简单易懂的Illustrious工作流/text2img:Simple Workflow for Illustrious(lora、fix、FaceDetailer) — v5.0

This Illustrious text-to-image workflow turns text into images through three steps, with LoRA, upscaling, and face repair.

Its stated aim is to give you more room to shape generation across steps when LoRA or artist tags would otherwise constrain a model’s output.

Version v5.0 further optimizes parameters and adds a noise node.

The graph includes LoadImage, PreviewImage, and SaveImage nodes for loading, previewing, and saving images.

The default model is described as being for testing only.

Named model files include 4x-UltraSharpV2_Lite.safetensors, Anzhc Eyes -seg-hd.pt, Anzhc Face seg 640 v3 y11n.pt, IllustriousXL_openpose.safetensors, lianhuanhua_fengshenyanyi_epoch_10.safetensors, MatureRitual-v2.0.safetensors, mistoLine_rank256.safetensors, and sam_vit_l_0b3195

Named node packs are comfyui_controlnet_aux, comfyui_essentials, comfyui-easy-use, comfyui-impact-pack, comfyui-impact-subpack, and efficiency-nodes-comfyui.

The description says you can use any model you like.

Before running it, check that the named model files and node packs are available in your ComfyUI setup.

Vorschlag

Check the face-detector filename before setup: the description names Anzhc Face seg 640 v4 y11n.pt, while the workflow lists Anzhc Face seg 640 v3 y11n.pt.

Vorschlag