Hi!
After switching from @pixi-spine to @esotericsoftware/spine-pixi-v8, I've got a significant increase in draw calls.
My problem is rendering about 20 different Spines one by one. Each has a few slots of different blendings (background-middle-foreground), where the middle one is normal and the other two have additive blending. This results in the blending being switched about two times when rendering each one.
It also makes every one of them generate about 2-3 draw calls. Rendering all of them gives me about 60 draw calls. They are not expensive by themselves, but I see significantly worse performance compared to the previous approach.
From my investigation, it looks like it's due to spine-pixi-v8 being much more straightforward in terms of rendering. I believe that @pixi-spine's approach of mapping slots to the built-in elements of Spine was for some reason more batcher-friendly.
Is there any known technique to make batching of those Spines slightly more performant?
I'm quite successful at experimenting with grouping the same blended layers together for multiple Spines, as they almost never overlap, but maybe I'm missing something much simpler?
