converted int8convrot of https://civitai.red/models/2895781/anima-29b-p1-turbo11-unofficial-merge
do check the model link for more info, there's barely any quality difference with int8convrot and comfyui seems more optimized for it
i'm using a 3050 rtx with 8gb vram, when using convrot please use the --fast cublas_ops argument at start on comfyui for better performance, its a lot faster compared to normal usage, and way less resources, i'm uploading this because not everyone knows or can convert models to int8convrot
currently using the vae 3.5 from here https://huggingface.co/oron1208/OOO_Anima/tree/main/VAE