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

Is everyone here paying for the API use or is there a free version I'm not seeing on ComfyUI?
 
I'm on Windows. When I search templates for MiniMax I get 4 results. 3 of them are the official MiniMax ones. When I download them and try to run them I get a notification that I have to sign in to use their API.
Is everyone who is using it paying for the API or is there something I'm missing?
 
Are you on the newest version of Comfy? The templates only showed up when I updated to the most recent version, before I also only had the API ones.
 
I also suggest updating the drivers for your card. The workflow was crashing on me about halfway through until I updated my drivers.
 
Im new to comfyui and im using the default template of the i2v minimax. 5 sec video generates around 10-12 mins. Are there anyways i could make it faster by changing some settings or adding something in the default workflow?
 
I can confirm this works. It also seems to take about half of the time. I generated a video with the same prompts and settings in both systems.
 
You can also just download the workflows from the website. Apparently typing the website address is not allowed here for some reason.
 
How long does it takes for you all? And in what GPU? I'm running the bf16 models on RunPod with a RTX PRO 6000 (98 GB VRAM, 188 GN RAM, 16 vCPU) for $2 and hour when the Pod is running, and it takes around 20 minutes for an 8 sec video at 1 megapixel for 9:16 aspect ratio using the Reference model with an image and a video input with audio.
 
My results for i2V were terrible quality. Not sure if I'm missing something or if it's just not good for i2V. Boobs looked okay I guess but everything else looked bad. No where close to the quality of Grok uncensored that was available briefly on Venice.
 
default i2v template, 0.4 MP 9:16 takes around 220s for 5s video on a 5070 12GB local
 
I figured it takes so long because of the 1080p 15sec video I'm using as video reference, the queality is really nice regarding character fidelity tho
 
isn't the smallest model over 20 GB? How can you make it work with 12 GB?
 
14 minutes per 10 second 352x608 video. I'm using a local RX 6900XT 16GB, 32GB RAM, with a 5800X 8-core. I also set up 100GB of Virtual Memory on an NVME SSD.
 
Is there a guide for noobs?
I tried I2V with prompt "animate" just to test it , 1 sec, standard settings and it took me 500+ sec
I'm sure it shouldnt be that long for 1 sec
i got 4060ti 16gb
 
it is probably the best out of the box model ever. great for T2V amazing for I2V, does everything you ask it to withotu hassle
 
You must log in or register to reply here.
Back
Top