Very good tutorial and thanks for the free workflow. After few tries I found it very difficult to control the movements of the camera and the character. lets wait with a vid2vid version
💯Thank you. Your videos are always correct, instructive and informative. Thank you for the links you added. I wish you health, luck and creative success this year. Thanks again. 💯
Thanks for posting this nice tutorial. Could it be possible to upload 2 images (or more) to make an interpolation btw them to produce a smooth morphing ?
This might sound advanced, but any chance for a keyframe to keyframe (inbetweening) workflow? As in two separate images but in the same world/stage/place setup that generates frames between the first and second frame in LTX? Could be useful to make extended footage.
I was actually looking for that specific thing as well. It would really be useful. I have a crude method for that, i copy the last frame from video as image and process it further, but if there was node for doing that then it would be awesome.
When loading the graph, the following node types were not found: LTXVConditioning LTXVScheduler ModifyLTXModel LTXPerturbedAttention LTXAttnOverride LTXVImgToVideo EmptyLTXVLatentVideo AddLatentGuide
Does not work with version 0.9.1, it issues a VAE-related error. I downloaded the VAE database for LTXV separately, added its download as a separate module, but I still get an error related to VAE. As a result, I downloaded the old 0.9 database and everything worked with it.
@@xclbrxtra, most likely, the problem is in my ComfyUI. It gives an error when trying to update it.(( ps: I downloaded the latest version of Comfy. Installed all the necessary modules, everything has been working since 0.9.1)
Can we just use Rife for frame interpolation to render a longer loop? I think i have that setup on my current WF... Ill try to incorporate it into this one.
brother, can you make a complete installaion video of LTX, i tried some nodes are missing in comfyui (also tried missing node install), Thanks @Xclbr Xtra
640x384 would work, and then just rescale it with another program to double size, like VideoProc, which is free for videos 5 minutes or less, which these would all be. Then it will be 1280x768, which is close to 16x9 proportions and can be cropped to that size without losing much.
Yes but I've noticed that smaller resolutions have more distortions (like messed up eye tracking). It's kinda like FLUX and SDXL, initial lower resolution may have some deformities.
Thanks for such a simple and easy to understand video.
Very good tutorial and thanks for the free workflow.
After few tries I found it very difficult to control the movements of the camera and the character.
lets wait with a vid2vid version
💯Thank you. Your videos are always correct, instructive and informative. Thank you for the links you added. I wish you health, luck and creative success this year. Thanks again. 💯
Amazing Video my friend, opinions are always there for a reason, Straight forward and simply explained. Thank you,
You are doing just fine!!! keep it up, SUPER
Thanks for posting this nice tutorial. Could it be possible to upload 2 images (or more) to make an interpolation btw them to produce a smooth morphing ?
This might sound advanced, but any chance for a keyframe to keyframe (inbetweening) workflow? As in two separate images but in the same world/stage/place setup that generates frames between the first and second frame in LTX? Could be useful to make extended footage.
Same question. It could be extremely powerful to get a smooth controled consistent morphing with ltx.
I was actually looking for that specific thing as well. It would really be useful. I have a crude method for that, i copy the last frame from video as image and process it further, but if there was node for doing that then it would be awesome.
Well done, welcome to the ltx club
I get an error GGUF Clip loader
Thanks buddy. Wow!
Thank you my friend.
where to dowload the ltx nodes? because in the manager>download missing nodes dont appear. Thanks
ComfyUI-LTXTricks you should be able to search this in comfy manager otherwise you can directly git clone the repo in custom nodes folder
@@xclbrxtra Thank you for your reply but I don't know what is the address for the repo.
Sorry I am new, ;)
When loading the graph, the following node types were not found:
LTXVConditioning
LTXVScheduler
ModifyLTXModel
LTXPerturbedAttention
LTXAttnOverride
LTXVImgToVideo
EmptyLTXVLatentVideo
AddLatentGuide
for example the LTXVConditioning is nowhere online
Does not work with version 0.9.1, it issues a VAE-related error. I downloaded the VAE database for LTXV separately, added its download as a separate module, but I still get an error related to VAE. As a result, I downloaded the old 0.9 database and everything worked with it.
I'm using the 0.9.1 in the video workflow, I think there might be some other issue with vae, I'll look into it 💯
@@xclbrxtra, most likely, the problem is in my ComfyUI. It gives an error when trying to update it.((
ps: I downloaded the latest version of Comfy. Installed all the necessary modules, everything has been working since 0.9.1)
LTX new node feta enhance where does it go
Ты супер, спасибо за воркфлоу
Is it possible to have the start image and the end image the same so you can make a looping video?
Yes
Can we just use Rife for frame interpolation to render a longer loop? I think i have that setup on my current WF... Ill try to incorporate it into this one.
Bro, easy use import failed. If you know how to fix it, I would be really thankful.
Excellent!
Thank you!
brother, can you make a complete installaion video of LTX, i tried some nodes are missing in comfyui (also tried missing node install), Thanks @Xclbr Xtra
florance2run is missing how to install that
BRO easy use import failed if you know how to fix i will be really thankful
How to contact you directly ? I have a few questions :)
Best greeting from Europe 👌
640x384 would work, and then just rescale it with another program to double size, like VideoProc, which is free for videos 5 minutes or less, which these would all be. Then it will be 1280x768, which is close to 16x9 proportions and can be cropped to that size without losing much.
Yes but I've noticed that smaller resolutions have more distortions (like messed up eye tracking). It's kinda like FLUX and SDXL, initial lower resolution may have some deformities.
@@xclbrxtra I didn't know that, I thought it would just make it faster, thanks for the tip.
@@xclbrxtra Your text to video one worked well with a size of 1280x736, took 6 minutes. I thought it would OOM or something.
@@xclbrxtra I used the Q8 model for that btw, and a RTX 3060 12 GB.
Video to video is best for motion.
2060s 12g takes an hour. What's going on😂
worthless only garbage out of this... everything over 1 seconds long will be crap
Ugh I'll try it. Not been having a lot success with how I want the movement from LTX. Hunyan needs to make the img to video model diffuser!