Oriel — AI & technology worth noticing, every day.

PrunaAI/P-H3-Image-Comfy

AI model · https://huggingface.co/PrunaAI/P-H3-Image-Comfy

First seen by Oriel on .

AI summary

This Hugging Face repository provides native ComfyUI files for P-H3-Image, enabling 8-megapixel text-to-image generation and multi-image editing in eight steps from 21 GB of VRAM.

  • P-H3-Image is based on MiniMax-H3 and supports text-to-image plus editing with one image and up to seven references.
  • bf16, int8, and w4a8 files can be mixed; the smallest set downloads 20 GB and peaks at 21 GB VRAM.
  • Native ComfyUI workflows use an optional Qwen3.5-9B prompt upsampler, no CFG or negative prompt, and only built-in nodes.
  • int8 stays close to bf16, while w4a8 changes more details and composition; on an L40S, it renders 2048 × 2048 in 6.3 seconds versus 19 seconds for bf16.
  • Prompt sensitivity, unreliable dense text, and the MiniMax H3 Community License, which excludes the EU, UK, Republic of Korea, and United States without separate licensing, limit use.

#comfyui #image generation #model quantization #image editing

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Likes

Total likes reported by the source.

  • Hugging Face Models: 435, latest observation ; +320 over observed period since

Downloads (30d)

Rolling 30-day downloads. A change in this value is not new downloads during the observed period.

  • Hugging Face Models: 1,200,813, latest observation ; +1,155,190 over observed period since

Aliases

SourceExternal IDURL
Hugging Face ModelsPrunaAI/P-H3-Image-Comfyhttps://huggingface.co/PrunaAI/P-H3-Image-Comfy Read discussion