V2.0 is out! This version has far better quality, consistency and stability than the previous versions. You can verify by checking the showcase images. It also learns her hair color properly so no need to add it to the prompt.
V2.0 achieves the quality bar I have - it is as consistent as my Asuka model on SD1.5, and even more faithful to the anime art style. Besides, Flux is so much better in following complex prompts. No need to use clothing LoRAs in most cases, simply describe the clothes and you can have Asuka try them on!
The trigger word is "ppasuka", to avoid the model picking up its existing data on Asuka.
Description
Retrained model with a further refined dataset and lower network dim of 4. The outcome is much more consistent and stable, making this model ready to graduate from prototype.
FAQ
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.