NOTE: IF YOU FEEL LIKE WORKFLOW IS LAGGY AND NOT LOAD,
1. MOVE THE LTXV PROMPT ENHANCER TO A NEW PAGE,
2. GENERATE THE PROMPT FROM THAT PAGE,
3. COME TO THE WORKFLOW, ADD A TEXT BOX AND PASTE THE PROMPT!
This is just a random workflow sharing with a friend, I didn't make it, i downloaded it, modified it and shared it due to a request. all credit should go to the original workflow
LTXVideo 0.9.6 Distilled Workflow with LLM Prompt | Civitai
I posted my results on reddit using above workflow and i was asked to share my workflow with some friends there
NOTE: change the output to VIDEO instead of GIF which i forgot to do...
mak56410 Thank You for 1K Buzz
floopers966 Thank You for 1K Buzz
Description
FAQ
Comments (15)
how to get the florence and lava bits working? they keep crashing the workflow
Why do you need Florence...? it might not give good results unless you know how to modify the output to make it sound like a motion prompt.. try Ollama, Install Ollama, download a vision model and use the Ollama Node, Ollama node has the instructions to make it work... Thanks!
In my case the workflow crashes when I try to look at the full workflow, I cant even edit the nodes.
@panorios Did you try the Official workflow..? ComfyUI-LTXVideo/assets/ltxvideo-i2v-distilled.json at master · Lightricks/ComfyUI-LTXVideo
@udappkuma Yes, the original workflow is working.
@panorios Its LTXV prompt enhancer, you can use something else or i will modify the workflow and upload tomorrow.
@udappkuma I cant modify the workflow it crashes, thank you.
How can you do landscape mode ? changing the dimensions on which node ?
Those nodes are are top of the LTXV prompt enhancer, I can see it on the screen capture i posted here
I'm still fairly new to the ai stuff, does this model generate nsfw ? having problem with that part
NOTE: IF YOU FEEL LIKE WORKFLOW IS LAGGY AND NOT LOAD,
1. MOVE THE LTXV PROMPT ENHANCER TO A NEW PAGE,
2. GENERATE THE PROMPT FROM THAT PAGE,
3. COME TO THE WORKFLOW, ADD A TEXT BOX AND PASTE THE PROMPT!
I would love to do this but have no idea how.
I tried to move some nodes to another tab in ComfyUI but ofcourse that broke it.
A version of this workflow with just an old fashioned positive and negative prompt would be really great. I'll then run my local Dolphin-Llama to create an uncensored prompt.
This is how you fix the issue properly
https://www.reddit.com/r/StableDiffusion/comments/1k1xk1m/comment/mns8xbj/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
If its complicated, This is the easy method
https://www.reddit.com/r/StableDiffusion/comments/1k1xk1m/comment/mnvqumm/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
@udappkuma I tried editing the .py file and adding both lines, but apparently I did something wrong because it broke the model.
I'll try the other method, thank you.
@tiskloteweer588 The best thing you can do when you make any changes to any file is to rename the orginal to filename.py.original and use a copy of that file to make changes so even it goes wrong, you can delete the modified one and use the original..
@udappkuma Thanks, luckily I learned that part decades ago, never managed to move much past that though :)