This model is trained as Tamaki, a character from Princess Connect! with her origin and summer swimming outfit. It's based on SD1.5 model, and would competiable with other checkpoints based on SD1.5. Safe to blend with other LoRA, LoCon based on SD1.5 as well.
Triggered with "tamaki" for character appearance
Origin Outfit:
"purple eyes, animal ear fluff, short hair, ((neck bell)), thighhighs, (((japanese clothes))), (geta), black legwear, (detached sleeves), (fingerless gloves), tamaki \(princess connect!\),"
Summer Swimming Suit:
"purple eyes, animal ear fluff, black bikini, hair_bell, neck bell, swimsuit, twintails, very long hair, tamaki \(princess connect!\),"
Suggested weight 0.8-1.0, works fine with 0.9.
Tested with Chilloutmix Ni, Anything 4.5, AOM3A3, CounterfeitV25, meinamix_meinaV8, perfectWorld_v2Baked, neverendingDreamNED_bakedVae, check the samples for details (Hires Fix with latent could be used, use 4x-Ultrasharp in Extras directly if you are fine with your sample outputs).
It's not trained to change to other outfits, still it's possible, lower the weights while enhance the alternative outfit description as my previous models.
**It's a quick cook, and not enough assets to train the model, so I blend 2 outfits together to balance a bit, but should be stable enough to use, play with the prompts and have fun. :)
Description
FAQ
Comments (3)
Can you share your training set? I recently trained for the same role,it would help me a lot if you could share it
please check my raw data set here: https://drive.google.com/file/d/1t4TV_rkCSoPxZGxYy9UUxdtwkyMF251X/view?usp=share_link, just a few images i scrap from google image and danbooru. hope you could improve it!
my workflow to curate the data sets includes upscale the source images with 4x-Ultrasharp for 2x to 4x, then corp and resize to 512x768 for trainning, i found this will improve quality. :)
@rieslin thanks
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.








