(For SOTAI Squad) Master Qwen Image Edit: Consistent Characters, Virtual Try on & Low VRAM Speed Hack

Video link: https://youtu.be/NaanLUAO3FA

In this video, I’ll show you how to fully master Qwen Image Edit with:

✅ Consistent Characters – keep faces and features stable across multiple generations.

✅ Virtual Try-on – easily change outfits while preserving character identity.

✅ Low VRAM Speed Hack – optimize workflows to run smoothly even on low GPU memory. You’ll learn how to:

- Build an efficient Qwen Image Edit workflow.

- Maintain high-quality outputs with limited VRAM.

- Speed up inference without sacrificing too much quality.

• Models:

  • Qwen_Image_Edit_VAE - put to VAE folder: https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/blob/main/split_files/vae/qwen_image_vae.safetensors

  • Qwen_2.5_vl_7b_fp8 - put to Text encoder folderhttps://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/blob/main/split_files/text_encoders/qwen_2.5_vl_7b_fp8_scaled.safetensors

  • Qwen_Image_Edit GGUF - put to Unet folder: https://huggingface.co/QuantStack/Qwen-Image-Edit-GGUF

  • LoRA model - put to LoRA folder : https://huggingface.co/lightx2v/Qwen-Image-Lightning/blob/main/Qwen-Image-Lightning-4steps-V1.0.safetensors

• Note:

  • Install custom node "ComfyUI_LayerStyle" to get "Purge VRAM" node

  • Install custom node "ComfyUI_PuLID_Flux_ll" to get PuLID nodes