Vertex Animation Toolset, crowd rendering

Can you explain animation frame compression a little bit.

What exactly will happen if we have too many frames of animation? Will the tool tell us when we have too many?

Is there some upper limit on how many frames can fit in the texture? Should we adjust the texture size to allow more frames?

Can i cupture skeletal mesh vertex positions and generate or update existing texture at runtime?

To everyone that has any questions about the plugin, It’s way better to message me at the support e-mail: hethger@gmail.com or ask in the plugin’s marketplace page.

I’m sorry but I just checked this thread in some time and now see the questions going back even some months, through e-mail and the marketplace page I get notified and typically answer within a day or two.

Cheers.

Can I get more info on how this works?

I was only “the user” of the system but I didn’t develop it so I don’t know any of the implementation details.
some details that might be useful:

  • the skeleton’s bones were mapped to a “vertex-bone table texture” (HDR, texture size X = number of mesh vertices, Y = 2px but I have no idea how the mapping was done)
  • the animations were 3 textures per anim (HDR, texture X = number of bones, texture size Y = number of animation frames, each texture corresponding to X, Y and Z matrices respectively)
  • the material would evaluate the passed anim texture against the vertex-bone table texture with a custom function (no idea how it did what it did) and plugged into WPO

That’s something, I’ll take it! Thank you very much for the insight.

What would be cool is if I could figure out how to get vertex animated static meshes to update their collision with the vertex animation.

Also, they were probably doing something related to this: SkinnedInstancingWhitePaper (nvidia.com)

Hey, is there anyway you could generate bone / weight mapping textures to use against generic animation textures so that if I use several meshes with similar skeletons I can reuse textures? I’ve been told this use to be a common practice on older generation stuff and it’s really efficient for something like oculus quest.

I ran into several people in unreal slackers who also told me to use this kind of set up so I’m looking for a way to get it done. Hopefully this plugin will get an update or I can pick it apart to make it happen on my end.

the setup we had required 100% same skeleton hierarchy and proportions even if the meshes were different. but really it’s up to you in how well you can develop such a system

pretty much figured it out at this point, only thing I need to do now is find out how to check which vertexes are weighted to which bones. I can think of a messy way, but I want to cut out as much uneeded data as possible. What I’m thinking about is smooshing some custom uv stuff, but really I’m not certain of a few things currently. Once I figure it out, I’ll be able to set up something that works on “same skeleton” set ups.

pretty sure you’re gonna need a 2nd UV channel to map the vertices to the bones. that’s how the UE4 (the per-vertex, without bones) vertex animation system does it, and that’s how we had it in our custom bone-texture approach as well

1 Like

Yes that’s expected. I’m looking into something further now where you can have a bunch of variations in one mesh and use something (probably some sort of masking function in HLSL) that gives you something where you can hide all your undesired variations by using WPO to scale the vertices to zero or move them all to one point.

You can do this with a regular mask ( as I have found in my experiments ) but I want to be able to have more dynamic variation instead of premade variations.

There seems to be a lot of issues with doing this with the standard nodes. I haven’t seen a good way of doing something like this yet and I’m not super familiar with HLSL so this will be a journey for me.

Can you please teach me how to resize static mesh created by this tool.

Hey guys,

I cam across this topic while I was facing a huge performance gap during my work on a zombie base horde/crowd game. More information about the game will be soon here [Plugins] WIP Horde Framework - Work in Progress - Unreal Engine Forums

Does somebody already have experience if this method.

I do have the following questions:

  1. Cann this technique be used with replication?
  2. Are there physical reactions (e.g. the pawn bumps into a wall) of the pawns?
  3. Should this technique only be used for far away pawns or can we use it also for close pawns?

Amazing! Can it blend animation?

Hi, I am really having problems using this with a different character. I followed the steps in the video 3 times and i just get a jumbled mess at the end…

The material functions no longer take anim speed, anim start, etc pins any more and now have an Anim Settings pin. I assume that is an RGB or RGBA of something else but I wasn’t sure what. I believe the the first three are the three parameters for each animation but since takes an alpha I wasn’t sure what the alpha would be. It seems to work with an alpha of 0 though.

Upgraded version to 4.26 has a great potential with Bone position/rotation extraction. But quite bad precision yet.

For example, this is the animation taken from Example project. Check the neck bones jiggling:

Do you have any ideas how we can avoid this?

Thank you in advance!

Vlad.

Hello there,

What about UE5 Early Access support?

Documentation no longer accessible in dropbox