🟣 Try it on RunPod

QwenImageEdit21

The Only Workflow You Will Need

🎨 Wildcard Edits | Multi-Reference | GGUF Support

ComfyUI-QwenVL-Mod — Qwen Image Edit 2.1 with Wildcard-driven auto-prompting

🎯 What Is This?

A Qwen Image Edit 2.1 workflow pack built for reference-driven editing — swap faces, heads, outfits, poses, expressions, backgrounds and more using natural-language edit prompts pulled from a curated wildcard library.

The pipeline:

Load Image(s) → WildcardProcessor → QwenVL Unified (prompt enhancer)
              → TextEncodeQwenImage21 → Qwen Image Edit 2.1 → Save

• 🎲 Wildcard edits: type __qwen21/turnaround__ and a full professional editing prompt appears — no prompt-writing skill needed

• 🧠 QwenVL auto-enhance: the unified QwenVL node rewrites/polishes the edit instruction before it hits the text encoder

• 🖼️ Multi-reference: <image1>, <image2>, <image3> inside wildcards bind to your connected images, in order

• 💾 GGUF support: run the Qwen Image diffusion model in GGUF (Q8_0 → Q3_K_M) on low-VRAM GPUs or Apple Silicon

📦 Required Models

Drop into ComfyUI/models/ — pick one diffusion model, one text encoder, the VAE:

diffusion_models

• qwen_image_2.1_bf16.safetensors — full quality

• qwen_image_2.1_int8_convrot.safetensors — lighter

text_encoders

• qwen3vl_8b_bf16.safetensors — full

• qwen3vl_8b_int8_convrot.safetensors — lighter

• qwen3vl_8b_w4a8.safetensors — smallest, low VRAM

vae

• qwen_image_2.1_vae_bf16.safetensors

🐋 GGUF option (low VRAM / Mac)

From Abiray/Qwen-Image-2.1-GGUF — drop in models/diffusion_models/ and load with Unet Loader (GGUF):

• Q8_0 · Q6_K · Q5_K_M · Q4_K_M · Q4_K_S · Q3_K_M

Repo: Comfy-Org/Qwen-Image-2.1 · ModelScope mirror

🎲 Wildcard Edit Library

Type __qwen21/<name>__ for a specific edit, or __qwen21/*__ for a random one. <image1/2/3> = your connected reference images, in order.

• __qwen21/turnaround__ (1 img) — character turnaround sheet: front / side / back + headshot

• __qwen21/prosheet__ (2 img) — pro reference sheet, image2 as extra detail ref

• __qwen21/newscene__ (1 img) — subject into a new environment (cyberpunk night market)

• __qwen21/background__ (2 img) — replace background of image1 with image2

• __qwen21/faceswap__ (2 img) — face of image2 ← identity from image1

• __qwen21/headswap__ (2 img) — head + hair of image2 ← head from image1

• __qwen21/outfit__ (2 img) — dress image1 subject with the outfit from image2

• __qwen21/pose__ (2 img) — copy pose from image2 onto image1

• __qwen21/expression__ (2 img) — copy expression from image2 onto image1

• __qwen21/multiref__ (3 img) — compose scene: image1 + image2 inside image3

• __qwen21/multiedit__ (1 img) — three simultaneous edits (outfit + light + accessory)

• __qwen21/timeweather__ (1 img) — day → midnight rainstorm, geometry preserved

• __qwen21/removeobject__ (1 img) — remove object + photorealistic inpaint

• __qwen21/insertobject__ (1 img) — insert object matching light / DoF / grain

• __qwen21/textedit__ (1 img) — edit text in image, typography preserved

• __qwen21/novelview__ (1 img) — new camera angle, same subject

• __qwen21/anime2real__ (1 img) — anime → photorealistic human

• __qwen21/real2anime__ (1 img) — photo → manhwa / webtoon illustration

How It Works

1. The WildcardProcessor node (TagForge) sits before the QwenVL enhancer

2. At queue time each __wildcard__ token is replaced by the matching entry from wildcards/pmp/qwen21.yaml

3. The expanded text goes through QwenVL Unified for a final polish, then into TextEncodeQwenImage21

4. Different seed = different wildcard picks — pin the seed for reproducible results

Tips

• Image order matters: for faceswap/headswap/outfit, image1 is the source (identity/outfit), image2 is the target photo

• Combine with text: __qwen21/background__, dramatic mood works fine — wildcards expand inline

• Add your own: drop a .txt in ComfyUI/models/wildcards/ (one prompt per line) → becomes __filename__; YAML files support nested keys like __name/subkey__

Required Custom Nodes

• ComfyUI-TagForge — WildcardProcessor + wildcard library: huchukato/ComfyUI-TagForge

• ComfyUI-QwenVL-Mod — QwenVL Unified enhancer node: huchukato/ComfyUI-QwenVL-Mod

• ComfyUI-GGUF (optional, for GGUF quants): city96/ComfyUI-GGUF

🌟 Why This Pack?

• Zero prompt-writing: 18 production-grade edit prompts, one token each

• Identity-safe edits: prompts are written to preserve face, proportions and style across swaps

• Flexible: BF16, INT8 ConvRot, W4A8 text encoder or full GGUF — scales from 24GB cards down to unified-memory Macs

• Extensible: your own .txt/.yaml wildcards plug straight in

📋 Credits

• Qwen Image 2.1 — Qwen Team / Alibaba

• GGUF quants — Abiray/Qwen-Image-2.1-GGUF

• Base template — Comfy-Org official editing workflow

• TagForge + QwenVL-Mod — huchukato

Built with ❤️ for the ComfyUI community