Depth2Diff
HEY HI HELLO FRIENDS! lmfao. This is an 8gb VRAM friendly Image to Vid workflow for comfyui that makes an animate diff depthmap and then diffuses that ''DiffMap'' along with depthmap controlnets.
You're gonna need this Depth Map Lora from Machine Delusions too: https://civitai.com/models/392921/depth-map-lora-sd15?modelVersionId=438287
example results:
https://civitai.com/images/13878829
If you start at 320x512 or 512x320 (resize node) then you can hires fix up to 1.6x
If you start at 512x512, 512x768, or 768x512 you can hires fix up to 1.2x or 1.25x
The workflow is color coded in rainbow so that you can sorta follow what it does and how it flows.
Its arranged in such a way so that you can understand it and see whats happening. It reads left-to-right...-ish
Its simple and basic. Good for beginners, or people who want a starting point to build off of.
If you change anything or add stuff it may or may not work within the 8gb VRAM Restraint so if you modify it and it breaks, thats on you to figure out.
It should go without saying but...
YOU WILL PROBABLY NEED TO GO THROUGH AND CHOOSE DIFFERENT MODELS AND LORAS BECAUSE YOU AINT GOT WHAT I GOT CUZ I GOT CUSTOM STUFF OR RENAMED SOME SHIZZ OK GEEZ BRO DAYUM SON I GOT UR BACK THO AND THIS IS THE PART WHERE YOU SAY GOOD LOOKIN OUT FAM GEEZ
(just playing but yeah, youll need to go select models, loras, diff models, clip model, etc)
OH OH OH OH OH WAIT WAIT HOLUP
Before you do anything, que it up once so that it can fail, then go to your comfy manager and...
INSTALL MISSING NODES
then do the whole ''UPDATE ALL'' thing
yeah
yeah
then restart comfy.
That will prevent like 90% of the problems that arent user error, like.. before you even waste your time coming up in here tryina be like BRUH I ERROR 4REEL HELP ME
(just kidding again but yeah it will prevent a lot of issues so there)
This is just a screenshot of the workflow. Youll need to download it from the button up there.
HOLLA BACK
Socials and Stores:
https://linktr.ee/pookienumnums
