Note: If you have ANY issues with nodes not downloading, read the notes or reach out. There's nothing that special about any of them that aren't core modules.
⛔⚠️🛑✋ Read the notes completly before using. Most common install and node problems are listed in the directions
Instagram: https://www.instagram.com/synth.studio.models/
Buy me a☕ https://ko-fi.com/lonecatone
This represents hundreds of hours of work. If you enjoy it, please 👍like, 💬 comment , and feel free to ⚡tip 😉
也有中文说明
This is a 12gb VRam or more workflow.
Features:
Allows you to use your own audio track for🫦 lip sync, preserving voices
Easy setup and use
Motion adjustment for more dynamic videos
Multimodal guider for fine tuning audio to video dynamics
Prompt generation from an image
Prompt enhancement
Handles both Normal and NSFW generation
也有中文说明
This is stright up one of the most badasss labor intensive workflows I've ever made. Please give me feedback on what does or does not work and the setting you guys use.
Description
FAQ
Comments (16)
I really learn a ton from your workflows. Thank you for imparting your knowledge !
This was a pain.... holy.....
Possibly the coolest thing I've ever made.
@lonecatone23 Genuinely, looking at and picking apart the examples you've presented has given me the equivalent of divine epiphany over the past few days. Like a chimp trying to dissect a smartphone LOL
I'm skipping bronze age straight to nuclear tech right now. Can't thank you enough !
@lucilucilu27765 That's how I figure stuff out. It's also why I wrute things teh way I do, sotha other's can just grab entire modules.
Glad you are learning. Makes me happy 😊
@lucilucilu27765 Hey, can you see teh workflow image and videos attached to this? I think the loader is acting up
@lonecatone23 I was very tempted to just grab and run it at first. But the more I try to open the seams, the more intriguing your whole system is ! Idk, it really lit the curiosity in me.
@lonecatone23 I can see the 1 image of the entire workflow up there, but nothing else. that was not by design ?
@lucilucilu27765 No. There are 3 or 4 videos uploaded, but they still say "processing" and I uploaded it 14 hours ago. I'll hit up support
@lucilucilu27765 Let me know if you can see them now
@lonecatone23 Yes, they are visible now. Finally.
Looks interesting - I'm experimenting with it now. If anyone has the same problem I had with this workflow (interpolation node failing to download rife47) make sure you have a folder in the following location - "custom_nodes/comfyui-frame-interpolation/ckpts/rife/", then download a copy of the rife you want - versions 4.7 and 4.9 are recommended by the author of this workflow, then fully restart comfyui. I spent ages troubleshooting this with chatgpt but got there in the end.
Wow, really? I didn't know that node was troublesome. it's in every Video workflow I've ever made and this is teh first I'm hearing about it.
Thanks for the update. I'll add something to my notes.
Out of curiosity, are you using the desktop version?
@lonecatone23 I'm using ComfyUI Portable on Windows 11 - maybe my firewall is blocking the repository for some reason (?). I thought it worth mentioning in case anyone else hit the same issue. It could save them a bit of time. Unrelated, do you have any specific settings recommendations for using image to video with realistic images? I've tried a couple and the quality is very crisp but almost too detailed to the point where it looks really odd (and the colour has also been washing out of the original image). I've tried reducing the strength of the lora detailer to 0.4 as that seems to be a common setting in workflows but it doesn't make much difference.
@bennyboy_77 I'm still experimenting. I like the lip sync and video consistency a lot, but it is clean.
Drop the steps down to 20 and use Euler or Euler A
As far as color wash, make sure SEED VR2 is set to LAB. It excels at color correcton, or slap in a color correction node between it and the final video
It seems this cannot run on an RTX 3090 graphics card; my error message is as follows:
FP8 quantized models is only supported on GPUs with compute capability >= 8.9 (e.g 4090/H100), actual = 8.6
No. On 6gb VRAM? No, it need twice that. I thought I put that in the notes