Want to try this one! see how it runs on my machines

Could you let me know how long a single generation takes on your computer? And which graphics card is used

Hollywood is so dead. This is OUR world now.

I just watched the Genmo video... and wept.

A dream of 40yrs is coming true. What a time to be alive.

unfortunately i am getting an out of memory error at the VAE encoder. with a 12GB Vram Nvidia3060

Can I run this with Forge or Auto?

thank you very much, your description helped me a lot to understand how mochi works, currently trying a couple of stuffs.

this is what i've been waiting for! awesome !

Is this only the text2video model? is there a way to make img2video?

I guess it's time to also upgrade my RAM 🙃32GB ram froze my PC for 10mins then GPU kicked in smoothly with 24GB Vram.
25mins total render.

Very cool so far. I am running it locally and wondering if anyone has some ideal settings they can share so far? Looking for how to increase duration and quality. Right now, I see that length is set to 43 in the default workflow, which equals a loop of about 3 seconds. Is there some kind of equation of how much length = actual seconds?

Amazing that we're seeing competitive open-weight txt2video models! Now just need to find where I put those H100s to start training LoRAs... 😭

@theally OP, Is there a node we can use that's similar to stable video diffusions "SVD img2vid_Conditioning" node that will allow us to do text+img2vid?

I made short animation via 4Gb VRAM 🤐

"If you don't want to run it locally,"

Only an idiot wouldn't 'want' to run it locally. If people don't it's either cuz they haven't been taught how, or cuz, thx to NVIDIA's proprietary gatekeeping BS, they can't afford a card good enough to do it locally.

Just wanted to clarify, that people really should try to do it locally if they have the hardware, cuz eventually, anything that can be locked behind insurmountable paywalls, will be.

So, get into it locally while you still can.