AI R&D, animation, gals and news. I make custom workflows and deploy comfy / webui. Feel free to DM me! Business inquires go here: [email protected]
Cheese Bunker
Joined April 2023
- Tweets13.8K
- Following630
- Followers7.3K
- Likes20.5K
Pinned Tweet
Testing out the MOSS by @MosiAI_Official
Mainly checking out style flexibility, which is what I care about.
Styles here are:
3D Render
Pixar Animation
Fortnite Style
Modern Super Mario (Nintendo) Style
Will keep posting.
Thoughts and prompt in feedback.
I'm always intrigued by the thought of MiniMax as an image model. Dabble a couple of times, but never too deeply.
Well, now P-H3, a image gen AND editing model based on MiniMax is here. Gonna try it.
Today, we open-source P-H3-Image, an image generation and editing model built from MiniMax-H3 by @MiniMax_AI. A SOTA video model, turned into an efficient open image model.
- 𝗛𝗼𝘄 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝘁 𝗶𝘀 𝗶𝘁? Generate or edit images 5× faster, in 8 steps instead of 50. A 4 MP image takes 2.6 s instead of 12.4 s. The model is 7× smaller than H3, so high-resolution images are a light download and run on a much smaller GPU.
- 𝗛𝗼𝘄 𝗴𝗼𝗼𝗱 𝗶𝘀 𝗶𝘁? Strongest at photorealism, documentary and editorial photography, film stills, and graphic styles with short quoted text. It renders 2 to 16 MP and edits with up to 8 input images.
-𝗛𝗼𝘄 𝗱𝗼𝗲𝘀 𝗶𝘁 𝘄𝗼𝗿𝗸? We reworked every component of H3: the prompt upsampler, the text encoder, the diffusion schedule, the transformer, and the decoder. Use it with Diffusers or ComfyUI.
📄 Tech Report: pruna.ai/blog/oss-p-h3-image…
🤗 @huggingface Open Weights: huggingface.co/PrunaAI/P-H3-…
👉 @ComfyUI Native Support: huggingface.co/PrunaAI/P-H3-…
The Clown Context Suite has changed its name to the Geiru Context Suite. Its a custom node suite specializing in long form MiniMax H3 vid gen and it helps reduce / eliminate vid gen degradation over long gens.
github.com/roycho87/geiru_co…
Sulfur 3, a totally sfw model for pure-hearted people who only want to generate puppies and rainbows has dropped.
Will I try it? Very yes.
huggingface.co/FusionCow/Sul…
lodestones, the creator of the Chroma model, has release chroma-h3, a MiniMax H3 fine-tune ~
huggingface.co/lodestones/ch…
Oh, ha, that was fast!
Idol, an image to gaussian splatts model has official Comfy support.
▼概要
単一画像から3Dを生成するモデル「Idol」のComfyUI向け再パッケージ版が公開。TripoSplatをファインチューンしたモデルであり、既存のTripoSplatワークフローに対応する。
▼ポイント
・画像から3Dへの変換（image-to-3d）を行う「speridlabs/idol」のファイルをComfyUI向けに再構成
・TripoSplatのファインチューン版であり、既存のTripoSplatワークフローをそのまま流用可能
・拡散モデル（idol_fp16）、CLIP Vision、VAEなど各モデルファイルの配置先フォルダを明記
▼使い所
既存のTripoSplatワークフロー資産を流用し、ComfyUI環境でIdolによる3D生成を行いたい場合向け。
#TripoSplat #ComfyUI
URLはリプ⬇️
Heh, another fella joining the open source higgsfield competition ~
This is a vid recam workflow. Got Gaussian splatts as an intermediate, something I'm tinkering with in the background as well.
Reshoot any video in 4D. itsmetamike.github.io/reshoo…
Convert any video into a 4D Gaussian Splat. Create a new camera path in Comfy. Get the same moment back from the new camera.
Four ComfyUI nodes, local, open weights, built and run end-to-end on a 3090.
#OpenSourceHiggsfield 🧵
Oh HYES!
Someone implemented time-to-move for MiniMax H3 called Motion Time-to-Move.
Allows for precise motion control within MiniMax H3!
Minimax h3 😲 Motion Time to move
implementation of time-to-move dual clock denoising for Minimax H3, allowing for motion control by dragging objects into motion, or mask part of a video
and change what happens.
github:
github.com/PxTicks/h3-motion
workflow:
drive.google.com/file/d/1mOH…
Nice ~
A motion transfer / char swap studio, free and open source!
Introducing OpenCast Studio!
A free and open source character generation tool with both character swap and motion transfer built in!
A custom Krea 2 lora lets you generate funny characters, then Flux 3 generates an anchor, finally the new Pika H3 Flex model performs the swap or motion transfer.
It's composed of ComfyUI workflows under the hood and only uses open source models.
This is my submission for the #OpenSourceHiggsfield competition!
Download free on github now: github.com/akatz-labs/openca…
I have vibe-coded a ComfyUI custom node For the Idol image to splat model.
Did this because I needed it for a side-project.
Results are good enough for certain uses.
github.com/Brie-Wensleydale/…
3D Gaussian splats from a single image AND it runs on consumer GPUs. Something I've been looking for for quite some time (and, it also makes what I'm working on obsolete. lol)
No Comfy Custom Node yet, but I bet it'll come around soon, no?
huggingface.co/speridlabs/id…
An MMH3 fine-tune from Pika.
Optimized for greater control and supports 12 refs.
Open and out, but awaiting reasonable quants.
I will try it once its out, since I've been trying various MMH3 models recently.
huggingface.co/Pika-Labs/Pik…
