简单易懂的anima工作流/text2img:Simple Workflow for anima(lora、fix、FaceDetailer) — v0.2

Use this Anima text-to-image workflow to turn text into images with a selected model through three generation steps; it supports LoRA, amplification, facial restoration, coloring, and image comparison.

The three-step method is intended to give you more room to adjust generation when LoRA or artist tags constrain the model.

Version v0.2 adds coloring, supports loading three models, adds image comparison, and adjusts some individual parameters.

Provide text, choose a model, apply LoRA, and set the generation steps and amplification coefficient to create images.

The image-generation process includes amplification, facial restoration, coloring, and image comparison.

The default model is for testing only; you can use any model you prefer.

Named node packs include comfyui-impact-subpack, comfyui-easy-use, comfyui-impact-pack, comfyui_essentials, comfyui_controlnet_aux, rgthree-comfy, and comfyui_ultimatesdupscale.

Named files include animaLLMLayerwisefp16_v1.safetensors, sam_vit_l_0b3195.pth, MatureRitual-anima-0.1.1aesthetic.safetensors, and TokiKado-anima-0.1.safetensors.

Other named files are Anzhc Face seg 640 v4 y11n.pt, controlnetLlliteAnima_lineart.safetensors, Anzhc Eyes -seg-hd.pt, qwen_image_vae.safetensors, and 4x-UltraSharpV2_Lite.safetensors.

The base model is Anima.

Adjust the number of generation steps and amplification coefficient to suit your needs.

Suggestion · non vérifié

Before using it, verify that the named node packs and files are available in your setup.

Suggestion · non vérifié

The stated 450–570 seconds on an RTX 3060 applies to the default steps and amplification coefficient; treat it as guidance, not a tested runtime for your setup.

Suggestion · non vérifié