This is a FLUX LoRa model trained with FLUX.1-dev-fp8, consisting of over 100 images.

The model can generate flat portraits with unique composition and a sense of flow, suitable for illustration design. To compare the effects of different parameters, this model was trained in two different versions, the difference being whether the T5 and CLIP, which are part of TE, were trained, for everyone to test.

The name of this model was chosen based on my intuitive impression of it.

Weights from 0.9 to 1.0 may produce better images (・・)ミ