🐨 SDXL version of this LoRA works best with ForCute checkpoint. You can check some of my old images here
⟿ https://civitai.com/images/22478585
⟿ https://civitai.com/images/22174978
⟿ https://civitai.com/images/21616340
🐻❄️ Best weight: 0.8 - 1
Trigger words: p1nkch1b1p3ts, chibi animal
🐵 Recommended settings that I used in all of my images generated with ForCute:
⟿ Sampling Method: DDPM
⟿ Sampling Steps: 20
⟿ Upscaler: SwinIR _4x
⟿ CFG Scale: 2
⟿ Denoise Strength: 0.3
⟿ Hires steps: 15, upscale by 1.5-2
🦄 I also tested this LoRA with SDXL FaeTastic checkpoint and the best settings I've figured out for this checkpoint to achieve the style is:
⟿ Sampling Method: Euler
⟿ Sampling Steps: 20
⟿ Upscaler: 4xUltrasharp_4xUltrasharpV10
⟿ CFG Scale: 4
⟿ Denoise Strength: 0.3
⟿ Hires steps: 15, upscale by 1.5-2
(If you use different settings or checkpoints, the results might or might not be more desirable. However, please let me know if you come up with a better settings).
･˚‧･+‧₊‧.°.⋆.🫧.•˚₊‧⋆:｡+.･
🐰 I had decided to release this LoRA for SDXL, so when I heard Flux on-site training was available, I was like "Why not making one for Flux first?" Hence, here comes my very first Flux LoRA 😅.
🦥 This LoRA aims to chibify animals and make them look cuter. I've been testing it over and over and it seems to be consistent. (Speaking of which, a very special thanks to @AIDigitalMediaAgency and his awesome Workflow to test the LoRA 💜). However, since this is my first time training a LoRA for Flux, please do be expected it won't be perfect. I might improve the LoRA later on if needed.
🐯 There is no trigger word but all of my prompts include cute + [animal].
🐼 There is another Flux version of this which is less chibi, tho I need to do more tests and I might release it if it works okay enough. SDXL version will be released later today or tomorrow.
🐧 Once again, thank you for trying out my LoRA. I hope it'd help you to create lots and lots of cuteness 😍
･˚‧･+‧₊‧.°.⋆.🫧.•˚₊‧⋆:｡+.･