A fine tuned diffusion model that attempts to imitate the style of late '80s early 90's anime specifically, the Ranma 1/2 anime. Use "80sanimestyle" in your prompt. This model benefits a lot from playing around with different sampling methods, but I feel like DPM2, DPM++ and their various ititerations, work the best with this model.
This model was trained using TheLastBen's fast-stable-diffusion for dreambooth. Model was trained on 121 screenshots from Ranma 1/2, with 3000 steps, and Anything-V3.0 as the base model.
Description
FAQ
Comments (14)
if you type in ranma ... in sd 2.1 you get images like this so i dont get what this exactly does
Well, if you read the description you'd know it was made with SD 1.5, (which isn't good at this style,) before SD 2.1 was released. And you know there are ppl who prefer to use 1.5.
how did you know about the ranma keyword, and what other keywords are there beside ranma! please help. Thanks
@mnazmi76408 for this checkpoint all you need is "80sanimestyle," for something like Anything V3 or one of it's many, many merges, adding "80's (style)", and by "Rumiko Takahashi," should do the trick.
@tasha28 But where are all these keywords listed!?
@mnazmi76408 You don't need those keywords for this model. It says it right there, in the second sentence of the description and in the info box to the right. All you need for this particular checkpoint is "80sanimestyle," somewhere in your prompt. That's it, this checkpoint was specially trained, so "80sanimestyle," triggers the effect. This is just a style checkpoint, so it wasn't trained with the character names. Though I am working on a V2 of the LORA that's trained with the character names.
@tasha28 Thanks... I am just wondering (regardless of the model) how do you guys know about all the keywords out there
@mnazmi76408 A lot of experimentation and trial and error, the only way you know a concept is in the model is to test it. Then people did research into artists and art styles and they tested those. If something is well known, then it's safe to assume the model knows it because the original data set is massive.
Is it possible to run this on Automatic1111? Keep getting error whenever I try to run the model. Thanks
It should work, because I use that too. Are you running it with x-formers?
@tasha28 got it working now! seemed like an issue with my automattic1111. Thanks for replying though :)
Works beautifully in ComfyUI!
I NEED TO GET A BETTER LIFE AND A JOB to download models like this ... XD Ranma 1/2 is life!
哦太棒了！√
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.