New game like render engine in the works?
cain-x
Posts: 186
Information from this DAZ video:
Going by the comments there, it seems DAZ Studio may be getting a new game render engine like eevee in Blender? Does anyone have insight into this and what the roadmap looks like? DAZ did reply that something is being worked on and check back for updates.
Post edited by cain-x on
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
interesting maybe an Unreal live link?
also the biggest issue for me using other softwares is actually instances and having to convert them to mesh
nearly everything instanced now in sets
if there is a way to use the instancing mechanisms of other softwares with DAZ sets it would be a game changer, even Octane render cannot see collapsed instances and uncollapsing them uses too much RAM.
Likewise Carrara, I have to manually add duplicated replicated stuff, figuring out where difficult iClone I can actually place duplicates at the instance placeholders but a few hundred it gets tedious
Hmm. I sure hope they don't start leaving us photoreal people out in the cold like they did to the 3DL folks!
Daz commented twice under thread "We're actively working on this update. Be sure to check back in soon.". That was around March 23, 2020.
Haven't seen anything. This is seriously exciting if true. Though will have to see form that it takes.
BTW, nice promo vid by Daz. And Thanx OP.
The Changelog for DS Beta 4.12.1.97 initially had listed the implementation of Google Filament which is indeed a real time render engine. As far as I remember there was something written abount material conversions from iray to filament as well. I can´t say more about it since I didn´t even know this engine exists until I did a quick search.
Unfortunately it has been removed before the release of this version of DS. It probably was not ready yet for publishing. But there defnitely seems to be something in the works.
well that looks pretty awesome
funny I was watching his video on Ogre the other day and even suggested it somewhere in this forum
Filament looks really nice for archviz, but they don't show renders of organic surfaces like skin, hair or leaves and that. Still. Looks really promising.
Checked out https://github.com/google/filament
It looks very promising. What would be so awesome is if it would allow for offline or network rendering. Perhaps even offload rendering of animations onto a different machine. I was hoping RIB files would contain animation (and is supported in the spec) but DAZ's implementation is broken.
I dunno. This would signal a frightening change in focus for Daz that I do not welcome. I don't want the PAs who create the eye-droppingly gorgeous too-heavy-for-realtime models on which I depend to be tempted to do anythind else but. I use Daz instead of iClone for a reason.
I can't afford iClone but I'd like the choice to render fast (especially for animations) or slow (high quality stills). To do so within DAZ Studio would be ideal as not all of us have the time or ability to transfer everything over to Blender to do it for free. I keep trying that but my efforts have not produced anything acceptable yet.
it is called choice
because some people want one thing does not mean things should be denied others
some hate Dforce, fiberhair, iray, 3Delight, HD morphs etc etc others want these things plus bullet physics, particles, VR, vertex modelling capabilities, you name it
giving people options should not be frightening to those who don't want them
and yeah I use mostly Carrara and iClone and other rendering engines because DAZ studio lacks most of what I want
sadly more and more content only works in DAZ studio which irks me a lot, not returned much yet but if it gets to that point I will be buying less and less
a faster render engine would keep me from spending at the Unreal Marketplace etc more
Oh yeah ... gimme all those! Well, perhaps except for vertex modelling as I would rather they update Hexagon (which has a bridge).
In an ideal universe, it wouldn't be.
But there are finite resources, especially time.
Look at the state of 3DL, for instance. Thank goodness I use Iray. But 5 years ago you might have said the same thing to those folks, and today they're basically SOL.
My time is finite and I spend a lot of it waiting for renders to complete.
But it's a zero sum game. More of something you don't want means less of something you do.
I was just expressing my opinion, already aware that it is completely irrelevant. It can't be used to infer that I would deny anyone of anything.
That would only be true in a universe where the only determinant for what features get implemented were that someone wanted them. There are always more desired features than developer man hours to implement, test, and document them, and so they must be divided into "what we're going to do" and "what we're not going to do". It is completely rational to fear that something you want might be put into the latter group.
Anyone has the right to have their own personal prioritized list of features. It is purely personal.
That was my point, precisely. Without Blender, I could get by with Maya. Without Motionbuilder, I could also get by with Maya. But if Daz were to stop making high poly characters, I'm out of business, so to speak.
But that's me. Just me. I really did not intend to force my opinion to resonate with anyone else.
Deleted wierd double post.
I don't really understand the development issues but I was under the impression that DAZ Studio will have Google Filament at some stage. This is Open Source so how much work does it take to integrate it? I've seen Vulkan mentioned too (I believe it was you who mentioned it) - is that an alternative or is it part of the Filament architecture?
As far back as I can remember with DAZ Studio, people have been demanding faster renders. People spend huge sums on new and faster GPUs so there is clearly a groundswell of support for something that would meet the need, if only, at the moment, to enable animation which is a pipe dream for those attempting to render hundreds of frames in IRay. Real time rendering could take DAZ Studio to a much bigger market and maybe draw some investment from the big boys like NVidia.
Of course, I can't say. Maybe we could ignore all the other variables like support levels and consider how long it took to get IRay support. I don't really know how long that was as it was already here when I showed up, but I don't imagine it was overnight.
It's a lower level framework, like OpenGL, that render engines like Filament could be based on.
Fair enough. But the realtime renderer is certainly going to, in some way, lack the realism that ray tracers produce. There's no free lunch. You're going to be giving up something. For a lot of people, the trade-off will be worth it, and in that sense, the choice will be good.
My only concern is that we can't have it all and, personally, I don't want to see a move towards lighter models intended for realtime rendering because I've seen what those models look like. I came to the conclusion that I didn't want a lighter model or a less faithful render engine, I wanted a faster render box.
But this is just my personal opinion and I'm not trying to yuck anyone else's yum.
some of us cannot afford high end computers either
The Iclone realtime Character creator Base male is 28,918 faces.
Your beloved Genesis 8 female is 17,776 faces.
She is already a realtime model in terms of polygon count.
She looks "better" than most other realtime models because of Daz subD ,propietary HD Details, JCM's, a well made clean quad topolgy and 4K textures in a high quality pathtracer running on purpose built GPU's.
And with all of that in operation ,she still drops to her realtime base res during posing.
A realtime engine ,in daz studio, would be not require reducing the Genesis 8 models in any way only an option to switch from 4K textures to something more suitable and cam Distance based LOD optimizations that exist in Unity &Unreal to reduce draw calls and texture recompilings
People not using the realtime engine could forego these realtime optimizations and carry on as usual for IRay stills.
G8s are low poly because they don't have to be high poly, not to make them lighter. DQ skinning, JCMs, and what some people are calling JCJs, are all computationally expensive, but they're what make G8s so gorgeous.
At 17K, and without all the innovation that G8s bring, they'd look worse than iClone because what people erroneously refer to as "base resolution" is nothing of the sort. There is no base resolution in Daz Studio; it's a SubD cage that does not represent the artist's vision for the character, and that MUST be subdivided, even at level 0. So a presentable un-subdivided model is having it both ways, and making one would represent a change in focus.
And that's what I don't want to see. I like my G8s smart and heavy.
well thats odd
my HD subdivision 4 characters still look detailed in openGL and render fairly fast
just sadly transmapping looks awful otherwise I would use it more, I can fake drop shadows in postwork
so honestly don't see how a game engine couldn't render highpoly mesh
what I cannot do, in iray, is render more than fits on my 6GB 980ti Nvidia graphics card
that is what a game engine might change if it's not limited to Nvidia hardware as I also have onboard Vega graphics other programs can use as well as a CPU
I don't think it's odd. I think we're not using the same definition of "realtime".
You would still be free to crank up your SubD/HD and use 4-8k textures and buy more &more& more hardware from NIVIDIA. and completely ignore the game engine optimized versions of the content
Reallusion did the opposite when they added IRay and the ability to set higher levels of subD for the CC3Avatars.
A tiny few embraced IRay for still renders in Iclone but the majoriy (the animated filmakers) were not affected by Reallusion offering some of their user base a different option.
I have moved completely to an Iclone to Blender/ EEVEE Pipeline and certainly wont be spending another six years to make one animated film, as I did with my old C4D(now abandoned) based rendering pipeline.
And I have the Option of switching over to Cycles or Radeon Pro Render plugging in some ridiculous 4-8K maps, turning up my SubD modifiers and sampling and render Hi res stills for product or Arch vis or portraits in a FREE software.
so any notion that one type of rendering option will be to the detriment of the other, is completely false... this is not 2006.. it is not a Zero sum game. The game industry is bigger than Hollywood and Music combined. an even non gamed dev artists/end users are beginning to see the value in near realtime viewport performance with FREE options like UE4, Unity&Blender EEVEE. All of which BTW have their own thriving content ecosystems and do not need Daz.
For any 3D content related company to ignore this is a fools gambit.
You cannot use the word "version" while arguing that it's not a Zero-Sum Game.
The rest of your post was predicated both on the caprice of decisiom makers you cannot influence and other things that cannot be known. I don't know when presuming to understand Daz's strategy has ever worked. I think I will be better off making conclusions from the object truths I've learned "the hard way" about the internals of how the Genesis 8 framework works and its limitations from working with the C++ SDK.
I used the decimator extensively for low res backgrounders in my feature length Animated Film "Galactus Rising"
I would need to see some actual production Data to beleive that The very existance of these end user OPTIONS have stopped or slowed the production of "Heavy" Genesis 8 content with HD morphs and 4-8k textures.
Daz once Had a game content branch called "morph3D" selling game optimized Genesis 2 content called "MCS" for Unity
Then rebranded it "morph ID"
Then rebranded it "the Oasis" IIRC
Now it is Called "Maketafi"( i'm not kidding)
All were/are dealing with game ready figure assets and web Avatars. to sell to those markets.
The market trends and potential for growth will determine "the focus" ..not you.. nor I.
If indeed Daz has publicly stated an interest in supporting the Open source Realtime "Filament " engine, it is reasonble to assume that they will at some point offer content optimized for that engine as they are in the 3D content business ..yes?? .....wether you like it ....or not.
since people keep demanding higher resolution more detailed mesh and HD morphs I really cannot see that trend ending because of a render engine option.
the only reason I actually want one is because more and more features cannot be exported from DAZ studio to use in other softwares or I would as I already do.
@DAZ devs : if you indeed upgrade the viewport with some realtime 3d shading it would be lovely. I liked the proper reflection display on the default shaders and I always wondered why you wouldnt use the same display type in the viewport for iray materials.
If you indeed upgrade the viewport please make sure to have iray shaders behave like these in the viewport to get good feedback before hitting the render button. Should help alot with material setups etc.
But that would still be a G8, dependent on the facilities provided by the framework. And the framework is decidely NOT realtime friendly. The polycount is not what make G8s heavy, it's the DQ, JCMs, JCJs, and SubD. Calculating the blended position of a vertex is a Sum of Products operation that CPUs can do 8 at a time, and GPUs by the thousands. The poly count is not the problem.
But if you remove what is the problem, what is left is a 17K character model that, shorn of the innovative framework that makes a 17K model be able to work, is inappropriate for just about anything. Even despite the quality of its topology. There's going to have to be another model because Genesis 8 characters don't even have a base resolution model.
There is a cost to having that OPTION. The genesis framework does not even have a "base resolution". It has a subd cage, which does not represent the artist's vision for the character, that MUST be subdivided, even if only at level 0. A usable base resolution model expected to be aesthetically pleasing without aby of the G8's innovations is a much different model that will not just spring into existence. I don't need production data to understand how the passage of time works; an artist can only do one thing at a time. It's going to take longer and he/she may opt not even to do one or the other. This is where the mere existence of the OPTION could affect the availability of the kinds of models I want.
Never have I even implied that my personal desire mattered to anyone nor anything beyond myself. But I'm sure you appreciate that market trends and potential for growth seem like empirically observable things, but there are many places for subjectivity to creep in between the business intel analyst on the ground and the CEO. Companies do things seemingly contrary to "what they should obviously do" all the time, and miss these signals altogether.
And holding man hours to produce it constant, the production of that content could reasonably be expected to affect the production of other types of content. Yes? And I'm afraid that you will have to point out where I claimed that my personal desire would have any effect at all on anything. I merely expressed my personal desire in the hopes of spawning conversation, and I suppose I succeeded better than I would have thought.