The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Wan seems to work with fp16.
comfyanonymouscb06e9669b1e
diff --git a/comfy/supported_models.py b/comfy/supported_models.pyindex 64611f58d..d39a03f25 100644--- a/comfy/supported_models.py+++ b/comfy/supported_models.py@@ -911,7 +911,7 @@ class WAN21_T2V(supported_models_base.BASE):memory_usage_factor = 1.0- supported_inference_dtypes = [torch.bfloat16, torch.float32]+ supported_inference_dtypes = [torch.bfloat16, torch.float16, torch.float32]vae_key_prefix = ["vae."]text_encoder_key_prefix = ["text_encoders."]