April 28, 2024: added V2 Rebirth pruned
v2:REBIRTH
Thanks to S6yx for the creation of this beautiful model. Enjoyed by millions. With their permission, I, Zovya, will be maintaining it moving forward.
April 4, 2024: fp16 and +VAE added
April 2, 2024: Rebirth
Update 3: Disclaimer/Permissions updated
Update 2: I am no longer maintaining/updating this model
Update 1: I've been a bit burnt out on SD model development (SD in general tbh) and that is the reason there have not been an update. Looking to come back around and develop again by next month or so.Thank you everyone who sends reviews and enjoy my model
Pay attention to the About this version section of model page for specific version information. ➡️➡️➡️➡️➡️
Model Overview:
rev or revision: The concept of how the model generates images is likely to change as I see fit.
Animated: The model has the ability to create 2.5D like image generations. This model is a checkpoint merge, meaning it is a product of other models to create a product that derives from the originals.
Kind of generations:
Fantasy
Anime
semi-realistic
decent Landscape
LoRA friendly
It works best on these resolution dimensions:
512x512
512x768
768x512
VAE:
Prompting:
Order matters - words near the front of your prompt are weighted more heavily than the things in the back of your prompt.
Prompt order - content type > description > style > composition
This model likes: ((best quality)), ((masterpiece)), (detailed) in beginning of prompt if you want anime-2.5D type
This model does great on PORTRAITS
Negative Prompt Embeddings:
Make use of weights in negative prompts (i.e (worst quality, low quality:1.4))
Video Features
Olivio Sarikas - Why Is EVERYONE Using This Model?! - Rev Animated for Stable Diffusion / A1111
Olivio Sarikas - ULTRA SHARP Upscale! - Don't miss this Method!!! / A1111 - NEW Model
AMAZING SD Models - And how to get the MOST out of them!
Disclaimer (Updated 10/31/2023):
The license type is CC BY-NC-ND 4.0
Do not sell this model on any website without permissions from creator (me)
Credit me if you use my model in your own merges
You can use derivative models which uses ReV Animated for Buzz points and site-based currency that does not convert over to real world currency.
Do not use this model to monetize on other platforms without expressed written consent.
Description
Inpainting version of model v1.2.1
FAQ
Comments (35)
This is one of the best models in civit. But the only thing which bothers me is that REV has a tendency to create wide foreheads... any suggestion on how the balance out that effect?
Now that I think about it... you're absolutely right!
oh no cant unsee
yeah also very sharp and defined Jaws. The model itself is amazing at lighting and a sense of depth. Just gotta fight against the same face big head sharp jaws.
This checkpoint is wonderful! My LoRAs look great in this model. Curiously I could not retrain them using this model, that generated errors and failed to complete, but I seem, to still be able to use the model to render them. Nice work!
Great model, used it for some Deforum clips like: https://youtu.be/uzgHixu7xRQ
I'm getting this error when using this model
" A tensor with all NaNs was produced in VAE. This could be because there's not enough precision to represent the picture. Try adding --no-half-vae commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check."
Same
try changing sampling method worked for me .
I get this error All the time, AFTER the first run;
if I redownload the model, it works again, for ONE time, then I need to redownload it again after...
@pixelrealmetics680 Change to Euler a worked for me. Thank you!
@NaughtySkynet2 i got this error and it gave me this solution.Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion. this worked for me
"Try adding --no-half-vae commandline argument to fix this."
@SmartNeto in cmd prompt or bat file?
@snakeviper18 if you run it using cmd just add it there, if you dont, then add it to webui-user.bat on COMMANDLINE_ARGS
@SmartNeto ok
lora loha lokr ia3...... can not train
Without a doubt my favorite new model, It's just amazing.
hey great model!
any feedback on the bayesian-merger?
i like the functionality of the tool however it seems to only keep unet weights and drops off vae, clip and sd weights which causes the model not to work well with invoke and comfyui
@s6yx Any plans on releasing a full version of 1.2.1 to work better with invoke and comfy?
EDIT: nvm, saw other comments about 1.2.2
The bayesian merge people have heard this was a problem and adjusted their tool.
@s6yx that problem should be fixed now!
@s1dlx new changes?
@s6yx bug fixes for now. We’re working on guided optimization and new merging modes. The former should be coming very soon
Different seeds generates almost identical results.
Any thoughts?
i noticed that too, its odd
another video with stable diffusion
Wow that is pretty awesome! Is this your own process or is there a documented process available somewhere?
did you launch a missile yesterday
@King_J0 Rev animated, did you launch a missile yesterday
I'm really torn between this model. On one side it generates absolute amazing detailed results. On the other side the faces are unfortunately always unusually "round" with big foreheads and have that anime like style somehow (which I don't like much). :/
I'm having issues adding the new build to InvokeAI, which keeps complaining about a size mismatch. Should the new version be less than 2gb, or did the upload get corrupted?
Details
Files
revAnimated_v121Inp-inpainting.safetensors
Mirrors
revAnimated_v121Inp-inpainting.safetensors
7371_revAnimated_v121Inp-inpainting.safetensors
revAnimated_v121Inp-inpainting.safetensors
revAnimated_v121-inpainting.safetensors
8_revAnimated_v121Inp-inpainting.safetensors
revAnimated_v121Inp-inpainting.safetensors
revAnimated_v121Inp-inpainting.safetensors
revAnimated_v121-inpainting.safetensors
revAnimated_v121-inpainting.safetensors
revAnimated_inpainting.safetensors
revAnimated_v121Inp-inpainting.safetensors