if you read comments but not the description, this is for you - this pack was already updated into a single workflow, with improved results - link is in description! UPDATE v2 !
I can see at the bottom there is a image viewer. I am trying very hard to find the plugin in node manager or is it a settings in comfyUI can you tell me how to get it?
My experience has been LivePortrait often breaks if your input video isn't clean enough for it. Any Background and/or just a tad too much movement or odd angle will often break it. The Eye and Lip tracking are just as finicky and will default to nothing if you don't go too far. I don't have enough info to be reproducable, is just what i recall off the top of my head.. cheers!
there will be updates on this workflow, i found that the white background is the best with mimicmotion, new compact all-in-one workflow will land tomorrow, i'm going to build a "faster" version with fire and forget function so you can set it up and "go outside" while it cooks. i'm aiming for an isolated character for the "top layer", then we can put it all back together.
@@bbrother92 RotoMaker is my custom built colection of workflows, it combines the MimicMotion (SVDXT1.1) with ReActor FaceSwap (inswapper/codeformer). I have another toolset named Loki which has all the info you need: ruclips.net/video/zfv0gH-RuB4/видео.html
thanks for the info, personally need a workflow for faceless RUclips, with an avatar, most likely sitting behind a desk and talking (I like to insert some special effects, the character is steampunk so like steam coming out of his head or something) and since English is my third language, hehe... I most likely use TTS, I really appreciate it if you could guide me as best as you can, I am building my PC for ML soon
Glad it was helpful! To address the rest :) we are slowly building towards some things like that, the speed is almost good enough and i will be building some webcam driven stuff. speed is usually a compromise of quality, or some tradeoff. New collection is planned in near future as for TTS, i have recently looked at some of them, but most are old or tricky to install (might break existing stack) so it's also on the radar !
@@FiveBelowFiveUK thanks for the reply, really glad to hear that, for the influencer avatar, some work doesn't need lip sink just a general good quality animation and talking, and some reaction animation, in my case, my avatar is a cyberpunk robot, so if I want to put steam coming out of his ears or shortcircuiting, I have my work cut out for me, that's why I need help hehe 😅
Thanks - you have some choices :) there is always new tutorials here on the channel, and soon for members some exclusive tutorials and guides. I am always available for one on one training, and have been doing this for over a year now, join the discord and send me a DM, link in description.
if you read comments but not the description, this is for you - this pack was already updated into a single workflow, with improved results - link is in description! UPDATE v2 !
I can see at the bottom there is a image viewer. I am trying very hard to find the plugin in node manager or is it a settings in comfyUI can you tell me how to get it?
image feed is part of Pythongoss pack: github.com/pythongosssss/ComfyUI-Custom-Scripts
My experience has been LivePortrait often breaks if your input video isn't clean enough for it. Any Background and/or just a tad too much movement or odd angle will often break it. The Eye and Lip tracking are just as finicky and will default to nothing if you don't go too far. I don't have enough info to be reproducable, is just what i recall off the top of my head.. cheers!
there will be updates on this workflow, i found that the white background is the best with mimicmotion, new compact all-in-one workflow will land tomorrow, i'm going to build a "faster" version with fire and forget function so you can set it up and "go outside" while it cooks. i'm aiming for an isolated character for the "top layer", then we can put it all back together.
@@FiveBelowFiveUK sir can you tell please - what did you use to make this talking avatar?
@@bbrother92 RotoMaker is my custom built colection of workflows, it combines the MimicMotion (SVDXT1.1) with ReActor FaceSwap (inswapper/codeformer).
I have another toolset named Loki which has all the info you need:
ruclips.net/video/zfv0gH-RuB4/видео.html
@@bbrother92 and don't miss the V2 update which landed yesterday, most people missed it :D
@@FiveBelowFiveUK jesus, are you programmer?
thanks for the info, personally need a workflow for faceless RUclips, with an avatar, most likely sitting behind a desk and talking (I like to insert some special effects, the character is steampunk so like steam coming out of his head or something) and since English is my third language, hehe... I most likely use TTS, I really appreciate it if you could guide me as best as you can, I am building my PC for ML soon
Glad it was helpful! To address the rest :) we are slowly building towards some things like that, the speed is almost good enough and i will be building some webcam driven stuff. speed is usually a compromise of quality, or some tradeoff. New collection is planned in near future
as for TTS, i have recently looked at some of them, but most are old or tricky to install (might break existing stack) so it's also on the radar !
@@FiveBelowFiveUK thanks for the reply, really glad to hear that, for the influencer avatar, some work doesn't need lip sink just a general good quality animation and talking, and some reaction animation, in my case, my avatar is a cyberpunk robot, so if I want to put steam coming out of his ears or shortcircuiting, I have my work cut out for me, that's why I need help hehe 😅
Just AMAZING! Bro, you can teach me? Please, my job depends of this! I dont have lot of money but I PAY for this!
Thanks - you have some choices :) there is always new tutorials here on the channel, and soon for members some exclusive tutorials and guides. I am always available for one on one training, and have been doing this for over a year now, join the discord and send me a DM, link in description.