These workflows demonstrate the use the new Audio Visualizers, as well as a wide variety of other reactive nodes, that I have introduced to ComfyUI by way of my node suite.
See model version details for corresponding information and YouTube tutorials --->
If you feel like it, give me a star on GitHub.
Feel free to join us at the YouTube channel; we are cooking with GASOLINE over there.
Common models used often:
Animate LCM https://huggingface.co/wangfuyun/AnimateLCM/tree/main
Ral Finger's loras https://civarchive.com/user/RalFinger
Detail lora https://civarchive.com/models/82098/add-more-details-detail-enhancer-tweaker-lora
Photon https://civarchive.com/models/84728/photon
Photon LCM https://civarchive.com/models/306814/photon-lcm
FAQ
Comments (20)
Well fuck me, I have always wondered if there would be something cool enough to force me to learn comfyUI.
Hold my beer, I'm going in.
FYI the videos you're sharing (when downloaded) behave maliciously in MPC-HC (details here) apparently the inclusion of metadata causes my video player to generate a unique dialog error box every few seconds, and since I have videos set to automatically loop there were dozens - it felt like an attack vector was being exploited.
Seems like it's fairly easy to keep that out of your MP4 but perhaps it's only the video player of champions that gets afflicted here.
ComfyUI optionally saves the workflow in the metadata of the files. I include if for people interested in the variation of the workflow that produced the video. Player of champions? VLC? I am not having this issue and its the first im hearing of it - edit: tbh VLC has be sucking lately
Yep, looks like I will have to learn comfy UI as well
@blhll yessssssss join us
Oh wow this look sick. Time to revive my 300 SD1.5 loras from the grave and play with this!
I am curious to know why the decision to do this with 1.5 was made? Is there any problem using this with xl?
you can use XL, I just find it to be totally inferior for video, and slow and memory intensive.
Only place I've seen XL thrive was with the Unsampling workflow from Inner Reflections. That workflow requires a video input through, I wonder if it could be used in these workflows.
https://civitai.com/models/544534/unsampling-workflows
@Catz i was considering adding sigmas as a target for the feature system in my nodes
@ryanontheinside That'd be interesting! I'm curious to see the different results. I'm still testing various lora mix with the Advanced Visualizer. Sometimes I get blurry/foggy results and not completely sure why, I'll have to play with the settings more.
For ease of use, I've replaced the lora stack with the Lora Loader Stacked Advanced (https://github.com/idrirap/ComfyUI-Lora-Auto-Trigger-Words) to automatically feed the trigger word in the prompts. Example of it can be see on 2/4 of the left of this image https://civitai.com/images/34778309
I love how you've extracted various parts of the music and reconstruct them visually, it's really interesting.
@Catz @Catz ill have to check out that auto trigger word, what a time saver cant tell you how many times ive looked up the same loras on civit
fyi, you can swap out audio for any other feature source (motion, MIDI, proximity, etc etc) :D
@ryanontheinside Here's the workflow with the lora triggers setted up
https://www.mediafire.com/file/mgr9cwvp7nqsb73/AudioReactiveDiffuser.json/file
@ryanontheinside now we are talking, I just dont have any 1.5 resources left on my system^^ its all pony now
@JohnnyB1 all of my node suite is independent of the underlying model, you can totally use any models!
size isnt everything fellas
I like it
Fanks
I love it
I am not sure what I am doing wrong. Please help.