These LoRas will help you create anime images using pretty much any SDXL & SD 1.5 checkpoints.
As far as I know this LoRa works with all SDXL & SD 1.5 checkpoints (or most of them at least).
For those who like a more modern style rather than the classic anime style, I've uploaded the LoRa AniME
The reason I don't upload them together is because they do have a distinct difference in style, especially on higher strength (1.0 and above). While AnimeAnything gets a more classic anime look at higher strength, AniME has a more modern 3D rendered style.
Get them both, they are both really good but for different things.
For SDXL (scroll down for SD 1.5 info)
Recommended settings (will most likely work on a wide array of settings)
Steps: 30-40
CFG: 5-9
Sampler: Euler a
LoRa strength:
Without the use of "Anime" in prompt ~1.0-2.2
With the use of "Anime" in prompt: ~0.5-1.3
Example:
Checkpoint: Hassan SDXL
This is a first test version of this LoRa and it's large (469 MB), and I'm planning to work on a smaller version without losing too much of the effect.
Note: If you are using Automatic1111, Forge or similar web UI, you might have to change your settings for the LoRa to show up in your UI.
Settings --> Stable Diffusion --> Extra Networks
For SD 1.5
Recommended settings (will most likely work on a wide array of settings)
Steps: 23
Sampler: DPM++ 2M SDE SGMUniform
Cfgscale: 4.5
LoRa Strength: 0.7-1.4
Example:
Checkpoint: Realistic Vision V 5.1
Description
Created on request.
FAQ
Comments (7)
I get this error from A111's networks.py in SD1.5:
AssertionError: Could not find a module type (out of ModuleTypeLora, ModuleTypeHada, ModuleTypeIa3, ModuleTypeLokr, ModuleTypeFull) that would accept those keys: diff_b
Odd. I haven't tried A1111 specifically though, only Forge, Comfy and Swarm. I'll try to look in to it tomorrow, if I have time.
Have you gotten the same error with different checkpoints? Which checkpoint(s) have you tried and gotten the error with?
I've searched around some, and issue seems to be with xformers. Try to re-install xformers with this:
pip3 install -U xformers --index-url https://download.pytorch.org/whl/cu121
@Creepybit I am not using xformers, but directml, and have never seen this error before or since.
@BitesizedTask64 Sorry, I have no solution and have no idea why you get this problem. As far as I know, no one else has had this or similar problem with the lora.
I uploaded another lora yesterday, that's similar to this one. Maybe you could try that one. If you do, and get any error messages on that as well, let me know.
https://civitai.com/models/567810?modelVersionId=632776
@Creepybit I'm getting an identical error. I also tried moving it into the LyCORIS folder to no avail.