text2vid img2vid H3 Minimax - Fully uncensored local AI video generator released today

ferris09

ferris09

Lurker
Joined
Aug 4, 2026
Posts
5
Reaction score
0
All you have to do is open ComfyUI Desktop and select templates to select H3 Minimax. It is NSFW capable with the default workflow but there is room for improvement with Loras. Check out www.reddit.com/r/stablediffusion to see the insane results people are getting. There are templates for image-to-video and text-to-video.
 
I tried directly from their website reusing / rephrasing a nano banana prompt and got a video with some tits (requires few attemts)
POV style video, first person view, view from my eyes. she stands before me. It is a pride month, so she decided to apply a full bodypaint on her. On the image we can see my own hands with brush, I'm helping her to paint a pride month flag on her. she stands with her hands behind her back, letting me finish work. Because of that action she stands frontal, without top clothes, allowing me to paint. Living room interrior. Day time.
 
If you run it locally it's fully uncensored with zero guardrails.
 
If you have runpod, you can use the template called "One Click - ComfyUI MiniMax H3" and it works like a charm, a 720p 8 sec video can take up to 10 mins depending on the GPU you use for your Pod but it's so worth it.
 
It's awesome! I have a 5090 and a 10 second video took me around 220 seconds. I did the prompt via Grok and it went full nudity. I'm trying out more hardcore porn soon.
 
where can i find a guide on installing this locally? ive never done this
 
use comfyui, it already includes some pretty good templates for t2v, i2v, and r2v
 
Yes, just use the reference worflow and put the video as input.
 
I only see it in Comfyui for Windows and not when I search in the Linux version. You guys have any idea how long it's going to take for it to show up over there? And yes, I did update the software.
EDIT: This was probably a stupid question. I am pretty sure i can just get the workflow on windows and move it over.
 
i have problems trying to connect a video and get it to work. Can you share your workspace and the connections?
 
Just add a "Get Video Components" node between the video and the Minimax Reference node, you'll get the video images and the audio ref with that
 
It's very simple. I don't think you need a guide. You install Comfyui Desktop and open it, click templates and select one of the H3 Minimax templates. Then you'll be in the workflow. You'll have to click the error button on the right side of the workflow to download all the stuff for the model. Once it finishes downloading, you just type a prompt for your video and click Run.
 
just google comfyui easy install and then download the model, text encoder, vae and a workflow.
 
Can it be replicated in comfy.cloud with the plan that allows you to upload external models from huggingface?
 
does grok chat allows for adult themed prompt generation or you had to trick him?
 
I don't have a X account yet I wanted to make sure it was legit and not just another workaround solution before making one.
 
You must log in or register to reply here.
Back
Top