What is a good size for an avatar?


#1

I am fine-tuning my avatar and have had some success at hand-simplifying vertexes, but the FBX is about 9MB. Is this excessive?


#2

If its 9mb with textures, thats about ok, even upto 15mb should be fine.

Remember though, the larger the avatar is, the longer it will take for you to appear for others :slight_smile: and the more likely you will be first truncated out


#3

That is untextured! (though I generally don’t use textures much anyway). Sounds like I need go harder on the vertexes! (I am trying to be careful to retain sub-divisibility).


#4

Yikes, thats a massive amount of polygons then and I’d revisit and tune it down.

15k-20k polygons is a good character target, with most of it on the face. You can go upto 30k without much issues, but further than that I suggest avoiding, especially since we do not have LOD targets!

Generate some Normal maps also to save on polygons, because atleast those will scale up/down


#5

My (square) face count is around 13.5k at present, so about 27k triangles (21k body, 5k hair, 0.7k eye). I just almost halved the poly count on the hands without sacrificing and detail I cared about, with an hour of fiddling, so will keep on nibbling away at myself :wink:


#6

I think I have gone as far as I can without ruining it. Clothed-body at 20k triangles, hair at 6k triangles and eye at 1k, for a total of 27k triangles. I’m going to call under-32k a good size since it is a nice round number in octal :stuck_out_tongue: (Actual fbx file - no textures - is just over 8MB! That still seems a big file to me! I don’t know how much cruft FBX carries around with it, though!)


#7

You have to remember that each vertice also contains UV information and weight infromation and each keyshape/blendshape multiplies the value.


#8

Yes. Though I am not a fan of blendshapes*: I’d rather use rigging for facial deforms (I assume there is no technical reason I can’t do so in HiFi other than that face-movement/emotes would not work without swapping in new scripts that use the extra bones).

*- Blendshapes, while a clever trick for pre-generated animation, don’t really seem to easily lend themselves to intercepting my actual facial movements for easy mapping to my avatar, if I ever want to do that.


#9

Basically to do it via bones you have to write everything related to driving the face your self.

Plus its more harder to do for many different avatars without the avatars having bundled animations for the face animations to begin with. Which makes Blendshapes driving easier when dealing with so many unique avatars.

As in it’s doable but it’s a lot of work.