Prophecy2k
Veteran
Dreams : create, share & play [PS4, PS5]
- Thread starter TEEDA
- Start date
I doubt fabulously well. Automated tesselation down into triangles will be messy, going by every other automated free-form modeller that creates lower res versions (try exporting a NURBS or SDS from your favourite modeller to a triangle format for use in Unity/UE). As Laa-Yosh reminds us, every bit of high res data, whether scans or fancy arbitrary modellers, has to be manually cleaned up artists to make something game friendly. It'll do for some content but I don't expect it revolutionise asset creation. It would be beyond incredible for MM to develop a process of downsampling that professional tools haven't yet achieved even specialising in that field.
Where Dreams will shine is exporting high-res models to modelling/animation apps, perhaps, and of course 3D printing.
Where Dreams will shine is exporting high-res models to modelling/animation apps, perhaps, and of course 3D printing.
Sure. I just don't expect the exported meshes to be highly optimised or a particularly useful basis for a full game. A character model created in Dreams won't be well positioned to be a skinned, rigged, animated character in an Unity game capable of using Mechanim animations. It'll be an incredible feat if it is!
MM confirmed in their first gamescom twitch stream that models created in Dreams will be accessible via their website, where those objects will be able to be exported in "common" formats that can be loaded in 3D modeling tools, game engines [I think they mentioned Unreal and Unity], or 3D printing apps.
Even if they [even non-textured objects] can be used for environments, props and similar stuff, it will be awesome!
D
Deleted member 86764
Guest
I just hope the "game" part of Dreams ends up as good as the engine looks.
Watching the videos they've produced there are a few anomalies at the moment; when possessing a character and walking it almost looks as if the animation shows that they're sideways walking up a hill, even if the ground appears flat. MM also had a few issues in the recent demo where the pointer would keep randomly going straight to the top of the screen. Maybe the camera lost sight of the DS4?
At the moment, it just looks beautiful - really, really nice. I just hope they're able to knock out a decent physics engine too.
Watching the videos they've produced there are a few anomalies at the moment; when possessing a character and walking it almost looks as if the animation shows that they're sideways walking up a hill, even if the ground appears flat. MM also had a few issues in the recent demo where the pointer would keep randomly going straight to the top of the screen. Maybe the camera lost sight of the DS4?
At the moment, it just looks beautiful - really, really nice. I just hope they're able to knock out a decent physics engine too.
Who is to say that Unity & Unreal won't also support point based game engines in 2016 - 2017.
They seem to be using biped inverse kinematics for the character animation & it's a lot less predefined as normal character animation so with that freedom you'll likely to see more strange animations. And yes they was having trouble with the light tracking I think it was too much light coming from somewhere or whatever they said.
I said support a point based game engine.
That will most likely be the case but I'm saying that maybe they will also have a point based engine in the future.
Extremely unlikely. It'd be a whole other engine, and their hands are full developing and maintaining their cross-platform engines. There's also no toolchain for creating SDF material, so it'd be an engine created requiring a PS4 and Dreams to create content.
However they do it, it'll be an automated tessellation solution with wonky triangles and broken UV wrapping in complex models - that's always what happens.
I guess that depends on whether proper artists find the modelling tools better than they're used to. I think it'll find its fans in game creation, but I doubt it'll be revolutionary and usher in a glorious new era of AAA-quality indie titles thanks to amazing modelling and texturing powers.
Will it be so hard to use range finding cameras to scan in objects using signed distance fields? Also why wouldn't the 3D CADs that are already being used work?
edit:
Last edited:
No. 3D scanning should work very well, creating an SDF that BubbleBath (the name of the engine in Dreams) can render.
Why wouldn't they work on what?
Can you please provide links to references. I've no idea if this image is from a real product or a paper. Either way, it's not dealing with triangles as a games engine would. Any SDF renderer or software could likely receive exports from Dreams and handle them nicely. It's non-SDF software where the conversion is required that'll limit what people can do with their dreams creations.
No. 3D scanning should work very well, creating an SDF that BubbleBath (the name of the engine in Dreams) can render.
Why wouldn't they work on what?
Can you please provide links to references. I've no idea if this image is from a real product or a paper. Either way, it's not dealing with triangles as a games engine would. Any SDF renderer or software could likely receive exports from Dreams and handle them nicely. It's non-SDF software where the conversion is required that'll limit what people can do with their dreams creations.
You said there's no toolchain for creating SDF material & I'm asking what would be so hard about creating a tool chain for a point based rendering game engine? you could scan 3D objects into point clouds & use them in the games also you can edit the objects like what is being done in dreams. Also why wouldn't Epic or the Unity people be able to create a game engine that's able to create materials like in Dreams?
The image come from a paper Coloured signed distance fields for full 3D object reconstruction.
Nothing particularly to some smart devs who can makes sense of the papers as MM did. It just doesn't exist.
Nothing's stopping this happening. But then nothing has stopped other devs creating a SDF renderer like MM have. They've done so, and there's a chicken and egg situation. Why create a SDF modelling app when there are no game engines that can use it? And why create a SDF game engine when there's no content for it? MM solved this by creating both halves themselves. Could Unity or Epic create a game engine that used Dreams exports? Yes. Would they want to? What's in it for them?
Okay. What's the citation in aid of? What argument are you supporting with this? I don't see how a paper on 3D scanning relates to Dreams having 3D modelling and possible export opportunities. You need to join the dots for us.
Nothing particularly some smart devs who can makes sense of the papers as MM did. It just doesn't exist.
Nothing's stopping this happening. But then nothing has stopped other devs creating a SDF renderer like MM have. They've done so, and there's a chicken and egg situation. Why create a SDF modelling app when there are no game engines that can use it? And why create a SDF game engine when there's no content for it? MM solved this by creating both halves themselves. Could Unity or Epic create a game engine that used Dreams exports? Yes. Would they want to? What's in it for them?
Okay. What's the citation in aid of? What argument are you supporting with this? I don't see how a paper on 3D scanning relates to Dreams having 3D modelling and possible export opportunities. You need to join the dots for us.
The 3D scanning wasn't about Dreams it would be a way to create materials for a game that used a point based rendering engine.
Looking at Dreams you can clearly see why smaller devs would want to use a engine that's not limited to polygons.
Then you needed to explain that when you made your point and reference. That aids discussion no end!
As for 3D scanning as a content creation system for a hypothetical SDF game engine created by a third party (Unity or Unreal), if we assume someone implements it beyond a paper and into a real product, it'll only solve some problems. Animation is still going to be difficult. Realistic people is likely a long ways off. That alienates a lot of games immediately. For creating static assets like scenery you could scan them in but you then limit yourself to what you can find/make and scan. It's not particularly useful outside of some specific game styles. Realistic open world adventure games might do exceptionally well where the characters needn't be realistic. VR haunted house games could be incredible in SDF using real world scanned materials.
Why? Polygons have a massive legacy providing tools and resources. Polygons solve every issue a game needs, including accurate human rendering and animation. How is an SDF engine better for smaller devs?
Oh, I don't see it as a serious new content creation tool. It's just a nice feature to have. Say a good sculptor was just messing around in dreams and modeled something that actually looks pretty good. He has this neat option to export it so he can re-model it in maya or what not if he so wishes.
Or, imagine your daugher made this beautyfull piece in dreams, and you can export it and 3D print it somewhere. Now that is very cool.
I don't think it will be super widely used, but definetly will come in handy for a few people here and there.
