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

Any1 know what closest lora to grok nudity quality, found default one very lacking for nipples etc
 
Sorry, been along time since ive used loras (so I dont remember unfortunately), Today I bought a new 5060,16 Gb card and all hell broke loose, everything messed up, nothing works nomore, Comfy, WanGp, nothing, just amazing, the card was to repllace my old 3070 card and was WORSE & even slower than my old card for generations... so guys DONT buy any 5060TI OC card, thoose sux big time!, I prob need to "downgrade" and start all over with new installations all over. (yes Ive spent Hours, hours to trying to get it functional... nope. didnt help.
 
I would suggest creating a separate minimal comfyui portable setup to check if the card is performing as it should. There's not much to download or setup with H3 anyway. 5060 Ti isn't THAT much better computationally than 3070 but it shouldn't be worse and obviously the RAM should help a lot.
 
Since nobody seems to share workflows, Here's one from "skill destiny" on yt which is supposed to work on 8 GB VRAM. idk what card he has but he claims to get 10 s renders in under 10 min. I have 12 GB VRAM and it takes me ~30 min for a 15s clip.
 
if you want to try, comfy cloud gives you 5 gens for free, they run on the rtx 6000 pro gpu, so everything is pretty fast.
lora upload is locked for premium but you can easily bypass that, including the gens counter.
 
I removed 5060 out and put my old in again, now everthing works like a charm again, think it must have somthing to do with the new cards cuda/kernel wich didnt correspond as it should.
I removed 5060 out and put my old in again, now everthing works like a charm again, think it must have somthing to do with the new cards cuda/kernel wich didnt correspond as it should.
 
May seem like a random question but how much will the "nvidia spark" improve ai capabilities give the amount of vram it has
.
 
Im new to this, is comfyui the best to use for ai image generation as well. Im downloading H3 minimax now but want images too since grok is nuked
 
Is FireRed the best for this on comfyui? Also I have a macbook with 32gb ram so not sure if I'll be able to use this, have been getting SamplerCustomAdvanced error
 
is there a minimax for i2i generation? Apologies if this has been discussed already I didn't see it when scrolling through
 
RAM is not really a factor , rather CPU VRAM and GPU. What model of a MacBook do you have? Maybe a smaller model would be okay if you have one of the newer ones. I'm no expert, though
 
Anyone know if M2 Max 32gb unified ram macbook pro can run minimax h3 on comfy, having a lot of trouble but I also know zero code
 
Give it some weeks for good LORAs to release and we finally have a good local uncensored free Grok alternative! What a time to be a gooner.
 
This model is amazing. R2V with zero brain/zero effort perfectly makes voiced 15 second clips that look and sound just like the references. Just click the template in ComfyUI and it works. I never even had such luck with just audio voice cloning before, let alone both video and audio.
Has anyone tried anything for stitching together generated clips to make a coherent longer video?
 
Does this only work for NSFW locally? Tried it on Venice a couple of times and when trying to prompt penetration, it just made the dick deform and grow forward into the woman.
 
You must log in or register to reply here.
Back
Top