It's an interesting alternative to the Tora/Spline idea, which has been very entertaining to use. I suspect it won't be so good at what's effectively pretty much impossible, creating a 3D of say a movie scene where it won't have the context of what someone or something looks like from the front Looking forward to Cog 1.5 if that can be persuaded to run on 24 Gig cards!
The default setting for each generated video is 6 seconds. Is there any limit to the maximum length? Why did I extend the time to 10 seconds (change num_frames from 49 to 80) and the generated videos are all grainy? The memory is 16G .
Hi, when i launch the QUEUE with all parameters like you indicate, i have this message ! "DownloadAndLoadCogVideoModel 'CogVideoXTransformer3DModel' object has no attribute 'encoder'" what can i do please?
Yes, we have Control Net .th-cam.com/video/757I1wf1Slw/w-d-xo.htmlsi=ovuFFjfvK94GL_Qy And there are some more updated CN Model available in updated version
I keep getting this error: Prompt outputs failed validation CogVideoTextEncode: - Exception when validating inner node: tuple index out of range Any ideas?
@@TheFutureThinker I did, I installed the updated comfy wrapper, updated the requirements, and updated my comfy, all the other cog video models and workflows work, not sure why this one is giving me trouble
I also encountered this after updating. Manually removing the custom node directory and reinstalling still didn't fixed it. 'My' version of the CogVideoXWrapper has a 'vae'-input instead of a 'pipeline'-input....
have anyone found a solution? i'm having this problem too, i installed from ComfyUI Manager Menu, then used update all, and update comfyui itself, but it doesn't help
Incredible. Thank you.
Love and blessings!
@5:15 Is it me or does DimX have trouble with the round table & the window bars behind it? It seems like the table morphs in and out of these bars.
Wow. Won't be long now before we can literally have an open sourced Kinng on our potato PCs. Amazing.
Fingers crossed!
Báo lỗi:
Prompt outputs failed validation
CogVideoDecode:
- Exception when validating inner node: tuple index out of range
It's an interesting alternative to the Tora/Spline idea, which has been very entertaining to use. I suspect it won't be so good at what's effectively pretty much impossible, creating a 3D of say a movie scene where it won't have the context of what someone or something looks like from the front Looking forward to Cog 1.5 if that can be persuaded to run on 24 Gig cards!
Yes , looking forward to 1.5, I tried in server side btw, looks ok at the level of Mochi server edition, and higher res.
Oftentimes I find the hallucinations more entertaining
Is it has NSFW Capabilities?
The default setting for each generated video is 6 seconds. Is there any limit to the maximum length? Why did I extend the time to 10 seconds (change num_frames from 49 to 80) and the generated videos are all grainy? The memory is 16G .
I did another method for extending video, yes its limited to 48 for img2vid
@@TheFutureThinker How to extend video time?
Hi, when i launch the QUEUE with all parameters like you indicate, i have this message ! "DownloadAndLoadCogVideoModel
'CogVideoXTransformer3DModel' object has no attribute 'encoder'" what can i do please?
can cog do generative image to video, or only the control net panning styles? O may have missed an earlier video or forgotten.
Yes, we have Control Net .th-cam.com/video/757I1wf1Slw/w-d-xo.htmlsi=ovuFFjfvK94GL_Qy
And there are some more updated CN Model available in updated version
great video, thank you!
Im gonna try this on my 3080ti 12gig vram and let you know!
Yes please
that is insane stuff
that is awesome !
try it tomorrow ;)
I keep getting this error:
Prompt outputs failed validation
CogVideoTextEncode:
- Exception when validating inner node: tuple index out of range
Any ideas?
Update your comfy, update the custom node.
@@TheFutureThinker I did, I installed the updated comfy wrapper, updated the requirements, and updated my comfy, all the other cog video models and workflows work, not sure why this one is giving me trouble
I also encountered this after updating. Manually removing the custom node directory and reinstalling still didn't fixed it. 'My' version of the CogVideoXWrapper has a 'vae'-input instead of a 'pipeline'-input....
have anyone found a solution? i'm having this problem too, i installed from ComfyUI Manager Menu, then used update all, and update comfyui itself, but it doesn't help
@@elukonoe990 No solution found, would love to hear a solution!
how can we extend video using Cogvideox ?
Thats great idea, will craft one later.
Can I run this on Linux and with just a 2060s RTX ? what are the requirements
No, I don't think you can use CogVideo5B on your PC. You can use Runpod though, maybe 24 Gigs VRAM?
I ran this and consume about 7-10GB VRAm
@@aerlantil i heard that in linux you will need more vram than in windows,
@@blender_wikiand you need to get a PC 😂😂