / Star Citizen is using (for interplanetary travel and planet size maps) and please "re-factor" this technology and optimize it, and make it part of Lumberyard core, so that ALL Indie developers can use this same exact technology in Lumberyard!  (Plus Amazon Lumberyard can further develop and optimize it for future releases).  Also after adding these features (so we can use cinematic quality textures and geometry/assets) please be sure to license the Quixel Megascans assets and textures for ALL Lumberyard users, for use in Lumberyard (just like Epic Games has done for Unreal Engine users).  This should be a NUMBER ONE PRIORITY for the Amazon Lumberyard Team!  Please get this done!  Please and thank-you!

Answer

[quote="WashedUpStudios, post:2, topic:8609, full:true"]
Hey @REDACTEDUSER
[/quote]

You don't seem to know ANYTHING about ANYTHING concerning game engines or consoles!  First off, the Unreal Engine 5 demo using Nanite is using cinematic quality assets, and yes it is running on a CONSOLE!  (the game demo is actually running on a PlayStation 5).  It is NOT a cinematic demo, it is an actual GAME DEMO for the new PlayStation 5 using Unreal Engine 5 with Nanite!  The PS5 and XBOX Series X console hardware (coming this fall) is equivalent of a AMD Ryzen 7 3700X and a GPU (RDNA2 based) similar in performance to the Navi-powered RX 5700 XT.

Unreal Engine 4.25 and Unreal Engine 5 are designed specifically for high-end gaming, and supports the very latest multi-threaded PCs and game consoles (something that Lumberyard does not do).  We would NEED something identical to Nanite for Lumberyard, but NOBODY at Lumberyard seems to know (or even understand) what EPIC GAMES is even doing, and this was already demonstrated years ago (by NVIDIA) with the Asteroids demo.  But Epic Games is now incorporating this same Virtual Geometry (with automatic dynamic LODs and automatic dynamic culling) technology (and calling it "Nanite").

Unreal Engine 4.25 already has Virtualized Texturing, and now they are adding Virtualized Geometry (Nanite) to Unreal Engine 5!

Amazon's Lumberyard team seems completely clueless, and always about 5-10 years behind EPIC GAMES Unreal Engine (or even Crytek CryEngine 5) when it comes to development or features.  Lumberyard doesn't even have support for PlayStation 5 or Xbox Series X consoles (even though developer kits have been available for quite some time now, and EPIC Games Unreal Engine 4.25 already supports both the PlayStation 5 and Xbox Series X consoles that are coming this fall).

Lumberyard team has fallen extremely far behind, and are still using an old archaic Cryengine 3.8.1 engine, and they don't seem to know (or understand) HOW to update it, make it current, or develop a real game engine (on par with CryEngine 5.7 or Unreal Engine 4.25 or Unreal Engine 5).

Lumberyard team has really fallen extremely far behind and is at least 8+ years behind in technology when it comes to comparing Lumberyard to CryEngine 5 (features and capabilities) and comparing Lumberyard to Unreal Engine 4.25 or Unreal Engine 5 (features and capabilities).

In 5+ years we can't even get large maps in Lumberyard!  We can't even get large map streaming!  We will NEVER see anything like Nanite in Lumberyard, because Amazon doesn't have the skills (or game engine developers) capable of even creating a decent engine (like EPIC GAMES or Crytek or even CIG).  Amazon refuses to hire good game engine developers, and instead hires "C-grade" or "D-grade" developers (and possibly even "F-grade") that can't even create a game engine on the same tier as CRYSIS REMASTERED!  Crytek can do this, Epic Games can do this.  Amazon Lumberyard team can not.  In 5+ years, we have seen absolutely NOTHING when it comes to feature requests (and game engine modifications) and all we have seen is a new GUI interface, that still isn't as good as CryEngine 5.7 or Unreal Engine 4.25/UE5, and the Amazon Lumberyard team doesn't have the "technical ability" to create an amazing game engine like Unreal Engine 5 (and Nanite).

Epic Games and Crytek are transforming the industry.   Amazon Lumberyard Team is just using an old archaic 8+ year old Crysis 3 (CryEngine 3.8.1) build, that is a complete mess, hasn't been upgraded, doesn't have the same features and capabilities as CryEngine 5, and is no where near EVER having the capabilities of a modern game engine like Unreal Engine 5 with Nanite!

Answer

[quote="JohnVanDerBurg, post:11, topic:8609"]
The devs we have are highly skilled and are up to date, I can assure you that.
[/quote]

I don't doubt you.  Having had access to the Lumberyard Engine source code for 6+ years, I would certainly hope (and expect) that you are familiar with it, and the problems with it, and I would have hoped that it would have been completely re-written by now.  😃

[quote="JohnVanDerBurg, post:11, topic:8609"]
But we don’t always get to work on what we wish to work on as there are other things going on, and I know many of the issues you mention also frustrate us. For example the installer that you mentioned as well as the build process.
[/quote]

**THIS IS THE BIGGEST PROBLEM!!**

Personally, I believe the engine needs to be re-written (From scratch) and completely free of any and all former Crytek "CryEngine" code, and be a completely new engine (that we can license under a permissive MIT license).

From this point, I personally would create a large "bullet list" of features that we need to add/accomplish in the core engine.  Then next I would sit down and talk to ALL THE DEVELOPERS, ask them if there is any features that they believe we need to add (and then add those features to the list as well).  [I would also make Chris Roberts / CIG a part of this process as well, and try to work together with CIG / Foundry 42]

Then we prioritize the features (in a top-down manner) based on which are the easiest/fastest to get knocked out, and also which features are the most requested (and most desperately needed) and also which features are most needed to get the engine "production ready" (in the fastest/quickest time possible).

Personally, I think CIG is "too busy" making ships, and building planets, instead of focusing their efforts on the core engine development!  This should be done FIRST!  Let's sit down, discuss ALL the features the engine needs, and then create a "To Do List" and then work on these features for the core engine!

Next we sit down with ALL THE DEVELOPERS and discuss their skills/talents, and then have a large discussion (on who wants to do what).  Let the developers pick and choose what "feature" that they want to work on (and implement) and let the developers do what they enjoy (and being part of a project that they ENJOY working on and ENJOY doing).

Let's make this FUN!

Then the features that are not being worked on (by Amazon Core Lumberyard team) are features that we can post (publicly) on a "help wanted" board, and reach out to Universities, students, and others (summer interns) that want to help tackle some of the smaller features, and work on getting those hammered out, and submitted to "core" (so core can review the code, make sure everything is clean, looks good, and then added to the core codebase in an upcoming/next release).

This way we have "production ready" engine (i.e. 2.0) and "experimental features" (i.e. 2.3, 2.4, 2.5, 2.6) that are in the pipeline (and ANYONE can use the "experimental" builds/features or build from source) for the 2.3/2.4/2.5/2.6 test/experimental branches.

But at least students that want to focus primarily on a "stable" game engine can use the 2.0 core (and being developing a game/content), and the core engine team can continue to work on the core engine (and features) for future 2.3/2.4/2.5/2.6 releases.

As features become more tested, and stabilized (and finally reach "production ready" status) they can be pushed out in a 2.1 or 2.2 or 2.3 release (as those releases become "production ready").

Always have a seamless "upgrade path" for developers, so content developers that are using 2.0 can simply download/update to 2.1 or 2.2 and can import their existing maps/content (and everything will be updated to the latest 2.2 compatibility).

This way content developers and artists can just focus on creating their content/games, and core engine developers can focus on the core engine development (and features / feature requests).

Then as more and more content developers (and schools/universities/students) accept "Lumberyard 2.0" as their "engine of choice" hopefully we can begin building up a large content base of "MIT licensed" content (game assets, textures, models, etc.) that are freely available (in the Marketplace) for Indie developers and artists and content creators to use.

We need to make this as "easy" and "turn key" as possible.  Remove ALL the barriers to creating a game (for small 1-11 person teams) and so small tiny Indie developers and small student teams can church out a AAA-quality game in 1-2 years (with a good core engine, and great AAA-quality assets available for free).

Unfortunately right now, the exact opposite is happening, and everyone is working in many different directions (creating highly modified engines) with nothing ever making it's way back to Lumberyard Core, and everyone is so far away from Lumberyard Core right now, that the divergence is making it impossible to ever "come back" to core, and core still remains broken and not updated!  This is a major problem!

[quote="JohnVanDerBurg, post:11, topic:8609"]
Recently an effort has been made to fix a lot of issues with the current installer. This should be released soon. It is a step in the right direction. We are getting rid of WAF as well.
[/quote]

This is EXTREMELY GOOD NEWS!  This is something that we have been SCREAMING about for years (since day one) and over 6+ years, and I'm glad to hear that this is coming!  (hopefully very soon).  But personally, I hate the current installer and think it should be abolished, and we need something much nicer (completely new) identical to the Epic Games Installer/Launcher!  That is what an installer and launcher (and updater) should look like!  Very nice, easy to use, quick build times, no build errors, and also all your add-ons show whether they are "compatible" with your version of the engine (from within the launcher/library).

[quote="JohnVanDerBurg, post:11, topic:8609"]
This all took way too long for very sure. I personally think Amazon took on too much work, like supporting internal game teams, other select teams, and a community, all with limited resources, with most focus on the internal projects, which made it feel like the community isn’t being listened to as much as they should.
[/quote]

I completely agree with you here, and thank-you for your honesty!  It's "refreshing" to hear an Amazon Lumberyard developer actually ADMIT to this.

As a member of the community, I can assure you that we have been waiting patiently (and fighting LONG AND HARD) for many of these things, and Amazon has been very "slow moving" and unfortunately tired to "do too much" all at the same time (with an engine that just isn't ready) and this is like trying to drive a train (when the tracks haven't even been laid yet).

In my opinion, **GET THE TRACKS LAID DOWN FIRST**, get the engine production ready, get the features added directly into core, and then once that is done, then you can take some of your community members (and small Indie teams that have been making the absolute best Game samples, and winning the "quarterly competitions" and look at the "Community Showcase" projects, and then you can see WHO IN THE COMMUNITY is "worth their salt" (and use these developers as "contractors" to work on various game levels, sample games, or whatever it is that Amazon wants to do with their Amazon Studios offices, but it would be better to use independent Indie artists/developers as "contractors" instead of setting up your own large Amazon Game Studios (and not even having a production ready engine) and then FAILING as a game studio.  It doesn't make sense.

Focus on the core Lumberyard engine FIRST!  I would shut down the Amazon Studios (personally) for now, and just move ALL of those team members over to Amazon Lumberyard Core, and focus on getting the core Lumberyard Engine (core) completely re-written (and completely free of any old Crytek/CryEngine legacy code) and then license the new engine (Lumberyard 2.0) under a permissive MIT license so others can contribute and make the game engine a "Community Project" instead of an Amazon project (with a real public roadmap and real forward-looking vision).  Next work on game samples, but get the engine re-written and done! (FIRST!)

[quote="JohnVanDerBurg, post:11, topic:8609"]
This all took way too long for very sure. I personally think Amazon took on too much work, like supporting internal game teams, other select teams, and a community, all with limited resources, with most focus on the internal projects, which made it feel like the community isn’t being listened to as much as they should. I think we can all agree on this, but I think some good changes are being made to address this now though.
[/quote]

I completely agree with you, and I really REALLY REALLY hope that Amazon is listening, and I hope that some MAJOR CHANGES are coming very soon, and I hope to see a Lumberyard 2.0 release (that is completely 100% new codebase that is 100% completely re-written and free from any old crytek code) and is an engine that we can license under a very permissive MIT license, and then release this "Lumberyard 2.0" (as well as the trademarks) to a Community-led organization, so that we can continue the development of Lumberyard 2.0 (as a community) similar to Android or Linux, but under a nice MIT-license (similar to GoDot).

Then Amazon can continue to "sponsor" it's own core developers (that will remain as core developers and core members of the Community Team) similar to how Google sponsors developers for Android or Chromium open-source projects.

That way Amazon is a being a "good corporate citizen" (to the open source community) and sponsoring developers (Amazon employees) that are members of the Community Driven project, and that way the COMMUNITY (and Core team) can decide what features will be worked on next (priority wise) and we can get this engine "production ready" and to become the ABSOLUTE BEST ENGINE out there!

Something that can BEAT Unreal Engine 4 / Unreal Engine 5, and beat Unity, and beat CryEngine 5.x (or even 6.x) and crush GoDot.

I believe Lumberyard has a LOT OF POTENTIAL (and that is what is so extremely frustrating) but unfortunately SIX YEARS is an EXTREMELY LONG TIME, and as a community we just aren't seeing ANY RESULTS come to fruition.

Personally, I just want this engine to get completely re-written (so it is free of any Crytek incumberances) and that way it is a completely new "Lumberyard 2.0" (that has NOTHING from the old cryEngine) and is a 100% new/clean codebase, and we can license this new "Lumberyard 2.0" under a very permissive MIT license, and let the community work together on this.

That way we can finally have a "Lumberyard 2.0" (that is similar to the GoDot community) and we can finally start making some decisions, and have volunteers dive in, and finally have a road map, forums, and a true community driven project (where universities and students, and companies like NVIDIA, AMD, Intel and Google) can ALL get involved and TOGETHER we can create something great.

When NVIDIA and AMD are sponsoring developers, and Intel is sponsoring developers, and Google is sponsoring developers and Amazon is sponsoring developers, then I would be more than happy to create a "bullet list" of everything that the community wants/needs done, and everything that we need to be "feature parity" with the other engines.

That way developers don't have to worry about "what is the BEST engine".  There will be ONE ANSWER and that would be "Lumberyard 2.0".  I would aim to have ALL the great features that Epic Games / Unreal Engine has, and all the great features that Unity has, and some of the features that GoDot is working on.

Look at everything that is out there, and just put those amazing features on a "To Do List" and slowly create the BEST ENGINE in existence.

This way we have ONE STANDARD ENGINE CORE that everyone can use and share.

Hopefully in the future, this could become the "basis" for other engines (and possibly Epic Games might abandon Unreal Engine 4 and Unreal Engine 5) and then "Unreal Engine 6" could simply be a "fork" of the Lumberyard 3.0 core.

That way Unity 7.0 and Unreal Engine 7.0 and GoDot Engine 7.0, and Lumberyard 7.0 could ALL share the same "core".

Similar to Linux distributions.

I would love to help manage that "core" team, and then we can focus on the features that the community wants/needs the most, and smaller features can be added (by the community / universities / interns / students) and hopefully we could get large enough, that we could sponsor students (similar to a "Google Summer of Code" but a "Lumberyard Summer of Code") where student interns (that want to work for 8-12 weeks during the summer) can pick a small feature that needs to be done, and come to work for 8-12 weeks (as an intern) and get that feature knocked out.

Plus it helps students get familiar with the engine and code base.  Eventually when they graduate, they can possibly join the core engine team (and take on larger core engine features).

I would love to get to the point where we have ONE ENGINE CORE that is a "common core" (similar to Linux) that ANYONE can use (and make that the "Lumberyard Core") and it is available under a very permissive MIT License so that ANYONE can take that "core" and develop a branch distribution (similar to Red Hat Linux, CentOS, Debian Ubuntu, Slackware, etc) and at least we have ONE solid core (similar to the Linux kernel) and then various distributions (that users can pick and choose from).

While "core" will focus on the main engine development (and most important features to core) as well as main tools.  And "distributions" can focus on packaging the core engine with additional addons that are stable, and also additional content that is free available (under MIT license) and that way developers don't have to "learn a new engine".

Since EVERYONE will eventually be sharing the same "core".  (Similar to linux).

That would be the most "beneficial" to developers and the community at large.  Then large companies like Epic Games (and even Unity) can adopt our core, and simply become distributions.

Unreal Engine 7 could be a fork/branch of our core, and Epic Games would simply "bundle" a Quixel Megascans content/license, and possibly a customized Epic Games Marketplace/Launcher.

Unity 7 could be a fork/branch of our core, and Unity Games would simply "bundle" a "XYZ" content/license, and possibly a customized Unity Games Marketplace/Launcher (as well as "Unity Ads").

CryEngine 7 could be a fork/branch of our core, and Crytek would simply "bundle" a crytek content/license (including all Crysis assets/content free of charge), and possibly a customized Crytek Marketplace/Launcher (as well as "Crytek Ads").

Amazon Lumberyard 7 could be a fork/branch of our core, and Amazon would simply "bundle" a customized Amazon Marketplace/Launcher (as well as "Amazon Ads"), and Amazon AWS, Amazon Twitch, etc. all included as part of the default Amazon distribution.

Google Chromium 7 (or whatever name Google would want to use / Stella? Starris?) could be a fork/branch of our core, and Google would simply "bundle" a customized Google Marketplace/Launcher (as well as "Google Ads"), and Google servers, etc. all included as part of the default Google distribution.

I suppose Apple and Microsoft could create their own distributions as well (if they so choose) and the nice thing is, at least game developers and students have ONE COMMON CORE BASE (that they learn and are familiar with) and it doesn't matter who's name is on "sticker" (as far as the distribution).

That way developers can choose the distribution (that in their opinion is best for them) and everything is based on ONE COMMON CORE ENGINE.  (Lumberyard core).

That in my opinion is the direction this whole project needs to be headed!

It's the ONLY WAY to move forward as a community, and it's such a mess right now, having to "choose" between Unity, or Unreal Engine, or CryEngine (blah).

NONE OF THE ENGINES are "perfect" and as a developer/instructor, it takes years to modify each of the engines to do what you want/need.

It would be better to have a common core, and then have distributions (from each of the various game companies like Epic Games, Unity, Crytek, Amazon, Google, etc.) that are ALL based on the 100% exact same core!

That way students only have to learn ONE CORE ENGINE.

It makes it easy on students (to study the code base and to modify the engine), and we focus on nice clean core code, that is easy to read, easy to document, and is modular/extensible and good APIs.

This way developers can focus on creating content/games/levels and whip out games (AAA-quality games) much faster (since we have engine core / engine distros done and ready to use) and have AAA-assets already designed/optimized ready in the marketplace.

Indie developers and students can focus on their storyline, and importing game assets, and creating their maps/levels/games and whipping out fantastic AAA-quality games (in cinematic quality) in a matter of weeks/months, and possibly a year or two at the most, versus the 5-12 years turn around (and $200+ Million to $300+ Million budgets / teams that we are currently seeing, just to churn out ONE AAA-quality game).

I would rather reduce the "overhead" and "duplication" that you see so many individual studios (and small studios) like CIG having to develop (internally) and instead bring ALL OF THAT to "core" (including the tools, and engine code) and then we just maintain the core (which Allows Chris Roberts and CIG to focus primarily on just creating game assets and storyline/content).

It takes game studios a large amount of money, and also it allows other game studios, indie developers, and students to use the "same tools" and "same engine" that the larger studios are using, and EVERYONE in the community can start whipping out AAA-quality games, without needing huge 1,000+ member teams (and $200+ Million dollar budgets) just to create a AAA-quality cinematic quality game.

That is my hope/dream, and I just hoping that we could make this happen!

[quote="JohnVanDerBurg, post:11, topic:8609"]
Lumberyard has a long way to go, and we all know that. I think a lot of mistakes were made in the past, we all agree on that. In my opinion those were mostly not mistakes by the devs though. I think we have some good changes coming up that haven’t been announced yet though. This should help LY improve in the future.
[/quote]

I agree.  I am NOT here to "beat up" on ANY Devs!  I love you guys, and I understand you are working hard, and I appreciate the Lumberyard Core Dev team, but I am just saying (from day one) that I really want this to be a COMMUNITY PROJECT, and I want "openness" and that is something that Amazon has NOT been good at.

Amazon needs to focus on the core engine.  Get it re-written.  Get it "free" from any former Crytek code, and get us a completely new and CLEAN game engine code base, that is extremely easy to read, and modular/extensible, and well documented.

Something NICE (like Epic Games Unreal Engine).

Because Crytek/CryEngine code has always been a COMPLETE MESS and a COMPLETE NIGHTMARE to work with.  It's the worst engine (and most "unfriendly" engine codebase) to work with, and it's so "slapped together" and "band-aided" (over the years) and it's a nightmare to deal with.

I feel sorry for Sean Tracy, and many of the core engine developers (over at CIG), but many of them came from Crytek and have a long history (with that complete garbage of a mess codebase) so it was easier for them to know where to look and what to modify and what to change.

But my hopes is that we can ALL get "free" of the Crytek mess and put that behind us.  I was very sorry to hear of Crytek suing CIG, and it really "scared" a lot of developers (and a LOT of the community) and nobody wanted to even touch (or mess with) the Lumberyard codebase and Lumberyard engine (until we could see what happens in the end/finality of this silly Crytek vs CIG lawsuit) and that was a complete nightmare for Lumberyard developers, as well as CIG, and the whole community as a whole.

Nobody wants to "mess" with Lumberyard if there are any "legal uncertanties" (or what could happen if Crytek goes on another "sue happy" lawsuit binge, like they did by attacking CIG).

It left a very bad taste in my mouth, and many (hundreds of thousands of angry developers, as well as MILLIONS OF ANGRY GAME FANS) that got REALLY upset with Crytek for suing CIG.

I'm hoping that mess is behind us, and I am hoping that we (as a community) can create a nice Lumberyard 2.0 engine (core engine) that is 100% completely free (completely re-written) and has ZERO crytek code, and that is a completely NEW ENGINE that we can license under a new permissive MIT license, and hopefully we can start drawing in users / developers that have abandoned Lumberyard (and CryEngine) due to the "complete mess" of an engine (as well as any legal uncertainties with Crytek) and as a result most developers left the Lumberyard project (as well as CryEngine) and most moved over to either Unreal Engine or Unity.

Unity is extremely fast build times (one click builds in seconds), and Unreal Engine is another great amazing engine (with the Quixel Megascans content available, and hundreds of millions of dollars in free AAA-quality game assets that Epic Games gives away for free each month to Unreal Engine users).

But my hope is that we can create a new engine (Lumberyard 2.0) and focus on the CORE ENGINE, and get everything we need added directly to Lumberyard CORE.  Then release this as a Lumberyard 2.0 (core engine) with bundled tools, and everything developers need to get started, and has "feature parity" with Unreal Engine 5, and Unity 6, and CryEngine 6.

That would be my request/goal for the Amazon Lumberyard Team, and what I hope that we can accomplish (as a community).  But we need OPENNESS from Amazon.  We need public roadmaps.  We need to be a part of this "requested features" and have the ability to request "feature" cards (that are added to the roadmap).

This way we can prioritize the cards/features, and at least we can see by looking at all the features (that need to be added) what needs to be done to the engine core, and we can start focusing on that!  (like what Epic Games does with Unreal Engine).

[quote="JohnVanDerBurg, post:11, topic:8609"]
Lumberyard has a long way to go, and we all know that. I think a lot of mistakes were made in the past, we all agree on that. In my opinion those were mostly not mistakes by the devs though.
[/quote]

I agree with you.  This is NOT the "devs" faults.  This is a MANAGEMENT PROBLEM and problems with Amazon uppper management.  I blame the LEADERS AT THE TOP, because this whole "Lumberyard Mess" is a 6+ year project/mess and nobody at the top seems to even know what they are even doing, and it's like a complete mess!

I actually FEEL SORRY for the developers (any any of the game developers that were even attempting to work at Amazon Studios) because this is NOT how you develop games!

Look at EA!  Look at how long EA focuses on a new engine (Frostbite) and focuses/dedicates developers specific to Frostbite engine (core development) and gets the core engine (features) developed FIRST, and then later Game development teams (and studios) use the Frostbite engine for their game development.

Amazon has ZERO EXPERIENCE in developing games (or game engines) and the leadership at the top, FAILED MISERABLE.

I don't blame the developers, I blame the FAILED LEADERSHIP at Amazon.

How could ANYONE think, "hey let's develop a game, when the engine is a complete mess!"  (and to somehow even think this was a "good idea" and NOBODY stopped it?)

Hundreds of millions of dollars have been lost/wasted, and I blame the Amazon leadership.

That's Jeff Bezos' money (thank God), because if it was MY MONEY, heads would be rolling!

But I'm far more frugal, and I believe in "doing more with less".  I would rather see a small lean amazingly talented "Core" development team, and then shut down the Amazon Studios and use those developers to come over to "Lumberyard Core" (so we have additional artists/developers) and use them to create assets/content/sample levels, and documentation, tutorials, videos and AAA-quality demos/samples (and help us get the Lumberyard Core completed and "production ready") and help us hit a completely new and completely re-written (from scratch) Lumberyard 2.0 release.  That is completely free of any former crytek code, and is something that is new.

That way we can license it under a permissive MIT license, and then finally make it a COMMUNITY driven project.  Make Lumberyard a new separate entity (outside of Amazon control).

Then have Amazon sponsor developers (just like Google sponsors developers for Chromium project or Android project).  This way Amazon is still being a good "corporate citizen" and contributing to the continued development of the core engine.

Then we can package a distribution (specifically for Amazon) with nice Amazon logos, and a nice Amazon launcher.

But we can also create customized launchers for Google, Epic Games, Unity, Microsoft, Intel, AMD, NVIDIA or anyone else that wanted to help sponsor the project (by funding core developers for the core engine dev team).

This way everyone can benefit, and AMD devs can focus on optimizing the core engine for AMD graphics cards and AMD processors (multithreaded).  The Intel team can focus on optimizing the engine for Intel processors and Intel graphics (Iris, Intel HD, Intel Xe, etc.).

The NVIDIA team can focus on optimizing the engine for NVIDIA graphics cards (GTX/RTX, etc.) and supporting Metal (for Apple/iOS/macOS), Vulken, etc. in core engine (optimized for their hardware).

Sony can sponsor a few devs (Core devs) that will optimize the engine/core specifically for PS4 and PS5.

Nintendo can sponsor a few devs (Core devs) that will optimize the engine/core specifically for Nintendo Swithc.

Microsoft can sponsor a few devs (Core devs) that will optimize the engine/core specifically for Microsoft Windows / DirectX 12 Ultimate.

Red Hat & Canonical can sponsor a few devs (core devs) that will optimize the engine/core specifically for Linux (RHES/Red Hat, and Ubuntu).
- Other linux distro developers can pull from this, and add to their own distros such as CentOS, and Ubuntu forks

Microsoft can add (and fund/support) Microsoft devs that can focus on HaloLens support, as well as optimizing the game engine for XBOX Series X, and Windows 10, and Windows server, and using Microsoft Xbox game controllers, and microsoft hardware, as well as DirectX 12 Ultimate features/support optimized in core Lumberyard core engine.

It takes an "Army" to develop (and maintain) a great engine, and unfortunately Amazon doesn't seem to understand this, and doesn't want to "let go" of control, and would rather see Amazon Lumberyard (sink and DIE a horrible death) than to be OPEN, and have OPENNESS, and let the community take over with the Lumberyard engine.

Reach a nice Lumberyard 2.0 (that is clean of all former crytek code) and is a new clean re-written engine, that has feature parity to Unreal Engine 5 and Unity 6.

Then release that engine to us (as a community) and let us create a "Lumberyard Core Dev team" that will be responsible for maintaining the engine, and then Amazon will sponsor a large team of devs (current Amazon Lumberyard developers) and just allow us (as a Community) to steer the project, and just "sponsor" the developers.  (Similar to Canonical and Ubuntu, or Google and Chromium or Google and Android).

This will benefit the community, and will also benefit Amazon, because we can create the distros for Amazon (the "skinned" launcher for Amazon) so that way Amazon has a nice themed/skimmed launcher, and Amazon Marketplace, and Amazon marketing materials, etc.

Plus twitch integration, AWS integration, etc.

While the core devs continue to work on the Core Engine, (and features) and get this engine into Production -ready status (so artists, studios, students and Indie developers can actually start using it).

I share the same "vision" as Jeff Bezos (and I'm glad that he decided to get into the engine business) but unfortunately Jeff Bezos doesn't know anything about game engines, and this is not his "area of expertise", but I am grateful for his sponsorship of the Lumberyard developers, and his purchase of that original cryengine license (as well as SpeedTree license, and Wwise license, etc.) to help get the engine started, and to help Lumberyard team developers "get their feet wet" with developing and working on a game engine.

Answer

It seems that ALL of my responses are being censored/blocked (flagged by the community).

REMOVEDUPLOAD

Hopefully this can be resolved, because every single post that I have made for the past week or two is being "flagged by the community" and being censored (and/or blocked).

I can't post ANYTHING or even RESPOND to anything, without it getting "flagged by the community".  Every single post is being "flagged by the community".

Hopefully this can get FIXED...

Thanks,
Mark

Answer

There has been some work in this area before. I think the main engineer of Nanite also mentioned this in some post he did. He was looking at Geometry Images:: http://hhoppe.com/proj/gim/

Perhaps there are some similarities to this and adjusted to modern hardware perhaps.
We'll have to see what hardware requirements there are. Also does it work only on static geometry, or also deformable objects like characters. It also might complicate some things like foot IK and placement (which they also did work on), and perhaps even collision detection.

And what is the size of the models on disk. Do games now suddenly be 10x larger as they have way higher poly counts? Or is it using the images and compressing them, as in the link above.

And what's the required disk speed. Did they choose PS5 because that has special ssd with compression support, which other platforms don't have?

It will be interesting to get to know more details. Also they aren't shipping it officially until the end of 2021 I think, with a possible preview at the end of this year.

Answer

The Nanite technology looks incredible, for sure! :) This is not just for cinematics though, it is going to be a different approach to things. The same as we had virtual texturing, but now for geometry. At this time, details still have to be provided on the exact implementation details. We have to see how much it depends on say the disk speed and required graphics hardware (or CPU as a part is using software rasterization I read).

I am sure the Lumberyard graphics team is closely monitoring this. Even though nothing can be promised I am sure eventually this will make its way in Lumberyard as well if this is really transforming the industry.

Definitely exciting times! :)

I am sure the Lumberyard graphics team is closely monitoring this. Even though nothing can be promised I am sure eventually this will make its way in Lumberyard as well if this is really transforming the industry.

Definitely exciting times! :)

Answer

I don't want to "bash" on Amazon too much, because Amazon DID pay for the original CryEngine license (and did sponsor continued development of Lumberyard for 6+ years), but this project should be a COMMUNITY LED project (not an Amazon corporate led project).

In my opinion, Lumberyard should be refactored/rewritten and released as Lumberyard 2.0 (under a permissive MIT-license) very similar to GoDot Engine, and then let the community get involved, and continue to lead development of Lumberyard!

Amazon likes to USE OPEN SOURCE SOFTWARE, but Amazon doesn't like to SHARE open source software (Amazon is not Google).  Google gives everything away for free (and makes their money on Ads, Google Ads/Adsense).

In my opinion, Amazon doesn't like to give anything away (when it comes to open source software development), and likes to base ALL of their products on open source software (even FireOS is based on Android), but you'll never see the FireOS codebase made public, and Amazon doesn't like to "share" or "play well" with others.  This is an internal problem with Amazon (and Amazon leadership).  Amazon has NEVER been a very "open" company that embraces OPENNESS.

Amazon benefits from the open source community, but it doesn't seem to fully embrace the open source community (the lack of a public Roadmap is one major example).

I was hoping that Lumberyard would be different and it is great that the source code is available, but the community can't do anything with it, because Amazon holds tight reign over it (i.e. license holder) and without a complete refactorization of that source code, and creating a completely new engine (with a nice clean codebase) that is completely refactored and designed for modern hardware (multi-core CPU, multiple CPUs, multiple-GPUs) that is designed to be highly threaded and highly optimized (for modern hardware) and release it under a MIT License as Lumberyard 2.0.

That is what we need.  Work as fast (and as hard/quickly) as possible, get this done, and release it to the Community as Lumberyard 2.0 (under a permissive MIT-license) and then let the community take the reigns, and let us ALL work together (as a community) and collectively with thousands of developers working together, we can create something AMAZING.

That is what I want to see happen, but unfortunately Amazon has wasted 6+ years already, and is still wandering around like a herd of aimless cats... with no clue, and no direction.  No guidance, and no real vision, and just wandering around aimlessly with no clue as to what they are even doing or where they are even headed.

I don't blame the developers, I blame the Amazon leadership.

Ultimately, I blame Jeff Bezos for not "nipping this in the butt" (several years ago) and bringing this nonsense to an end, and pushing to get this done.  I would be "lighting a fire" under everyone's butt to get this refactored, and get a Lumberyard 2.0 (under a permissive MIT-license) pushed out to the community so the COMMUNITY can start to lead the development of Lumberyard (exactly like the GoDot Engine community), and then Amazon can take a deep breath, relax, and just let others (AMD, NVIDIA, Intel, Google, Toyota, GM, etc.) jump onboard (as sponsors/developers) and allow others to work collaboratively on the COMMUNITY Lumberyard Project / Engine.

GoDot has over 100,000+ developers working on the core engine!

https://godotengine.org/article/status-godot-gdc-2019

Look at the GoDot Development Process here:
https://godotengine.org/article/shedding-light-on-godot-development-process

Their focus is on OPENNESS and TRANSPARENCY!  (two things that Amazon is TERRIBLE about)

As crazy as it sounds, I believe in FIVE YEARS that GoDot will leave Amazon (and Amazon Lumberyard) in the dust... in the shambles, and Lumberyard will probably go up in flames and burn itself into the ground as a "dead engine", while the community works together and turns GoDot into something Amazing.

That is what is currently happening, and in 5 more years, it might actually be a reality.

But Amazon has to decide, whether they want to refactor Lumberyard and release it under an MIT license (and release a "Lumberyard 2.0" to the community under a permissive MIT license) and then let the Community take charge (as a truly open source community-driven project) and then Amazon can continue to sponsor (and fund) developers (as Amazon Lumberyard employees) that work on the Open Source Community Projects (just like Google does with Chromium or Android, or Oracle does with MySQL or Java, or Canonical does with Ubuntu, or Red Had does with RHEL / CentOS.

We NEED OPENNESS.  We NEED A COMMUNITY DRIVEN PROJECT, not an Amazon driven project.

There is a BIG DIFFERENCE between community members and corporate members (and corporate leadership).  This can't be done via Amazon leadership.  It won't work.  It has no future.  It WILL FAIL.

Amazon doesn't understand what the community wants (or needs).  Amazon is NOT LISTENING to the community.  Amazon doesn't care about the community.  Amazon is too "disconnected" and too "out of touch" and too "clueless" and I am talking about LEADERSHIP (not developers but leadership).

The ONLY way to SAVE this project, is to quickly refactor Lumberyard into a completely new codebase clean of any existing Crytek code, and completely clean, and release this new version as a Lumberyard 2.0 (under a permissive MIT License) and then Amazon leadership needs to shake hands with the community and let's work together on Lumberyard (as a community driven project).

That's when you'll see Lumberyard (2.0) pick up some steam, and begin to move forward, and I'm hoping that we could keep up with GoDot, and in time eventually beat Unreal Engine, Unity, and hopefully GoDot as well.

That would be my hope/dream/goals for Lumberyard, but unfortunately I can't seem to get anyone at Amazon to UNDERSTAND THIS, and to get Amazon leadership to MAKE THIS HAPPEN.

Stop screwing around.  Just get the engine refactored, get it clean, and get it released as a 2.0 release.

Don't worry about anything else.  NOTHING.

Then we create a public roadmap, and put EVERYTHING that we want on the "To Do List" and then start prioritizing features, and then we let the developers work on whatever it is that they want to work on.  (The Joys of community development).

In open source projects, most developers work on their free time, on whatever they want, and when they want. No one has any authority over them, be it moral or monetary, and this is for the best.

Look at Linux, and some of the brightest minds are working for these projects.

Personally, I would allow ANYONE to create a FEATURE REQUEST (using GitHub) by opening an issue.  If someone makes a forum post (with a FEATURE REQUEST), I would help move this over (and translate the post into something that developers can understand) and move it directly to GitHub (as an "open issue").

Then I would post the "issue" to a developers forum, and seek community help/support, and hopefully someone interested in the issue will step up, and take charge and work on that issue/feature.

Many developers often need the same features (and many are duplicating their efforts that other developers are also working on) and just like CIG has added many features (that others want/need) but that is not being shared.  Instead, it needs to be a community driven project, and let CIG and Amazon contribute to the community driven (Lumberyard 2.0) and let Star Citizen be based completely on a Lumberyard 2.0 core!

That is what I would love to see!  (Plus you'd probably see lots of other great games pop up in record time, as others are able to use the amazing engine and features in the core).  But as it stands, things are just moving too slow, and everyone is working in opposite directions, and there is NO COMMUNITY and NO COMMUNITY LEADERSHIP.

Amazon is holding onto Lumberyard extremely tight, and everyone else is working on their own forks and nothing is getting fed back upstream (back to Lumberyard core) and as a result, we are seeing NOTHING come of Amazon Lumberyard.

Just lots of failed projects!  None of those assets, or anything from any of those failed projects even makes it back to Lumberyard core!  (which is terrible).  Just lots of wasted energy, and nothing seems to come of this.  I'm really hoping that this can change (fairly quickly).  We need a Community Driven project!  (not a corporate Amazon led project, but a Community led and community driven Lumberyard project!)

[quote="JohnVanDerBurg, post:11, topic:8609"]
I think LY is making some good moves now, which you should see later on. We can’t change the past, but I’m more optimistic about the future.
[/quote]

I still don't believe that.  In 6+ years, I'm just seeing "more of the same" and nothing seems to change.  Just one failed project after another, and it seems like a big mess, and I don't see anything changing anytime soon.

The ONLY WAY to save this project, is to remove it from Amazon and make this is a COMMUNITY DRIVEN project (instead of a Corporate project) and just ask Amazon for sponsorship (Amazon sponsors "X" number of developers for "X" number of years).  We would ask Google, Microsoft, AMD, NVIDIA, and other large corporations to do the same (so that the number of developers can increase, and the financial load is "shared" among sponsors) and ultimately the community benefits from the project, and we can finally move forward (at much faster speeds) especially as Universities and students get involved, and contributions to the codebase start growing, and over time we have a nice collective effort.

I believe this is the ONLY way forward for Lumberyard.  Amazon is NOT Epic Games.

[quote="JohnVanDerBurg, post:11, topic:8609"]
I personally think Amazon took on too much work, like supporting internal game teams, other select teams, and a community, all with limited resources, with most focus on the internal projects, which made it feel like the community isn’t being listened to as much as they should.
[/quote]

That's the problem!  A big huge waste of money, and wandering around like a herd of cats, and in the end nothing gets accomplished!  Amazon is trying to drive the train (without laying down the tracks).  It doesn't work out well for ANYONE involved!  Developers are spread too thin, internal projects ultimately fail (nothing is ever released to the public) all the time/effort/assets go up in smoke, and the core engine is still a mess, and the community is still stuck in the same place that it was 6+ years ago!

It was a terrible idea, and terrible direction that Amazon leadership chose to take.

But in my opinion, the BEST THING that Amazon could do is give the Lumberyard engine to the community, and the Community developed Lumberyard (free of Amazon leadership) and we are a community led project (but Amazon sponsored), then we can move forward and be focused more on being community driven (and community led).  Anyone that wants to contribute can.  Anyone that wants a feature, can add a feature (and submit a pull request).

A small handful of core developers would create the features (feature cards) and put them on the public roadmap and any pull requests that complete a feature will be reviewed (for code review, and ensure the code is neat/clean) and then pulled into core.

To reach a consensus, we regularly hold meetings to review PRs (on chat/IRC).  If an agreement is reached, the PR is merged.  We can move fast and hard.  (without all the nonsense of Amazon failed leadership).

If a PR creates some problems (or is challenged by other developers) it can be reverted so that a new PR can be worked on addressing the shortcomings of the first one.  (This is what is great about the community projects).

As for "direct commits" we'll have several core developers that are "owners" of various areas of the engine (and additional maps/content/assets).  This means they have enough knowledge of the engine that allows them to commit directly to keep improving or working on the engine (and these devs will hopefully be "sponsored" by Amazon).

Likewise, many of these same "core devs" would have the ability to usually decide to merge a PR in an area of their responsibility (if they approve of it).  This will speed up the process, and we'll hopefully have thousands of community users building from source (to get the latest "experimental" features) in the same way that Epic Games and Unreal Engine does it.

This way "stable builds" are available (by content creators/artists) for creating games, and "experimental builds" are available for features that are not fully tested yet, and for testing new bleeding edge features (that will eventually make it into a stable build).  But this allows both developers, artists and content creators (and Indie developers) to decide whether they want to use experimental builds or production builds.  If everything they need is in a production build, that's fine.  If a feature they need is "experimental" they can build from source, or use an experimental build.

With hundreds or thousands of community members building from source, bad commits are easy to find and bisect and they can be quickly reverted.  Which will help keep Lumberyard stable (while still being able to be developed quickly).

But I believe it is going to take a large community to develop Lumberyard, and we can't do that when Amazon is NOT OPEN, and won't even publish a public roadmap.  As a community, we need a public roadmap, and need to know the direction that we are headed (as a project).  This is where the major "disconnect" between Amazon and the Community is.

[quote="JohnVanDerBurg, post:11, topic:8609"]
That said though, I think LY is making some good moves now, which you should see later on. We can’t change the past, but I’m more optimistic about the future.
[/quote]

I've been here for 6+ years (before Lumberyard even existed) and the day that Amazon inked the deal, I was extremely optimistic and had great hopes for this project.  After 6+ years, I'm definitely not feeling as optimistic anymore (after seeing the direction this project has taken), and six years is an extremely long time, and we still don't have a roadmap, and we are still not a community driven project!  We are still corporate led, and corporate governed, and for this reason... I don't believe Amazon leadership will ever do anything with Lumberyard, and it will most likely fail.  Amazon would rather let this fail (and die) and be a complete waste of money, time, efforts and resource... then to turn this over to the community (for community driven project) and then just agree to be a Amazon Corporate Sponsor (to fund the core developers) working on the project.

That would be the "best scenario" for the community, and it also allows other corporate sponsors to get involved, and helps distributes the load and allows others to work on areas of the engine that they want to work on, and the community as a collective "owns" the engine, and we work together on making something great.

I personally love the GoDot Engine's development model, and that is would I was hoping for (for Lumberyard) and in 6+ years, it seems like that is NOT the case, and there is a big huge "disconnect" between the community and Amazon, and we can't seem to "bridge" that gap.

The current model/direction of being a "Corporate run" project is a big mess, and it's a nightmare for anyone that wants to even try and use Lumberyard.  We still have NO ROADMAP in 6+ years!  How can ANYONE even take this project seriously?  I hope we can resolve this, and work together and make Lumberyard amazing.

[quote="JohnVanDerBurg, post:11, topic:8609"]
I think LY is making some good moves now, which you should see later on. We can’t change the past, but I’m more optimistic about the future.
[/quote]

I hope so.  I look forward to speaking with you on Thursday, hopefully we can work together and come to some form of resolution, and I would truly love more than anything to see Lumberyard grow and succeed as a project (but I believe it needs to be a "Community Effort" and not a "Corporate Effort".  I believe a "Community Effort" with a "Corporate Sponsorship" would be the absolute best way to move this project forward for everyone (for the community, as well as Amazon) and it would be the most beneficial to both sides.  We need to refactor the codebase, create a new clean Lumberyard 2.0 engine, and we need a permissive MIT-license.  That is what we (as a community) need.  Once we have that, we can look for other corporate sponsors (Google, Intel, AMD, NVIDIA, Sony, Microsoft) that can hopefully sponsor a few additional core developers that can help, and I would love to see a great engine that will be designed/optimized for the very latest consoles, as well as take advantage of the very latest processors (highly multi-core/multi-threaded) and take advantage of faster and faster hardware (PCIe4 and PCIe5 NVMe VNAND) and I hope we can reach cinematic quality real-time rendering (with real-time ray tracing) at a decent frame rate using virtualized texturing and dynamic geometry (similar to Nanite / NVIDIA Asteroids demo).

Many great projects including Linux, MySQL, WordPress, Drupal, Apache, PHP to name just a small few, are community driven open-source projects.  I really like the direction that GoDot Engine is heading (as an open-source community driven project).  I really wish/hope that we could completely refactor the Lumberyard engine (source code) and clean it up, rewrite/refactor it, and create a new clean codebase (free from any old Crytek code), so that way we have a completely new clean engine, that we can call Lumberyard 2.0 (as a new engine engine release) and release it under a permissive license like the MIT-license, exactly like GoDot Engine!  Then have a Lumberyard 2.0 (licensed under a permissive MIT license) that the community can work on and develop (with core engine developers being sponsored by Amazon) and this way we can grow the community, and be a community driven project!  Universities, and other corporate entities (Google, AMD, NVIDIA, Intel, Microsoft, etc.) can sponsor developers, and contribute to the core development, and we can start moving forward (and hopefully beat GoDot Engine!).  That is my "hopes" for the Lumberyard project.

I truly hope that we can work together as a community.

Thanks,
Mark

Answer

Hi John,

To answer some of your questions:

1) Yes Nanite is designed to take advantage of modern hardware (PCIe4 NVMe) with throughputs in the 7,000+ MB/s range.  Modern current-gen AMD CPUs **ALL** support PCIe4 NVMe.  (i.e. Ryzen 7 3700X).

2) The PlayStation 5 and Xbox Series X game console hardware is very similar to a AMD Ryzen 3700X (8C/16T) and a custom Navi-based GPU (similar to an AMD RX 5700 XT) with a custom 12-channel VNAND SSD controller (similar to PCIe 4.0 NVMe) but it supports 6 "tiers" (priority tiers) instead of 2 tiers (priority tiers) like standard NVMe 4.0.  (Which is good for pulling large data, and smaller audio in real-time, just-in-time loads for unexpected things like death, etc.)

To answer your questions about the PS5 hardware see (Part 1) here:
https://www.youtube.com/watch?v=erxUR9SI4F0

Then see Part 2 here:
https://www.youtube.com/watch?v=v1SRHrCa4XQ

Then see Part 3 here:
https://www.youtube.com/watch?v=ESQyAhooBLM

Those 3 videos listed above, are a good "reaction" videos (giving the actual hardware discussion, and breakdown as to how it relates to game engine development) by a former EA Game Engine engineer (who formerly worked on creating EA's Frostbite Engine).
As for "disk space" the geometry and assets are compressed

Mike Cerny (from Sony) gave a pretty in-depth review on the PS5 hardware (to game developers) and you can watch the actual video (from Sony) in its entirety in the three video links that I posted above (with reaction/explanation/breakdown).

I'm quite surprised that NOBODY from Amazon/Lumberyard even understands what is coming (for consoles) this fall.  EPIC Games and Crytek are already supporting the next-gen consoles (PS5 and Xbox Series X) and it seems strange that nobody at Lumberyard is even working on supporting the PS5 and XBOX Series X (that are coming this fall) and Epic Games Unreal Engine 4.25 (current production release) already supports creating games and game development for the PS5 and Xbox Series X!  You can actually do one-click builds and run your Unreal Engine 4.25 games on PS5 and Xbox Series X dev kit consoles right now using Unreal Engine 4.25!

Many of these same features (used in the Unreal Engine 5 demo) are current "experimental" (beta) features that you can use NOW in Unreal Engine 4.25.  These same exact features will be "production ready" in Unreal Engine 5, but are currently available now as optional features (such as Chaos destruction physics, Niagara GPU particles, Virtualized Textures, and DCC on consoles) that you can use NOW (if you build from source).  Most of these features shipped with UE 4.23, and are available in Unreal Engine 4.25 (if you build from source), but ALL of these features will be default (in Unreal Engine 5) and be considered "production ready" for UE5 release.

But most of the features from the UE5 demo are currently available (as "beta" features if you build Unreal Engine 4.25 from source)

You can see here:
https://docs.unrealengine.com/en-US/Engine/Rendering/VirtualTexturing/index.html

Runtime Virtual Texturing here:
https://docs.unrealengine.com/en-US/Engine/Rendering/VirtualTexturing/Runtime/index.html

RVT Quick Start Guide here:
https://docs.unrealengine.com/en-US/Engine/Rendering/VirtualTexturing/Runtime/QuickStart/index.html

Streaming Virtual Texturing here:
https://docs.unrealengine.com/en-US/Engine/Rendering/VirtualTexturing/Streaming/index.html

What is used in the Unreal Engine 5 demo is simply "Streaming Virtual Texturing", Global Illumination (Lumen), [Chaos Physics](https://trello.com/c/oHdcH6oH/496-chaos-physics-production-ready), [Niagara VFX/Particles](https://trello.com/c/26ttfAdO/521-niagara-simulation-and-iteration-stages-experimental) and also "Streaming Dynamic Geometry" (Nanite).

Chaos Physics here (Beta in 4.23 / Production Ready in 4.25):
https://trello.com/c/VqRYT26b/455-chaos-destruction-beta

Chaos supports cinematic quality destruction and physics (in real-time).

Whenever a building is damaged, objects fracture, smoke and dust is generated as well as more tiny fractured bits that enhance the physics simulations' visuals.

Niagara Particle Systems can now be generated by the physics simulation in Chaos.  See here:
https://trello.com/c/i51lp6R5/389-niagara-integration-into-chaos-physics

Also does it work only on static geometry, or also deformable objects like characters. It also might complicate some things like foot IK and placement (which they also did work on), and perhaps even collision detection.

If you watch the demo, and watch the discussion/developer reactions (on YouTube) this seems to work on both static geometry, as well as deformable objects (like water, characters, etc.).  Foot IK and placement has already been worked on (to prepare for this in current 4.25 builds), and collision detection has already been worked on (with Chaos physics) to prepare for this.

If you watch the 3 videos that I posted links to (above), and listen to Mike Cerny (from Sony) discuss the PS5 hardware, he discusses that the custom AMD Zen2 CPU/APU used in the Sony PS5 has

And what is the size of the models on disk. Do games now suddenly be 10x larger as they have way higher poly counts? Or is it using the images and compressing them, as in the link above.

EVERYONE at Amazon Lumberyard Team should be REQUIRED to watch the three links that I posted above (EVERYONE!!) from Mike Cerny's detailed PS5 console hardware breakdown (for developers) so that Amazon Lumberyard Team (EVERYONE) can understand what they need to be developing the Lumberyard game engine for!  (Next-gen consoles and next-gen computers).

It seems "odd" that nobody at Amazon Lumberyard Team even knows what is going on, and doesn't even have dev consoles (from Microsoft Xbox Series X, and Sony PS5 dev consoles) which Epic Games, and even CD Projekt Red have had for over a year now!  (So they could work on optimizing Epic Games Unreal Engine 4 (UE 4.25) specifically for the PS5 and Xbox Series X games consoles hardware!

You should also look at the audio (which supports hundreds of sound sources) in both PS5 and Xbox Series X.

You can see here:
https://www.youtube.com/watch?v=ESQyAhooBLM&t=12m6s

Watch this clip here on the 3D Audio (with good locality) and HRTF here:
https://www.youtube.com/watch?v=ESQyAhooBLM&t=14m16s

These are ALL features that Epic Games has been working on for the past 2 years, and are now included in Unreal Engine 4.25 (Production Ready) builds for creating PS5 and Xbox Series X games.  (They were used in the Unreal Engine 5 demo, and will also be included in UE5).

*And what is the size of the models on disk. Do games now suddenly be 10x larger as they have way higher poly counts? Or is it using the images and compressing them, as in the link above.*

Yes, the models on the disk are much larger, but if you watch the PS5 video, you'll see that game developers have been using compression to compress the models and images (100x) so the actual disk space usage is smaller (even though the models are much larger) due to use of compression.  Compression combined with faster SSD (PCIe4 NVMe) (with read/writes of up to 7,000MB/s) allows for 100x faster performance (in disk reads/writes and file I/O, as well as mapping, and coherency) and you can learn more about this here:

[https://www.youtube.com/watch?v=v1SRHrCa4XQ](https://www.youtube.com/watch?v=v1SRHrCa4XQ)

Watch this here about the PS5 SSD Controller here (which uses 4 lanes of PCIe4) and is very similar identical in performance to modern AMD Zen2 CPUs that support current generation PCIe4 NVMe M.2 drives (which are capable of up to 7,000MB/s read/write):
https://www.youtube.com/watch?v=v1SRHrCa4XQ&t=7m47s

As for Compression, game developers are now using "KRAKEN" from Rad Game Tools

You can watch/learn about this here:
[https://www.youtube.com/watch?v=v1SRHrCa4XQ&t=8m20s](https://www.youtube.com/watch?v=v1SRHrCa4XQ&t=8m20s)

You can learn more about Oodle Data Compression Family (Kraken, Leviathan, Mermaid, Selkie) here:
[http://www.radgametools.com/oodle.htm](http://www.radgametools.com/oodle.htm)

*And what’s the required disk speed. Did they choose PS5 because that has special ssd with compression support, which other platforms don’t have?*

ALL modern computers (AMD Zen2/Zen3) and consoles (PS5 & Xbox Series X) support PCIe4 NVMe M.2, capable of 7,000MB/s read/writes.

A modern Gigabyte AORUS Gen4 NVMe (PCIe4 NVMe) for AMD Zen2 or AMD Zen3 computers supports at least 5,000MB/s reads and 4400MB/s writes (and has been available for over a year and a half now).

Here:
[https://www.newegg.com/gigabyte-aorus-1tb/p/N82E16820009012#](https://www.newegg.com/gigabyte-aorus-1tb/p/N82E16820009012#)

Which is comparable in performance to the PS5 SSD.  (Newer multi-channel SSDs that have 12 and 16 and 18 channels, which are capable of read/write speeds in the 7,000MB/s read/write range were demonstrated at CES, and should be coming to market in Fall 2020, which will outperform even the PS5 SSD, and this is why Sony has included a standard PCIe4 NVMe socket/slot, so that you can install a newer PCIe4 NVMe inside the PS5 for faster 7,000MB/s read and write times, and also for increased disk storage).

But yes, modern computers (2018 and later) with AMD Zen 2 or AMD Zen 3 processors are now using PCIe4 NVMe M.2 (with 5,000+MB/s read/writes) and in the near future the newer generation VNAND multi-channel controllers will be supporting 7,000+ MB/s (in fall 2020).  Also in 2021, AMD is implement PCIe5, and the newer PCIe5 NVMe controllers will be supporting 14,000+ MB/s read/write times (using 4 lanes of PCIe5, and supporting up to 14,000MB/s read and write times, using newer multi-channel VNAND controllers and supporting 16 to 36+ VNAND chips using a single controller).

So yes, faster disk read/writes is normal for 2018/2019/2020 hardware, and game engine developers should be developing their game engines to support these newer hardware technologies (and games are expected to be around 100-150GB in size, even using Kraken or Leviathan.

And what is the size of the models on disk. Do games now suddenly be 10x larger as they have way higher poly counts? Or is it using the images and compressing them, as in the link above.

All ultra-high-resolution game models (geometry), and ultra-high resolution textures are compressed (reducing the file size about 85%) and Leviathan decodes about 10x faster than 7z/LZMA, and Kraken decodes about 3-5x faster than zlib, with way more lossless compression!

I'm quite surprised that NOBODY at Lumberyard even knows this, are you guys NOT game developers?  When is the last time you tried creating a AAA-quality game?

Kraken has been used for well over 5+ years now, and is used extensively on PS4 and Xbox One, to decrease the size of game downloads, as well as loading compressed data and much faster I/O of compressed data, using hardware decompression on PS4 and Xbox One! (as well as Oodle Network Compression, which reduces network bandwidth for multi-player games, and game streaming).

Oodle Kraken has been around for quite some time!  Oodle Leviathan is the next generation compression, and that is what newer games (2019 and later) are now using.

Amazon Lumberyard is quite "behind the times" when it comes to game engine development, and creating AAA-quality games.

*And what’s the required disk speed. Did they choose PS5 because that has special ssd with compression support, which other platforms don’t have?*

ALL systems support Kraken hardware compression (including PS4, Xbox One, PC, etc.) dating all the way back to 2014-2015 era.  Kraken is used in most games (anything 2017 or newer) for PS4 and Xbox One.

Leviathan is much newer, and is used on newer games (anything 2019 and newer), since it's incredibly fast and has the highest lossless compression ratios, and decodes even 10x faster! (using modern CPUs/GPUs).

All modern CPUs support decompression.  Decompressors make use of SIMD and work with all current generation (and next generation) hardware.

Mermaid, Kraken, and Leviathan are MUCH MUCH MUCH faster (over 10x faster) than loading uncompressed data from disk!  So compression is ALWAYS used!

You can read more here:
http://www.radgametools.com/oodlecompressors.htm

[quote="JohnVanDerBurg, post:5, topic:8609, full:true"]
There has been some work in this area before. I think the main engineer of Nanite also mentioned this in some post he did. He was looking at Geometry Images:: http://hhoppe.com/proj/gim/

Perhaps there are some similarities to this and adjusted to modern hardware perhaps.
We’ll have to see what hardware requirements there are. Also does it work only on static geometry, or also deformable objects like characters. It also might complicate some things like foot IK and placement (which they also did work on), and perhaps even collision detection.

And what is the size of the models on disk. Do games now suddenly be 10x larger as they have way higher poly counts? Or is it using the images and compressing them, as in the link above.

And what’s the required disk speed. Did they choose PS5 because that has special ssd with compression support, which other platforms don’t have?

It will be interesting to get to know more details. Also they aren’t shipping it officially until the end of 2021 I think, with a possible preview at the end of this year.
[/quote]

Mermaid, Kraken, and Leviathan are MUCH MUCH MUCH faster (over 10x faster) than loading uncompressed data from disk!  So compression is ALWAYS used!

You can read more here:
http://www.radgametools.com/oodlecompressors.htm

Also newer CPUs (AMD Zen2 and Zen3) support PCIe4 (and PCIe4 NVMe) with 4 lanes of PCIe4, a normal M.2 (PCIe4) NVMe supports up to 7,000MB/s read/write speeds!

Newer multi-channel controllers are available that support 7,000MB/s read/write speeds (using PCIe4 NVMe M.2's) and these were demonstrated at CES, and should be available this fall (Fall 2020).

Gigabyte already has the Gigabyte AORUS PCIe4 M.2 x4 NVMe capable of 5,000MB/s read and 4,400MB/s writes here:
https://www.newegg.com/gigabyte-aorus-1tb/p/N82E16820009012#

Using Toshiba BiCS4 96 Layers 3D TLC (800MT/s) and using a Phison PS5016-E16 controller (with eight NAND channels with 32CE targets), DDR4 DRAM caching and a PCIe4.0 x4 interface.

Phison has newer PCIe4 controllers (that are now available) that support up to sixteen channels, and support read/write speeds of 7,000MB/s!  Also Phison is currently working on newer PCIe5 controllers (PCIe5 x4) that support 14,000MB/s read and write speeds (for the new AMD Zen4 processors that will be released in the fall of next year), so AMD Zen4 (Ryzen 5000 series) processors will support 14,000MB/s read and write speeds!  (next year in fall 2021).

I'm EXTREMELY surprised that the Amazon Lumberyard Team is so incredibly behind, and NOT "keeping up" with modern hardware, and modern technology.  Epic Games seems to take this seriously, and is developing their game engine to support modern (current gen) (which came in 2017-2019) and modern next-gen (fall 2020-fall 2021) hardware that is coming in 2020 (and future PCIe5 hardware that is coming in fall 2021 with 14,000MB/s read/write speeds).

If you look at PCIe5 M.2 NVMe (with 14,000MB/s read/write speeds) combined with Oodle Leviathan compression, you can have an insane amount of geometry and 8K or 16K textures compressed (that would normally be about 8TB to 10TB of disk space) that can be compressed down to around 100-150GB of disk space (using Leviathan compression) and then accessed from the disk (about 80-100 times faster I/O due to the data compression) and is decompressed (on the fly) using modern CPU or GPU cores (with Oodle compressors/decompressors).

Loading one gigabyte with 7z/LZMA takes about 20 seconds.  The same gigabyte of data with Oodle Kraken takes LESS THAN 1 SECOND!  Oodle makes it possible to use powerful compression all the time, without making your users wait for decompression!

Oodle runs great on every platform (PC, PS4, XBox One, PS5, and Xbox Series X) and it drastically increases your I/O capabilities (by using compressed data), and you can easily do 80x or 100x the I/O using compressed data (using newer Leviathan compression) that decodes 10x faster with way more compression!

Kraken is by far the fastest (and most common) compression, but in another year or two, Leviathan will probably be used (predominately) for PC and console compression use.

Here:
http://www.radgametools.com/oodleleviathan.htm

As well as Oodle Network Compressor (for online multiplayer games):
http://www.radgametools.com/oodlenetwork.htm

Especially on servers with a large number of connections, like MMOs, it's a huge reduction in memory use (lowest per-channel memory use of any comparable packet compressor).  Oodle for TCP uses half the memory of zlib.  Oodle for UDP uses no per-channel memory at all.

On servers with a large number of connections, like MMOs, this results in a huge memory savings, several gigabytes!

I'm quite surprised that NOBODY at Amazon Lumberyard Team even understands this!

I have been complaining about this for several years now, but NOBODY at Amazon Lumberyard seems to even understand (or even comprehend) how to make a modern AAA-quality game (let alone develop a game engine for a AAA-quality game).  Look at what Epic Games is doing with Unreal Engine, and that's the direction that the Lumberyard development team should have been heading (5+ years ago), but unfortunately it seems like the Lumberyard Team is just too far behind, and they haven't even THOUGHT ABOUT these features, and how to create a modern day game engine (for modern hardware), and modern consoles (PS5 and Xbox Series X).

I really wish that Amazon Lumberyard Team would "wake up" and "stop sleeping".  If you NEED HELP with a roadmap (and list of needed features), we have been screaming nice and loud in the forums (for the past 5+ years) but NOBODY at Amazon Lumberyard Team seems to be listening.

If you want to learn more about Dynamic Geometry (dynamic LOD's), NVIDIA gave a good presentation several years ago (2016 era?) about it, and this is the same technology that Epic Games is now implementing in their Unreal Engine 4.25 (experimental / beta builds from source) and also the upcoming Unreal Engine 5 (production builds).

NVIDIA's Asteroids Demo showcases Mesh Shader technology being used to create smart culling systems to manage the level of detail (LOD).  Creating dynamic LODs.

This is the same type of technology that Epic Games is working on (Mesh Shaders with a smart culling system, for dynamic LODs) which is what Epic Games is calling "Nanite".  Epic Games is creating a hardware agnostic version (doesn't require NVIDIA hardware), and is a software version that runs on ANY CPU or GPU.

You can learn more about Mesh Shaders with a smart culling system for dynamic LODs) here:
https://devblogs.nvidia.com/using-turing-mesh-shaders-nvidia-asteroids-demo/

It was discussed in 2016, and the hardware demo was available in 2017 (for researchers) and available to the public in August 2018 (and was discussed in great depth by Christoph Kubisch at 2018 SIGGRPH conference) which you can watch here:
http://on-demand.gputechconf.com/siggraph/2018/video/sig1811-3-christoph-kubisch-mesh-shaders.html

Is nobody at Amazon Lumberyard Team even following/watching what is going on with modern hardware? (and Techniques for Real-Time Graphics?) including Mesh Shaders?

This was available (research) since 2016, and has already been discussed (publicly) and been available (via NVIDIA Developer / GameWorks) since at least August 2018.

Here:
https://devblogs.nvidia.com/using-turing-mesh-shaders-nvidia-asteroids-demo/

This NVIDIA GameWorks demo is using a modified version of Unreal Engine 4 (UE4) back in 2016-2017 era, and the Asteroids Mesh Shaders Demo highlights the use and capabilities (as well as technical introduction to Mesh Shaders / Turing Mesh Shaders).  The Asteroids demo application can achieve very high frame rates using modern CPUs, but it also is optimized to move key performance bottlenecks of object list processing off of the CPU and into highly parallel GPU mesh shading programs.  Using an extremely large dataset comprising of several trillions of potentially visible triangles at any given time, the shaders efficiently eliminate primitives that will never be seen (culling) and shad only those contributing to the pixels displayed.

In the demo (by NVIDIA using a modified Unreal Engine 4.20), each individual asteroid model consists of 10 levels of detail (LODs), with the highest level containing up to 6 million triangles (per model/asteroid), and the largest asteroids (several kilometers in size) have an extremely detailed look (while up close), and you can inspect the triangle meshes by switching to wireframe mode.  Activating "Visualize LOD levels" paints each asteroid with a color, based on its current LOD (dynamic LODs), showing how the mesh shaders transition smoothly between LODs (dynamic LODs).  Finally, you can also "turn off" the "dynamic LODs" system, and then be able to select the display of any of the first 7 levels of detail (for comparison).

It's a great demo, and something that Epic Games and NVIDIA have been working together on for several years now (over 5+ years) and it's been demonstrated in 2018 (publicly) at SIGGRPH 2018 Conference in August 2018.  You can view the conference/demo discussion here:
http://on-demand.gputechconf.com/siggraph/2018/video/sig1811-3-christoph-kubisch-mesh-shaders.html

This has been around for several years now, and NVIDIA created a hardware-optimized implementation (for NVIDIA Turing GPUs) so that GPUs could be used to help to achieve very high frame rates (using either modern CPUs, or modern GPUs).
 
The Asteroids demo application can achieve very high frame rates using modern CPUs, but it is also optimized to move key performance bottlenecks of object list processing off of the CPU and into highly parallel GPU mesh shading programs.  Using an extremely large dataset comprising of several trillions of potentially visible triangles at any given time, the shaders efficiently eliminate primitives that will never be seen (culling) and shad only those contributing to the pixels displayed.

This is what Unreal Engine 5 (Nanite) is using/doing.  I hope this helps!

I hope I have answers all your questions/misunderstandings (as far as hardware/compression, and the use of new rendering techniques for real-time graphics, including mesh shaders).

I really wish that the Amazon Lumberyard Team wasn't so far behind, and would actually attend SIGGRAPH Conferences, and LEARN about modern hardware, and develop their Lumberyard game engine to support the very latest hardware (so that when the hardware is released, Lumberyard supports it), instead of being 5-10 years behind (in game engine / hardware support), making it IMPOSSIBLE to even use Lumberyard to create a "modern day" AAA-quality game.

Things like Nanite (Mesh Shaders) are things that Lumberyard should have been working on 4-5+ years ago, and it's difficult to understand WHY Amazon Lumberyard Team doesn't already have a working demo using this technology in Lumberyard yet?  C'mon guys, wake up!

We have been asking/demanding a current roadmap (so we can see what the Amazon Lumberyard Team is currently working on, and what is currently planned/expected for future releases) and we can't get ANYTHING from the Amazon Lumberyard Development Team.

It's impossible to develop a game (or even use Lumberyard for TEACHING/Instructing students) when Amazon Lumberyard is so incredibly far behind when it comes to game engine development (and supporting the newest/latest technologies, that are being used in next-generation consoles and modern game development).

SIGGRAPH 2018 - Mesh Shaders here:
https://www.youtube.com/watch?v=72lYVTlPfI8

Why You Should Use Mesh Shading here:
https://www.youtube.com/watch?v=PAwCqrtHxL4

Reinventing The Geometry Pipeline: Mesh Shaders in DirectX 12 (by Shawn Hargreaves) here:
https://www.youtube.com/watch?v=CFXKTXtil34

Compression of Geometry and simplification of geometry data structure by triangle template computation (by the mesh shader) are good/normal/modern day optimization tools/techniques, and it seems "odd" that Amazon Lumberyard Team doesn't follow along and is incredibly far behind in technology.

Advanced Mesh Shaders (by Martin Fuller) here:
https://www.youtube.com/watch?v=0sJ_g-aWriQ

You should probably reach out to the Microsoft Xbox Advanced Technology Group (at Microsoft) and have a talk with them, about needing help with Lumberyard (because nobody at Amazon seems to understand what they are even doing, or what is coming).  Ask Microsoft Xbox Advanced Technology Group (at Microsoft) for HELP with Amazon Lumberyard (they have already helped with adding support/features to Unity and Unreal Engine), maybe they can help you as well.  Because Lumberyard Team seems a bit "clueless" and don't seem to understand what is even going on in modern day game development (for console and PC).

Microsoft Windows supports ALL DirectX 12 Ultimate Features!  These features should ALL be implemented in Lumberyard!

If you want to take a college class, or learn more about Culling with Mesh Shaders, here is a good introduction and performance comparison class/course/discussion here:
https://www.youtube.com/watch?v=lijo-UqZu4Q

And you can watch the NVIDIA "Geometry Reinvented with Mesh Shading" here:
https://www.youtube.com/watch?v=rLEbO0Vrzz4

This was already previously discussed at SIGGRAPH 2018 (in August 2018), but was discussed again (more in depth) at a SIGGRAPH 2019 (NVIDIA presented talk) entitled "Applications of Mesh Shading with DX12", which explained how developers could take complete control over geometry processing.

Quick 12 minute synopsis can be found here:
https://www.youtube.com/watch?v=rLEbO0Vrzz4

Full one hour presentation can be found here:
https://developer.nvidia.com/siggraph/2019/video/sig918-vid

ONE Engineer and ONE single scupt artist spent 8 weeks working together to build this "Asteroids Demo", with a fully controllable space ship (and fully controllable scenario) in which a spaceship is best in space and has an endless cascade of asteroids.

The video demonstrates how "Catmull-Clark" subdivision surfaces, a custom meshlet export plugin, and stochastic dithering were used to deliver the finished product.

You can watch this full SIGGRAPH 2019: Applications of Mesh Shading here:
https://developer.nvidia.com/siggraph/2019/video/sig918-vid

You can watch the full video (by Rahul Sathe & Manuel Kraemer) on July 18, 2019 here:
https://on-demand.gputechconf.com/siggraph/2019/video/sig944-applications-of-mesh-shading-with-dx12/

A quick 12-minute excerpt can be found here:
https://www.youtube.com/watch?v=rLEbO0Vrzz4

But this is what is being used (what Epic Games is calling "Nanite") in the latest Unreal Engine 4.25 builds (experimental) and also will be production ready in Unreal Engine 5.

"Geometry Reinvented with Mesh Shading".

You can learn more here:
https://www.youtube.com/watch?v=rLEbO0Vrzz4&t=2m0s

The space ship was purchased from TurboSquid.

1 Sculpt Artist ( Nathan @REDACTEDUSER

1-5 NVIDIA developers did the creation of the Unreal Engine 4 demo (entitled "Asteroids") and this was completed in 8 weeks (from start to finish) and everything was built from scratch 8-weeks prior to the 2018 SIGGRAPH Conference, so that it could be discussed at the conference, and the Asteroids Demo was made available for download and given to everyone in attendance at the 2018 SIGGRAPH Conference, so they could use/run the demo and see the performance (on their normal hardware/laptops/desktops) and also compare the performance on newer NVIDIA Turing GPU hardware (with GPU hardware acceleration).  But NVIDIA hardware doesn't need to be used, but the demo is optimized to support modern NVIDIA GPU hardware (including Turing and NVIDIA RTX-based GPUs).

In the Asteroids demo (by NVIDIA using a modified Unreal Engine 4.20), each individual asteroid model consists of 10 levels of detail (LODs), with the highest level containing up to 6 million triangles (per model/asteroid), and the largest asteroids (several kilometers in size) have an extremely detailed look (while up close), and you can inspect the triangle meshes by switching to wireframe mode.  Activating "Visualize LOD levels" paints each asteroid with a color, based on its current LOD (dynamic LODs), showing how the mesh shaders transition smoothly between LODs (dynamic LODs).  Finally, you can also "turn off" the "dynamic LODs" system, and then be able to select the display of any of the first 7 levels of detail (for comparison).

You can learn more about the creation of the "Asteroids Demo" (using UE 4.20-4.25) by NVIDIA here:
https://www.youtube.com/watch?v=rLEbO0Vrzz4&t=2m0s

I really wish that the Amazon Lumberyard Team could "wake up" and create this same exact demo in Lumberyard and give us a Lumberyard "Asteroids Demo" exactly like this one created for Unreal Engine 4 (UE 4.20 - UE 4.25).  But this is the same technology that is going to be used in UE 4.25 (experimental) that is used in Unreal Engine 5 (what they are calling "Nanite") and Unreal Engine 5 will have the "production ready" version (final release).  But this has all been available for several years now (2016-2019) for technology demos, and it's quite surprising that NOBODY at Amazon Lumberyard Team even knows what is going on in the modern GPU rasterization pipelines.

You can watch a synopsis here:
https://www.youtube.com/watch?v=CqsFgksWkEs

Answer

Seems to be a type of REYES rendering.

Have to bear in mind that Unreal also now caters to the archviz and special effects markets, where this could be a big benefit to their workflow.  It's impressive that they've got this running in realtime on a PS5, but I'd be surprised if it's a practical choice for games anytime soon. It's a very expensive way of doing things.

Answer

I realized that, that was my bad I just remember nanite right when it was starting, and it was not even possible to get that to run on a console. You are right they do run smoothly on a console at this time.

Answer

John,

I REALLY appreciate your post!  Thank-you for LISTENING!  It's frustrating, and having been here (from day ZERO before Lumberyard even inked the deal with Crytek), I have been a BIG HUGE ADVOCATE of Lumberyard, and I was extremely excited (at the time) about the prospects of Amazon purchasing a CryEngine license, and allowing us to use it.  I was extremely "hopeful" and it breaks my heart to hear about the "internal struggles" that a large company like Amazon has to deal with (especially with "limited resources" as you say).

If you look at Epic Games, and look at the success of Fortnite (and PUBG), and just the money being generated from Fortnite is enough to "fund" the continued development of Unreal Engine.  Fortnite has been a "windfall" of income for Epic Games, and allows Epic Games to sponsor development teams, and purchase companies (such as Quixel) and give away things like Megascan assets (for free) for Unreal Engine development.

These things are HUGE for students and small Indie Developers (and Instructors) because it allows us to point our students in a particular direction (i.e. Quixel Megascans) and students have access to photorealistic 4K, 8K and 16K cinematic quality textures and assets (for free) that they can use for Unreal Engine development (and even commercial use) for FREE (as long as they are using Unreal Engine).

I wish we had this same type of "ability" with Lumberyard, and I do agree that the license for SpeedTree (many years ago) was a great idea, and helped with early Lumberyard Indie users, but I would think after several years that the Lumberyard team would have been able to develop their own dynamic vegetation system, and their own physics engine (similar to Unreal Engine's "Chaos" system), and their own VFX system (similar to Unreal Engine's "Niagara" system), and their own 3D spatial audio system  (similar to Unreal Engine's "Spatial Audio" system).

I personally don't have a problem with Wwise Spatial Audio (and I'm thankful that Lumberyard included a Wwise license for Lumberyard use), but I personally do like the "integration" that Unreal Engine provides where everything works together "seamlessly".  Which is something that we don't see with Lumberyard.

If you look at the new Unreal Engine 5 demo, you can clearly see how "nice" it is to have all of your systems created by one company (Epic Games / Unreal Engine) and see how each of the various systems (Physics/Chaos, VFX/Niagara, Spatial Audio/Audio Occlusion System, Streaming Textures, and Streaming Geometry/Nanite) all work together to create a nice "out of the box" development platform that has everything small Indie developers need!

The ability to import cinematic quality assets (such as Quixel Megascans) in 8K or 16K directly into the engine (without having to create ANY LOD's or optimizations) is HUGE!  For small content artists and very small development teams, this is HUGE!

My biggest fear is that Amazon is NOT focusing on what DEVELOPERS need, and instead they "bit off" more than they could chew by spreading themselves "too thin" by creating game studios and trying to "do too much" without even working on the core Amazon Lumberyard Engine FIRST!

This is like trying to drive a train (without even laying down the tracks).  The studios can't even use Lumberyard (to develop a AAA-quality game) in an "efficient" manner, without years of modifying the engine, and instead of developing games, they are each working on modifying the engine.

This should be done at the "core" level (i.e. Lumberyard dev team) and then one the engine is developed (and mature) and is similar (in feature parity) to Unreal Engine (and even Unity), then you can look at developing your own games (i.e. Amazon Studios).

But in my opinion, Amazon should "shut down" most of its studios, and instead concentrate it's efforts and resources on the Lumberyard Engine, and also offer "bounties" (like $50,000 bounties) to Indie Developers for great "photo realistic" game samples/levels.

This will do TWO THINGS, it will get the community EXCITED, and get developers "motivated" (to learn the engine, and use the engine) and will also create some "competition" (among students/Indie developers) and then with maybe 50 or 100 submissions, the community can "vote" on a winner, and then first place gets $50,000 and second place gets $25,000 and third place gets $15,000 and fourth place gets $10,000.  Then have about 20 "Honorable Mentions" (for 5th place thru 20th place) which can be just Amazon Gift Cards ($100-$150 gift cards, and Amazon FireHD 10's, and Amazon Echo Dot's) as "door prizes" for the 5th place to 20th place teams.

Then have these competitions FOUR TIMES per year.

Have a Q1 competition (with a set of specifications on a particular game sample), then have a Q2 competition (with a different set of specifications for a particular game sample), then have a Q3 and Q4.

Always have these listed ONE YEAR IN ADVANCE (or 1 1/2 years in advance) so Instructors can get students ready (and assemble small 7-11 person student "Indie" teams) and Instructors can get maybe 8-10 different teams enrolled in the competition, and get them working together (as a large group) creating assets and textures, and models, and preparing highly realistic environments (i.e. maps), and then the smaller teams can work on their individual game levels (using the same assets, textures, landscaping, etc.) and focus more on buildings, and interior assets, and gameplay/sounds/lighting/storyline, etc.

This way in 1 year to 1 1/2 years, we can "churn out" very nice AAA-quality game levels (fairly inexpensive) by using students, and this helps grow the community, and then the winning teams make their contributions available (to the community) for FREE UNLIMITED USE, so that anyone can download the projects, game assets, samples, etc.

In exchange the teams get their prizes, and everyone "wins".  Amazon can create a nice "Marketplace" (similar to Epic Games / Unreal Engine Marketplace) and then offer these projects/assets for FREE on the marketplace, and also work together with developers (Marketplace developers) to offer FREE game assets EVERY MONTH (free giveaways) identical to what Epic Games does!

Every single month, Epic Games has several game samples, maps, game assets, and various blueprint-based systems that Epic Games offers FOR FREE to the community (monthly).

It keeps developers involved and ACTIVE, and also keeps them coming back (every month) for the free assets that they can add to their account.

See here:
https://www.unrealengine.com/marketplace/free

See here:
https://www.unrealengine.com/en-US/blog/epic-releases-4-million-of-infinity-blade-assets-and-renews-featured-free-unreal-engine-marketplace-content

See here:
https://www.unrealengine.com/en-US/paragon

It's GREAT that Epic Games can give users $17 Million worth of Paragon assets, and another $4 Million worth of Blade assets, and has a great Epic Games / Unreal Engine marketplace and works with Marketplace developers to offer FREE monthly giveaways (every single month) so

See April 2020 - Free Marketplace Content here:
https://www.unrealengine.com/en-US/blog/featured-free-marketplace-content---april-2020

See May 2020 - Free Marketplace Content here:
https://www.unrealengine.com/en-US/blog/featured-free-marketplace-content---may-2020

See June 2020 - Free Marketplace Content here:
https://www.unrealengine.com/en-US/blog/featured-free-marketplace-content---june-2020

Just look at each of these links (listed above) and LOOK AT THAT CONTENT that is being given away for FREE (every month) and a small Indie Developer, or a student... those assets will make your eyes pop out of your head!

It allows very small teams (of 3-11 developers) to create AAA-quality games (levels) in under a year (often times in just several months) once a good pipeline is developed, small teams can start churning out high-quality levels in just one school year!

Unfortunately we have been asking for these SAME EXACT THINGS from Amazon / Amazon Lumberyard, but in 6+ years Amazon hasn't taken the development of Lumberyard SERIOUSLY, and instead has headed in all different/opposite directions (trying to hire employees, instead of using Indie/Community developers FOR FREE, and incentivize the community by offering "Bounty Programs" instead of trying to setup your own studios!).

Stop wasting millions of dollars on your own Amazon Studios, and instead cut ALL THAT MONEY and shut that stuff down, and instead focus on creating "Bounty Programs" (it's almost like "out sourcing" your development for game samples) to the Community!

It's MUCH MUCH MUCH CHEAPER to have Community Created Content instead of internally created content (that NEVER seems to get released to the public) and Amazon just continues to flush money down the toilet, and the community NEVER BENEFITS from anything that Amazon is doing (or spending/wasting it's money/resources on).

Instead offer "Community Bounty Programs" (for game sample development, tutorials development, youtube videos/tutorials development, public Wiki pages that the community can edit and work on, similar to Wikipedia for updating documentation, and adding links to game samples), and also a GitHub where we can upload our own game samples (for the community to use/share) and update links in the Amazon Lumberyard wiki page/tutorials (with links to our community developed game samples, youtube videos, and step-by-step getting started guides, etc.)

This way the community can help with documentation, videos, tutorials, game samples, etc.  We can have a "contributor agreement" similar to that of CARLA, so that students can "contribute" to the development of various community developed projects/game samples (such as CARLA).

That way ANYONE (new students, indie developers, etc.) can download any of these large projects, and use the assets in their own projects (free of charge), and also see nice AAA-quality large map samples.

When you have 3,200+ students "working together" (despite being from different universities, and being located all around the world/globally) but CARLA is one good example of what I would like to see for Lumberyard.

CARLA is developed using Unreal Engine 4, and is freely available for download.

See here:
https://www.youtube.com/watch?v=dvPhN0Gbb0w

The project has nice Unreal Engine 4 tools that allow you to import procedural map generation (using

Introduction to CARLA here:
https://www.youtube.com/watch?v=Z7PLkcY9Mtk

You can learn more about CARLA Talks 2020 here:
https://carla.org/2020/06/09/talks_2020/

CARLA Art Improvements: Environment and Rendering here:
https://www.youtube.com/watch?v=ZZaHevsz8W8

CARLA Core Implementations: Synchrony, Snapshots and Landmarks here:
https://www.youtube.com/watch?v=nyyTLmphqY4

CARLA Data Ingestion here:
https://www.youtube.com/watch?v=mHiUUZ4xC9o

CARLA - Sensors in CARLA here:
https://www.youtube.com/watch?v=T8qCSet8WK0

CARLA - Pedestrians and their implementation here:
https://www.youtube.com/watch?v=biQfgj2PQ-U

CARLA - Improvements in the Traffic Manager here:
https://www.youtube.com/watch?v=n9cufaJ17eA

CARLA - Introducing Scenarios Runner here:
https://www.youtube.com/watch?v=dcnnNJowqzM

CARLA - OpenScenario Support here:
https://drive.google.com/file/d/1g6ATxZRTWEdstiZwfBN1_T_x_WwZs0zE/view

I would love to see more "Community Involvement" (as well as students, researchers, etc.) using Lumberyard, but unfortunately Lumberyard is very difficult to "get started" with (very large learning curve), the source code is a mess, and build times are horrendous, and just getting a build to even complete (without errors) is a problem in itself, and Lumberyard is currently a "mess" (compared to how easy Unreal Engine 4 is to use, and the very easy to use "one click build" process) and how easy Unreal Engine 4 is to install (and update) and the nice Epic Games Launcher really makes it EASY to use (and to update) Unreal Engine 4, and to keep your projects (and dependencies) updated.

[quote="JohnVanDerBurg, post:11, topic:8609"]
Just please keep in mind that we also have limited resources (yes even at Amazon). It is not exactly fair to compare us to Unreal for example.
[/quote]

I would have a hard time believing (as large as Amazon is) and with as "deep pockets" as Jeff Bezos has, that this "mess" of a project, couldn't be "turned around".

[quote="JohnVanDerBurg, post:11, topic:8609"]
For example the installer that you mentioned as well as the build process.

Recently an effort has been made to fix a lot of issues with the current installer.
[/quote]

We need a completely new installer.  Look at Epic Games Unreal Engine installer as an example!  Look at the Epic Games Launcher!  That is what we need!  Something identical to this!

Personally, I would shut down all the studios (for now) and focus primarily on the game engine core itself!  Get Lumberyard to equal parity as Unreal Engine 4 and/or Unity and/or CryEngine 5.

Basically you want the "best of all worlds" in ONE game engine, and make that game engine Lumberyard!

That is what the PRIMARY FOCUS should be for the Lumberyard team.  As far as game studios, shut those down (temporarily) and move those developers (and content creators/artists) over to the Lumberyard Core team, and focus primarily on the engine development.

Attempting to create games with a game engine that is not even production ready (and is a complete mess) is like trying to "drive the train BEFORE you lay the tracks".  It doesn't work well.

It's FOOLISH to try and open game studios, and attempt to work on games when the Lumberyard Core engine is a complete mess (and is unfinished and not production ready). [And so many drastic changes are happening to Lumberyard core]

It's similar to what Chris Roberts' is having to deal with (at CIG).  Imagine having to "recreate" your game (for the most part) between versions (i.e. Star Citizen 1.0, 2.x series, and then 3.0) and the messy process of creating all that content and having to redo a lot of it, and re-writing the backend engine, and creating drastic changes, and then having to go back and redo a large portion of the content.  It's a mess (and very time consuming).

We need a "standard" and I was really hoping that Lumberyard (and not Unreal Engine) would have been that "standard" (engine) that serve as a great/best general purpose engine for large photorealistic modern MMOs.

But the Lumberyard team still hasn't dealt with the small tiny maps issues.  Unreal Engine supports 20km x 20km maps (and even larger/infinite size if you stream).

Lumberyard is still stuck on 4km x 4km maps.  This is absurd in today's day in age, and how can you even create a flight simulator in Lumberyard (without huge massive modifications/undertakings like what Chris Roberts/CIG is having to do).  This is not even possible for small Indie developers and small content artists/developers that just want an engine to "work" (out of the box) without requiring hundreds of developers (just to work on engine source code modifications).

The NUMBER ONE FOCUS should be getting the Lumberyard Core Engine refactored, add most of the CIG features (large world maps, planets, universes, 64-bit precision, planet editor, universe editor, etc.) and get these features completely re-written, cleaned up, refactored, and put directly into core, and also reach a feature-parity with Unreal Engine 4, Unity, CryEngine 5, and make Lumberyard a "hybrid" (cream of the crop) engine that has EVERYTHING that all these other engines have, but combined into ONE ENGINE!  Then release this under a Lumberyard 2.0 release (with a permissive MIT-license) and let the Community continue development!

That will attract users (and a large community), developers always want to use the BEST ENGINE possible (feature rich, most features, most/best tools, import tools, content creation tools, procedural generation tools, etc.) and also add the same features as Unreal Engine 5 (Dynamic Geometry / Nanite, Virtual Texturing, map streaming so you can have unlimited size maps and planets and universes, etc.).

Put ALL of these features into Lumberyard Core, and then make them available for students and Indie developers to use (and also once Core is more updated/feature rich) then offer "bounty programs" (to students and Indie developers) and let the community develop game samples/levels/assets that will be contributed to the Marketplace (for free) so other users/members can download them, use them, share them, and eventually allow marketplace members to post their content (for sale) similar to Unreal Engine 4 or Unity, so people can make a living developing models and game assets, and content for Lumberyard (that other developers can purchase and use in their own games).

This really "cuts down" on development time, and allows very small teams (7-11 developers) to create AAA-quality game samples, and even small games (in a fairly quick period of time, 1-2 years).

It's great for students and small Indie Developer teams, and it will help grow the Lumberyard community.

[quote="JohnVanDerBurg, post:11, topic:8609"]
please keep in mind things aren’t as easy as it seems. The devs we have are highly skilled and are up to date, I can assure you that. But we don’t always get to work on what we wish to work on as there are other things going on, and I know many of the issues you mention also frustrate us. For example the installer that you mentioned as well as the build process.
[/quote]

I understand, and this is extremely "frustrating" since Amazon is such a LARGE company, and they make it "difficult" to maneuver and do rapid development (like a small 7-man "startup" could do) because of all the bureaucracy, meetings, managers, overhead, and nonsense.  I also understand that "resources" are extremely limited, but for this reason I would suggest refactoring the engine and having a completely new codebase (that doesn't share any code with the former Crytek CryEngine) and is a clean new codebase, and release this engine as Lumberyard 2.0!  (under a permissive MIT-license)

This way the COMMUNITY can help with development of the engine, and other corporate sponsors (Intel, AMD, NVIDIA, Sony, Microsoft, Google, etc.) can sponsor additional developers, and we can get the core Lumberyard Engine developed, feature rich and improved.

[quote="JohnVanDerBurg, post:11, topic:8609"]
please keep in mind things aren’t as easy as it seems. The devs we have are highly skilled and are up to date, I can assure you that. But we don’t always get to work on what we wish to work on as there are other things going on, and I know many of the issues you mention also frustrate us. For example the installer that you mentioned as well as the build process.
[/quote]
 
Unfortunately this "happens" when dealing with large companies (such as Amazon).  It took us several years of SCREAMING AND YELLING just to get the source code posted to GitHub, and also took several more years of MORE SCREAMING AND YELLING just to get a decent forums!  (instead of Amazon's propriety garbage forums that were pure crap).

Unfortunately, even after 6+ years of SCREAMING AND YELLING, we still can't even get a public roadmap!

I think it would be TRULY BEST for Amazon to "transfer ownership" of their CryEngine licence to a "Community Project" (instead of Amazon ownership) and then allow the community to begin working on the engine and help "steer" the direction of the engine (similar to what Google/Apache does) and this would allow other companies to "join in" on the effort (look at GoDot).

See here:
https://godotengine.org/

This would be the ABSOLUTE BEST THING for the Engine and for the Community as well.

This would allow ANYONE (not just Amazon) to be involved in the core development of the engine, and would allow us to help shape the engine (based on what the COMMUNITY NEEDS/WANTS), and allows EVERYONE to help contribute to the development of the engine (not just a very small group of Amazon developers).

I believe the IMMEDIATE FOCUS would be to completely re-write / re-factor Lumberyard, into a new Lumberyard 2.0 (that is free from ANY former Crytek code/dependencies) and is a completely 100% new codebase / newly written engine that we can license under an MIT license (just like GoDot), and make this "Lumberyard 2.0" available (similar to GoDot) to the community under an MIT license.

That is the least possible restrictive license, and allows the community to do whatever it wants.  But at the same time, it also allows other companies (Intel, AMD, NVIDIA, Toyota, GM, etc.) to get involved and begin to use the engine (as well as Universities) and students (and Indie Developers) and slowly the community can get involved (in core engine projects / features / add-ons) and slowly develop an AMAZING ENGINE (and amazing content) that is ALL available FOR FREE (to anyone) for use.

This will help move everything along, and eventually lead to an engine that can BEAT Unreal Engine 4 (and Unreal Engine 5) and BEAT Unity, and BEAT CryEngine 5, and be community driven.

Similar to an "Android" (or "Linux") that grows.  I have looked at GoDot, but personally I had great hopes for Lumberyard, and in the 6+ years that Amazon has been leading this project, it's like watching a "sinking ship" that is just going nowhere.  Amazon doesn't seem focused on the core Lumberyard Engine, and instead is blowing money in all different directions on silly "internal" projects, and the Lumberyard engine continues to remain the same exact "mess" that it's been (for the past 6+ years) and the community is going NO WHERE (under Amazon's FAILED LEADERSHIP).

Unfortunately I blame the "poor leadership" at Amazon, and this is ALWAYS TO BLAME for projects that fail (and are not going anywhere).  With proper leadership, I believe this project could be turned around, and could be extremely beneficial to the community (as well as to Amazon).

The AWS, and Twitch features are great, but unfortunately those features are meaningless if the engine itself is not being updated/maintained and doesn't remain in feature-parity to other engines (like Unreal Engine 4 / Unreal Engine 5).

In my opinion, we need to move Lumberyard (away from Amazon) and into a completely separate entity (new entity) similar to the open-source GoDot Engine project.  A community driven project.

Then have the Amazon Lumberyard (core team) and developers (sponsored by Amazon) help contribute to the development of the new "Lumberyard Engine" (that is completely free of Amazon control / management).

The BIGGEST PROBLEM with the project is the FAILED Amazon Leadership (it's like a herd of cats just wandering around aimlessly with no real direction at Amazon).  Too many different failed internal projects, and the core Lumberyard Engine is still a mess.

Amazon doesn't seem to know how to make a game engine, or a AAA-quality game, and their FAILED LEADERSHIP is like a "herd of stray cats" wandering around (and wasting money).

Instead we need a Community-Driven project (like GoDot) that is NOT under Amazon control, and that way we have a completely community-driven Game Engine (under a very permissive MIT license) just like GoDot.

No strings attached, no royalties, nothing.  Completely free and open-source under the very permissive MIT license.

Then contributors can contribute (and sign a contributors agreement) and we can slowly turn "Lumberyard 2.0" into something GREAT.

This way AMD, NVIDIA, Intel, Google, and large universities (and Indie Developers/students) can ALL get involved and be a part of the project, and TOGETHER AS A COMMUNITY we can develop something great.

Answer

[quote="Zaphod_Beeblebrox, post:6, topic:8609, full:true"]
The next gen is going to be interesting, particularly in regard to the streaming speed of the console SSDs, and how game designs could change. I could imagine that it’s going to be PCs actually holding back cross-platform games in that regard - it will be difficult to rely upon the PC user base having a fast enough storage solution, due to all sorts of different configurations.

I suspect it will be only in PS5 exclusives that we really see that potential pushed.
[/quote]

For the near-term, I actually agree with you!

Highly optimized games (optimized specifically for the PS5 hardware) will probably be better than many/most PC games (for the near future), due to the baseline hardware being AMD Zen3 Ryzen 7 3700X (8C/16T) and the 12-channel VNAND controller, allowing for 5,500MB/s read/writes which is great for a console!  Plus having a PCIe4 NVMe M.2 socket (for future PCIe4 NVMe M.2 expansion).

Sony made some GREAT MOVES with the development of the PS5!  Truly amazing!

As for Microsoft, I was extremely impressed with the XBOX Series X (a few weeks prior) and it wasn't until watching the fully Sony PS5 Developers Conference, that I realized that Sony is actually correct!

Sony actually did a much better job with the PS5, and the decision to create a proprietary VNAND SSD controller (for I/O) and still having / using the standardized PCIe4 NVMe M.2 socket for expansion (much better / faster option then Microsoft's decision to go with a proprietary expansion).

I was extremely impressed with the Microsoft Xbox Series X, but after seeing the Sony PS5 developers conference (and watching both), I can honestly say that Sony made the better moves (this time around) with the PS5 hardware.

I believe the PS5 will be in a class (all by itself) for the short term!

With PC's trying to play "catch up" for at least one year.  Plus I don't think you can even build a PC (with the components used in a PS5) for the price that Sony is selling the PS5 for!

So truthfully, it's MUCH CHEAPER to actually use a PS5 (instead of building or buying your own gaming PC) at least for the next year or two (till at least Zen4 is released with PCIe5 and newer PCIe5 NVMe VNAND controllers that support 4 lanes of PCIe5 with 14,000MB/s read and writes!), but that won't be till at least late 2021 or early 2022.  Zen 4 is scheduled for late 2021 (fall 2021) but we don't even know if Phison will have the PCIe5 controller chips into the hands of the SSD makers (before the AMD Zen 4 launch) and it will be interesting to see if we even have PCIe5 NVMe SSDs at launch (fall of 2021).  But if we do, that will be a "game changer" for PCs!

Just imagine 14,000MB/s read and write times on a PCIe5 NVMe M.2 flash drive!  Plus 8TB (and larger) storage on a single M.2 NVMe!  Just imagine putting TWO of those in a RAID0 configuration (similar to what Apple does with the Macbook Pro) and having 28,000MB/s read and write times!

Combined with a nice AMD "Big Navi" 6800 XT series graphics card, and you'd see some amazing "killer performance" on cinematic quality games!

But unfortunately, we need the game engines to be fully optimized for this new upcoming hardware (multi-threaded physics, etc) and I just don't see this currently happening in Lumberyard (yes, Chris Roberts and CIG have started working on doing this for Star Citizen, but it's no where near ready and still NOT making it's way back into core Lumberyard, as far as I can see).  So these are MAJOR problems for Lumberyard (as a project).

Lumberyard needs a LOT LOT LOT of work!  I really wish that Amazon Lumberyard Devs would work closer with CIG, and take the CIG codebase and take all of the features that CIG has added to their fork, and I wish that the Lumberyard Team would try to neatly rewrite and refactor it, and put it into core Lumberyard.  So we can ALL have these features in nice cleanly written Lumberyard 2.0 core! (which I was hoping to see an announcement for about 1-2 months ago, with a tech demo for next-gen consoles).  But it didn't happen, and I'm beginning to believe that Amazon Lumberyard might not even have a Lumberyard 2.0 even in the works right now.

Plus Lumberyard core devs really should have been re-writing and optimizing the Lumberyard engine for PS5 and Xbox Series X (over a year ago), to prepare for next-gen PCs and consoles, which clearly isn't being done (as of yet) [as far as I know] and doesn't even seem to be on the roadmap (as of yet) [as far as I know] and we still don't even have a public roadmap!  LOL!  😂

But I do agree that "next gen" is going to be interesting, but unfortunately only Unreal Engine seems to be "prepared" for it, and only Unreal Engine 4.25 (and Unreal Engine 5) are currently designed for the next-gen consoles!  (and next-gen PCs).

Epic Games (Unreal Engine) has been putting quite a bit of effort in refactoring Unreal Engine 4.25 (including Epic Games own internal **Chaos** physics, and **Niagara** Visual Effects (VFX) system, UE4's/UE5's Virtual Texturing system, and **Nanite** virtualized geometry (dynamic geometry) system!  (with automatic LODs and culling).

But unfortunately without a public roadmap, I'm just afraid that the Lumberyard Dev team is "lost" and "not prepared" and "not forward thinking" when it comes to game engine development.

Amazon Lumberyard is several years behind Unreal Engine (already) and I'm afraid that when UE5 is released, that it might be the "nail in the coffin" for Lumberyard.

[Lumberyard might end up going the way of GoldSrc!](https://www.youtube.com/watch?v=ut4DHtcs3Qs)

I don't see a completely re-written "Lumberyard 2.0" on the roadmap, and Amazon has been completely SILENT for the past 3 years (at GDC and SIGGRAPH) so it doesn't seem like Amazon has anything new updated to compete with a highly optimized and completely re-written UE5.

It doesn't seem that Amazon is ready for "next gen" and this is a major problem, and just based on past history (with how long it takes to get updated Tools support, (i.e. 3ds Max 2017/18/19 and Maya 2017/2018/2019, etc.) and even getting Visual Studio support updated (usually takes Amazon 3-4 years) and it's quite possible that Amazon Lumberyard might not be ready for "next gen" consoles till at least 2025, which by then would make Lumberyard fairly outdated (and obsolete).

Without a public roadmap, there is no way to tell what Amazon Lumberyard devs are even doing (or lack thereof) as far as Lumberyard development, but it's certainly NOT looking good for Lumberyard's future (as a game engine).

Crytek will have CryEngine 5.8 updated (or possibly a CryEngine 6 release) with PS5 and Xbox Series X support probably way before Lumberyard.  Unreal Engine 4.25 is already ready for PS5 and XBox Series X console development!

REMOVEDUPLOAD

Unfortunately we don't see this same type of "forward thinking" with Amazon and it's Lumberyard Engine.  Nobody at Amazon was working with Sony (years in advance) on future console development.

Amazon's approach seems to be "Let's wait till the console hardware is already released FIRST" and then we'll spend 2-3 years developing for the new hardware, and then we'll have a game engine in 2026-2028 (in 6-8 years from now) that is "fully optimized" for the new next-gen hardware (that was released in 2020).  While game developers are expected to twiddle their thumbs for 6-8 years (or do all of the optimizations and complete re-write of Lumberyard themselves, like CIG and Chris Roberts is doing, with a $300+M dollar budget).

So in the near term (at least the next 6-8 years, possibly even 10+ years) it seems that Lumberyard might be "dead in the water" when it comes to being fully optimized for next-gen consoles and next-gen PC's (Fall 2021 era with AMD Zen4 and PCIe5 NVMe).

Amazon is always 5-8+ years behind (the competition) when it comes to game engine development, and tools development.  So I am NOT "holding my breath" to see Lumberyard truly optimized for next-gen PC's or next-gen consoles, and for this reason... I put my faith in Tim Sweeney and Epic Games (for at least another 5 years) especially with Unreal Engine 4.25 and Unreal Engine 5 (versus the current state of an old archaic 2008-era CryEngine 3.8.1 / Lumberyard fork).

It's NOT looking good for Lumberyard.  😥

For this reason, I have been "SCREAMING AND YELLING" for a public roadmap, so as developers we can at least SEE what is going on, and SEE what is happening (and make good judgement calls as to which engine we choose to develop with/on).  Especially for next-gen console, and next-gen PCs!

Lumberyard is getting "lost" in the weeds, and is just falling farther and farther behind.  WAF is a complete joke, and several hour build times are just a "thing of the past" when it comes to modern-day game engines like Unreal 4.25 or UE5!

Even Unity kicks the pants off of Lumberyard in build times, and I'm afraid "Team Amazon" (Lumberyard) is just NOT READY for next-gen consoles and next-gen PC's.

These tools should already be in our hands (last year) and Unreal Engine 4.25 already supports PS5 and Xbox Series X game development.  Microsoft and Sony worked together hand-in-hand with Epic Games to ensure that Unreal Engine is the "de facto" game engine for next gen consoles.

Sony even used Epic Games (Unreal Engine 5) specifically for it's Tech Demo and the PS5 launch!

Microsoft has already been using Unreal Engine 4.25 (and now Unreal Engine 5) for over a year already, for it's development of Hellblade II (by Ninja Theory) for Xbox Series X consoles.

See here:
https://www.youtube.com/watch?v=pp98SUBZYH8&t=8m52s

This is a great interview with Tim Sweeney (discussing UE5) and the capabilities of the new engine!

Amazon is just too far behind, and it doesn't seem like Amazon is prepared (or even ready) for next-gen consoles or next-gen PC's, and haven't been putting any thought into creating a Lumberyard 2.0!

Answer

The next gen is going to be interesting, particularly in regard to the streaming speed of the console SSDs, and how game designs could change. I could imagine that it's going to be PCs actually holding back cross-platform games in that regard - it will be difficult to rely upon the PC user base having a fast enough storage solution, due to all sorts of different configurations.

I suspect it will be only in PS5 exclusives that we really see that potential pushed.

Answer

[quote="Zaphod_Beeblebrox, post:6, topic:8609, full:true"]
The next gen is going to be interesting, particularly in regard to the streaming speed of the console SSDs, and how game designs could change. I could imagine that it’s going to be PCs actually holding back cross-platform games in that regard - it will be difficult to rely upon the PC user base having a fast enough storage solution, due to all sorts of different configurations.

I suspect it will be only in PS5 exclusives that we really see that potential pushed.
[/quote]

From what I have seen so far, both consoles (Xbox Series X, and PS5) are both capable of real-time photorealism (using Unreal Engine 4.25 and Unreal Engine 5).

The increased throughput (of 3,400MB/s for XBOX Series X, and 5,500MB/s for PS5) are both sufficient for real-time photorealism (especially when used with [Kraken or Leviathan compression](http://www.radgametools.com/oodle.htm))  It will create some insanely high I/O's for large data transfers (large massive geometry and insanely large cinematic quality 8K and 16K textures) and the culling of geometry is a huge feature (that even Chris Roberts and Star Citizen have been struggling with trying to implement).

I just don't see this coming (anytime soon) for Lumberyard, but it's definitely done and finished for Unreal Engine 4.25 and Unreal Engine 5 users that want to create games for PS5 and Xbox Series X!

I just expect new "next gen" (cinematic quality) AAA games to be in the 100-150GB range (even with Kraken compression).  Most likely with 1TB (or more) worth of textures, models and assets, that are compressed down to about 100-150GB's (using Kraken compression or Leviathan compression) and then uncompressed (in real-time) on the fly (which gives it even faster I/O performance, almost ten fold!)  So in reality, you are moving well over 2GB/s (raw) and about 20GB/s (compressed) using both compression and the performance of the newer PCIe4 NVMe SSDs!

Keep in mind that both consoles are using AMD Zen3 processors (nearly identical) and both support PCIe4 NVMe.  It just that Sony implemented their own 12-channel VNAND controller, which is what giving the PS5 the faster I/O throughput/speeds (at this time) but both are using 4 lanes of PCIe4 for throughput, so in theory... Microsoft could release future external storage (using the proprietary connector) for faster drives (in the future), but personally I think Sony went a much better route by using a standard PCIe4 NVMe M.2 connector (so you can use a standard PCIe4 NVMe that is "Sony Approved" for future storage expansion) and with 4 lanes of PCIe4, there is a potential for 7,000MB/s (read and write speeds) once we have better VNAND controllers (which are coming from Phison and are already available in low-volume production, and should be available in high-volume production later this fall, so it's quite possible that next-gen PCIe4 NVMe M.2 drives will most like support 7,000MB/s read and write times (for PCIe4 NVMe using AMD Ryzen Zen 2, Zen 3 and Zen 4 processors) that all support PCIe4.

Intel is completely crap right now, and is really far behind (still using PCIe3 for years now), but AMD is way ahead of the game, and PCIe4 is going to be a game changer for next-gen consoles and next-gen PC's.

Later next year, in 2021 AMD is releasing Zen 4 which supports PCIe5!  So it's most likely that we'll begin to see PCIe5 based M.2 drives support 14,000MB/s for newer "next-gen" PC's (2022 and later).

Combined with newer graphics (Big Navi and later) and 6800 XT's and 6800M's (for laptops) and possibly a [6900 XT (Big Navi) coming this fall](https://www.youtube.com/watch?v=H8W_0gFzRGk&t=1m53s), I expect to see better graphics on next-gen gaming PC's (in late/fall 2020) and PCIe4 NVMe is already available for current AMD Zen 2 and Zen 3 processors/motherboards (with 5,500MB/s read speeds and 4,400MB/s write speeds) which is identical in performance to current PS5 specifications.

Throw two of them in a RAID0 configuration, and you already have 11,000MB/s read and 8,800MB/s write speeds (in current AMD Zen 2 and AMD Zen 3 processors/systems).

https://www.youtube.com/watch?v=H8W_0gFzRGk&t=1m53s

[quote="Zaphod_Beeblebrox, post:6, topic:8609"]
I could imagine that it’s going to be PCs actually holding back cross-platform games in that regard - it will be difficult to rely upon the PC user base having a fast enough storage solution, due to all sorts of different configurations.
[/quote]

I would just view the current/previous gen PC hardware (as PS4 / Xbox One type hardware/class), and I would view the upcoming fall 2020 AMD Zen 3 (and 2021 AMD Zen 4) as the "next gen" hardware (that is comparable or faster than PS5 and XBOX Series X) and what I would call "next gen" hardware.

As a developer, I would be focused on Next-generation hardware (right now) because next-gen consoles are coming this fall (PS5 and Xbox Series X) and "modern" PC hardware is already here (NVIDIA 2080 Ti, and PCIe4 NVMe with 5,500MB/s read/writes).  Plus even faster NVMe coming this fall (with newer Phison controllers supporting 7,000MB/s read/writes for AMD Zen 3 PC's).

I do agree as game developers that we should already be developing for photorealism and next-gen hardware NOW (PS5 and XBOX Series X consoles are already available for developers) and will be available to the general public this fall.  But anyone can start developing a "next gen" game using Unreal Engine 4.25 (right now) and Tim Sweeney confirmed that Unreal Engine 4.25 games compile and build (and run on) PS5 and Xbox Series X right now!  (so you can start developing NOW for next-gen consoles using Unreal Engine 4.25 and later).

Unreal Engine 5 will have Nanite (out of beta and in production).  But UE5 is a real game changer!

I'm very afraid (without seeing any type of public roadmap) that Amazon Lumberyard Team is probably "not prepared" for this, and it doesn't seem that Amazon Lumberyard Team has been "working closely" with Sony or Microsoft (over the past 2-3 years) on next-gen console development, so it's most likely that Lumberyard is "not ready" for next gen consoles, and isn't optimized for next-gen PCs or next-gen consoles, so more than likely it might be another 3-4 years before Lumberyard is completely re-written and optimized for multi-core processors and multiple GPUs (highly parallel).  I believe Amazon Lumberyard would have to completely re-write the physics engine (similar to what Epic Games has done with their new Chaos physics engine) and also Lumberyard would need quite a bit of work (which I believe Chris Roberts and CIG / Star Citizen have been working on for quite some time already) but Lumberyard is definitely not ready (right now) and most likely not anytime soon for next-gen consoles or next-gen PC's.

The performance would be terrible trying to use a legacy game engine like CryEngine 3.8.1 (or Lumberyard which is based on old archaic legacy CryEngine 3.8.1).

We don't have a public roadmap and no way to see what is being worked on (in the future) so there is no way to tell what Amazon is currently doing (behind the scenes) but based on my gut instincts (and how long it takes to even get development tools updated, which is normally 3-5 years behind current tool releases) so it might be a long while (probably 2023 or possibly even 2027 or maybe even 2035 timeframe) before we see a version of Lumberyard that is highly optimized for next-gen console and next-gen PC hardware (with a completely new physics system (Chaos), new VFX (Niagara) and new dynamic geometry (Nanite) and virtual texturing (that UE4 and UE5 currently use).

So we just won't be seeing "cinematic quality" in Lumberyard for quite some time (at least not till Lumberyard is completely re-written from scratch, and optimized for modern "next-gen" hardware).

I don't know if we'll ever see the "tight" systems integration that you see in UE 4.25 or UE5, coming to Lumberyard.  It would most likely require a completely new re-write (similar to what Epic Games has done) of the Physics system, VFX system, definitely streaming textures and global illumination and possibly even a re-write of the spatial audio sound (similar to what Epic Games has done).

Epic Games has been developing game engines since 1996, and they are NOT a "newcomer" to developing game engines.  Tim Sweeney knows what he is doing, and UE5 is definitely the near-term future (and what game developers should be using right now) for current-gen PC and next-gen console (which will be released this fall, and will be available before your game is completed, so UE5 is what most developers should be using right now for game development).  Sony, Microsoft and most studios are using a production-ready version of UE 4.25 or UE5 for development.

Without seeing any Tech Demo's from Amazon, and no news of a new Lumberyard 2.0 release, it seems like Amazon doesn't have anything in the pipeline (near term) as far as I know.

This is WHY we need a **public roadmap** for Lumberyard!  😉

Epic Games is extremely forthcoming (and they tell/warn developers several years in advance) especially at GDC or SIGGRAPH, but Amazon doesn't seem to be present and

But Amazon Lumberyard seems to have "fallen off the map" as far as development is concerned.

The last time Amazon even spoke at SIGGRAPH was 2016 (from what I recall) and I haven't seen anything since 2016 (from Amazon) which leads me to believe that not much is happening at Amazon.  If things were being worked on, and a new Lumberyard re-write was near completion, or if Amazon had new technology (comparable to Nanite) then I believe Amazon would have discussed it several years ago at SIGGRAPH (in 2016-2017 era) instead of waiting till 2020, or 2021 to announce (and show a tech demo) of something.

Unfortunately it's not looking good for the Lumberyard Community or "Team Lumberyard".  At least not in the near future, and without a roadmap it's very difficult to even know how long it will be before we see a complete re-write of Lumberyard (Lumberyard 2.0) that is highly optimized and designed for highly threaded muticore processes and modern GPUs (ray tracing, etc.) and Global Illumination, as well as updated Physics system, and a **dynamic geometry system (identical to Nanite)**.

I just don't think it's coming anytime soon, or else we would have heard something at GDC or SIGGRAPH at least 1-4 years ago, and we haven't heard a single peep from anyone at Lumberyard for 4+ years now, so it's mostly likely that Lumberyard isn't ready for next-gen consoles (and most likely won't be ready until it is re-written and highly optimized, and possibly a Lumberyard 2.0 release).  But without a public roadmap, we really don't/won't know anything from Amazon.

Amazon has never been very "forthcoming" with their developers conference announcements and we just don't hear much or see much from Amazon.   😥

I would think that if Amazon had something, that they would have already "shown it off" by now.  Or at least I would hope.

Nanite technology in Lamberyard engine

Relevant content