image to video using Hunyuan checkpoint.
32GB ram and 16GB vram/ laptop 3080TI.
Description
FAQ
Comments (22)
Can you share the video generated from the image?
The video that is added is that. I think you maybe looking for something similar to this workflow.
seems regular vae loader having trouble recognize hunyuan vae. How do you get around that: WARNING: No VAE weights detected, VAE not initalized.
I have not had that issue. I would try to delete the node and add it back. Don't use undo to add it back.
Also having vae issues accept a different error. 'VAE' object has no attribute 'vae_dtype'
This worked great, thank you!
It took 21 minutes to generate the example video on my 3060ti (8GB VRAM, 32GB RAM).
Oh my God, 21 minutes. Thanks for letting me know. I was going to try it on my 3070, but now I gave up and didn't even download it.
When loading a video model, comfyui just quits. Does anyone know the cause?
Not enough RAM/VRAM/SWAP or pagefile too small
Depthflow nodes can not be found, even if I reinstall it or try to fix it, any solutions?
Take a look at these resources. I didn't have that problem so I am not sure.
Try changing the version, seems to fix it for me.
Change version when attempting to install missing from "Latest" to "1.0.0" worked for me.
hello can someone help me troubleshoot? my input image seems to be disregarded entirely.
Edit: Bring Denoise down, note says 1 is totally new image, .01 is very similar to input
Very good workflow! works like a charm and its quick and easy to use. Just having 1 question:
Is it possible to add movement to a picture and keeping the original but make it animated?
Ps: if i put it on 0.1 denoise the picture keeps the same but there is 0 movement. If i put it higher there is or still no movement, or then the picture changes. But i just want to add movement to the existing picture
you can try to find a good balance for the noise value to get you what you need. Also check out https://civitai.com/models/1131397/v4-dejanked-image-2-video-hunyuan-magic-with-flux-and-refiner-speed-hack this adds movement to the image before it is sent over to the encoder.
I've run this a bunch of times now and gotten an actual moving picture like twice, every other time it has output a still image. Anyone know whats going on? Or have a way to diagnose
same for me. denoise lower then 0.9 and its a still frame.
Having a hard time getting it to not try to allocate more than 20GB VRAM