Test LoRA that I created just to learn, turned out pretty decent.
The style is anime-like with painterly, expressive brushwork.

I highly recommend trying this lora with the new Flux Krea model, it produces better results than base Flux.D with this lora in my opinion!

HiDream has 2 versions, a "Creative" one which will hallucinate more, but will give more creative and detailed results. And a "Stable" version which will give more consistent reliable results, with the drawback of being less creative.

Flux (applies for both v1.0 and v2.0):

• Recommended strength is around 1 - 1.5 range (lower values have smoother texture, while higher values give you more textured brush strokes)

HiDream:

• Recommended strength is 1.0

Trigger word is "m3ssyStyl3", for example you can start your prompt like "m3ssyStyl3 art style image showing a...", "stylized m3ssyStyl3 illustration showing a..." etc.

Update: Flux v2.0 has been released, it has been trained on a dataset that is about twice the size of v1.0, with better captions, longer training with higher resolution. This results in noticeably higher quality generations, with less artifacts and hallucinations. Also v1.0 had bias towards purple hair, if you didn't specify hair color it would gravitate towards purple, with v2.0 this has been fixed. I hope y'all like it!

Any kind of support is very much appreciated! If anyone is interested in custom LoRAs, you can always DM me! (˶ᵔ ᵕ ᵔ˶)

Image comparisons for Flux bellow (strength comparison between 1.0 and 1.5):