The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Better fix for lowvram issue.
comfyanonymousa3cc3267489b
diff --git a/comfy/model_patcher.py b/comfy/model_patcher.pyindex 5577ca4bf..56e85f990 100644--- a/comfy/model_patcher.py+++ b/comfy/model_patcher.py@@ -393,9 +393,9 @@ class ModelPatcher:if m.comfy_cast_weights:wipe_lowvram_weight(m)- param = list(m.parameters())- if len(param) > 0:+ if hasattr(m, "weight"):mem_counter += comfy.model_management.module_size(m)+ param = list(m.parameters())weight = param[0]if weight.device == device_to:continue