Trained on Tsubaki, 15 images.
Applies for DiT V1: Remove the "multicolor hair" prompt before you begin generating. I tried, but the AI is still not near to replicating her colourful hair. Otherwise, I'm happy. Another LoRA I bring to DiT V1, I'm very satisfied with this evolution. The characters look much closer to the shows. Though I have to confess that I did enjoy the different art style model back in the SD era, and I miss it.
I use Euler A. 28 steps. Because the results are pretty good, I think you can generate 3 images in parallel, and you will still be fine. Unlike SDXL, DiT is pretty consistent from my initial experience. Of course, if you demand a lot of detail, then maybe go cheap first.
I use 3, 4.5 and 6CFG. 0.7 Rescale CFG, same as every DiT V1 LoRAs I use.
Thumbnail model: Tsubaki
