Test the model in TensorArt

I created this merge with the goal of building a versatile model capable of handling a wide range of art styles — from comics and graphic novels to manga, digital paintings, 3D renders, and traditional media like oil painting or watercolor. While anime models dominate the space (and I’ve mostly worked with anime myself), I felt it was time to explore and support other aesthetics that are underrepresented.

This model isn’t perfect — it can sometimes misinterpret prompts or lean too far into realism — but with a bit of patience and prompt tweaking, I believe it can produce some unique and impressive results. I’ll continue refining it as I experiment further to reach the vision I have in mind.

Showcase image settings:

• Sampler: DPM++ 2M SDE Karras / Euler A

• Steps: 20-40

• CFG: 4-6

Hires. fix:

• 4x Ultrasharp

• 10-20 steps

• Upscale by 1.5

• Denoising strength of 0.3-0.6

These settings are a suggestion, feel free to experiment and find the combination that works for you.

I hope you enjoy using it! 🩵🌻 Feedback and suggestions for improvement are always appreciated.

EDIT: Using artist names seems to give very good results, give it a try ✨