I have tried to compare performance when loading glb models which have embedded textures and models which have separate texutres (in separate folders). and as a result, models with embedded textures are better than models with separate textures.
Can you explain a bit more about how are you measuring and comparing the performance? I can imagine that the embedded textures model would be faster to load, as it would only need one load for everything, but once both models are ready, I’d expect them to have similar performance. Unless the texture size is different in the models, then, the one with smaller textures would have better performance.
After further research, I found that models that have separate textures have glass objects, and models that have embedded textures don’t have glass objects. I speculate that the performance drops (based on the FPS) because there is a glass object, so is it true that glass can make performance decrease?