lmfao engine upgrade looking good. What model(s) did you use? What was the workflow like?
I also spent a few months messing around with comfyUI last year and of course I had to test it with APB despite not having touched the game since 2013...
I believe this was on a on heavily optimized qwen model (for fast iteration) and a lora I found that helps with the stylized > realism conversion process.