circlestone-labs/Anima

Updated 1 day ago • 338k • 1.12k

Note SOTA. Best prompt following you'll get out of an anime model. Downsides were mitigated by the distillation loras (some inconsistencies, slow gen).

hanzogak/Anima-Comradeship

Text-to-Image • Updated 14 days ago • 5

Note Cosmos-Predict2.5-2B-base-distilled-LoRA is by far my favorite distillation lora, but they all have their own strengths so I'll keep a list here.

Laxhar/noobai-XL-Vpred-1.0

Text-to-Image • Updated Nov 19, 2025 • 11.1k • 66

Note Where everything really started. Still useful to keep for the very large ecosystem of controlnets, lora etc. SDXL based models are also better at mixing styles.

tianweiy/DMD2

Text-to-Image • Updated Jun 11, 2024 • 52.9k • 236

Note the reference, for SDXL based models

nblight/noob-ft

Updated Jun 13, 2025 • 2

Note noob-ft-1536x-extract.safetensors brings higher resolution generation to noob models. Use at 0.5 strength, it's surprisingly good. More effective than whatever Illustrious XL 2 tried.

2vXpSwA7/iroiro-lora

Updated Nov 10, 2025 • 687

Note 2vXpSwA7/iroiro-lora/anytestv4*illustrious are simply the best multicontrolnets for the model. It's a bit like xinsir controlnet union.

MIC-Lab/illustriousXLv0.1_controlnet

Updated Nov 25, 2025 • 11

Note illustriousXLv0.1_Tile_fp16.safetensors is THE model for your upscaling needs.

SmilingWolf/wd-v1-4-moat-tagger-v2

Updated May 20, 2023 • 128 • 106

Note fast booru tagger

heathcliff01/Kaloscope2.0

Updated Nov 7, 2025 • 23 • 22

Note Saw a picture you like but want to know which artists in the booru datasets are most like it (or even the right match)? This unique model has the answers.