Recommendations and good practices to create a full body avatar selector with a couple thousands layers
Hi there!
I'm testing a file that is going to be an avatar selector similar to Duolinguo's but with a couple of thousands layers and full body. I'll have multiple outfits for the avatar, so people can select (for example) a dinosaur shoes (from the dinosaur outfit) and a shark hat (from the shark outfit) plus skin colors and hair color.
I just have a couple of animations and I think this won't be a heavy aspect of this file, just a state machine, 1 idle, 1 idle breaker and 3 animations. But multiple layers to make the logic for the different outfit combinations.
My file has multiple svg for the body parts and outfits (like 3) and everything runs smooth to this point. But if I test duplicating the solo layers for each outfit to mock a 50 or 100 layers file the performance drops...
Possible solutions from my POV:
working the outfit layers with embedded or referenced assets
reduce the look of the avatar to shoulders and head instead of full body
Do you have any other recommendations that I'm missing to make this better,
Thanks!
These both seem like good ways to begin optimizing the customizer, but some other things you can consider:
Where possible, reduce the number of shapes you are using. For example, if you have 5 hats that all use the shape basic shape, but the colors are different, just use 1 hat and key the color.
If you're using bones, only bind and weight the shapes you need to deform. Create hierarchical connections where possible.