MiniMax H3 Face Fix! Remove Blurry Faces in ComfyUI (6GB VRAM WORKFLOW)
retained text
Hello everyone, I’ve developed a new workflow designed to improve blurry and distorted faces in MiniMax H3 videos using Face Refiner groups and custom nodes to isolate faces from low-resolution frames, followed by a second sampling stage with a regular sampler at 0.4 denoise. The workflow also integrates the SelfLift sampler and VEDA Sparse Attention to optimize generation performance for low-VRAM users. In my tests, MiniMax H3 generation at **0.6 megapixel,** CFG **1**, and **8** steps takes around **30 minutes**, or **35 minutes** with an input image, while the Face Refiner stage takes approximately **24 minutes** with either a text prompt or two input images. Actual times depend on your hardware and settings. If you’re struggling with face quality in MiniMax H3, check out my latest tutorial and let me know how it performs on your setup! ***Workflow Links*** [https://drive.google.com/file/d/1J7V8pJ5XWpKx41\_imIgMVN\_PupFAVDbF/view?usp=sharing](https://drive.google.com/file/d/1J7V8pJ5XWpKx41_imIgMVN_PupFAVDbF/view?usp=sharing) [https://civitai.com/articles/37544/fix-blurry-faces-in-minimax-h3-videos-with-comfyui-face-refinement-low-vram-workflow](https://civitai.com/articles/37544/fix-blurry-faces-in-minimax-h3-videos-with-comfyui-face-refinement-low-vram-workflow) ***Video Tutorial Link*** [https://youtu.be/pClKMe\_5Ar8](https://youtu.be/pClKMe_5Ar8) ***Donates*** [https://www.patreon.com/CGPIXELAI/posts/new-comfyui-fix-173591780?utm\_medium=clipboard\_copy&utm\_source=copyLink&utm\_campaign=postshare\_creator&utm\_content=join\_link](https://www.patreon.com/CGPIXELAI/posts/new-comfyui-fix-173591780?utm_medium=clipboard_copy&utm_source=copyLink&utm_campaign=postshare_creator&utm_content=join_link)
document fetched 2026-10-29 00:43 · canonical https://reddit.com/r/comfyui/comments/1xhmt6x/minimax_h3_face_fix_remove_blurry_faces_in