Wildminder

@wildmindai

Physicist, Programmer, Designer

Berlin t.me/TokenDiff Joined December 2024

104 Following 13K Followers 2.6K Posts

Latest posts 5

Banger!
DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic - cool model, using it as my orchestrator from now on.

- Fast and reliable
- Zero rejections
- Accurately routes messages/tool calls between heavier models

huggingface.co/xtools-at/Deep…

Wildminder @wildmindai 3h

ComfyUI Open-H3-Regenerate-2K-IC

nodes + ComfyUI ready weights

github.com/slmonker/Comfy…

0:16

Wildminder @wildmindai 3h

New amazing NVIDIA drop- IC High-Res Video Refinement.
holds the draft clean in the transformer context while generating 2K/4K detail from noise

- faces match
- lip sync holds
- text stays readable at 2K and 4K

beats prior regenerate models on drift.

nvlabs.github.io/Sana/Open-H3-R…

View quoted post on X

Wildminder @wildmindai 3h

New amazing NVIDIA drop- IC High-Res Video Refinement.
holds the draft clean in the transformer context while generating 2K/4K detail from noise

- faces match
- lip sync holds
- text stays readable at 2K and 4K

beats prior regenerate models on drift.

nvlabs.github.io/Sana/Open-H3-R…

0:52

Wildminder @wildmindai 21h

ComfyUI-KEEP, Video Face Super-Resolution

Got asked to update this old node for upscaling faces (vid & img).

Fully reworked and optimized for the latest ComfyUI.

It definitely doesn't match modern methods, but it still gets the job done!

github.com/wildminder/Com…

0:38

Wildminder @wildmindai 2d

MOSS-Image Huge LoRA Archive

A Civitai dump of all available LoRAs. Updated daily.

…and tracking all this stuff for MOSS? No thanks.there are way too many of them, lol.

huggingface.co/mossloras/moss…