Kalpana XL is trained on over 12k images from MidJourney and Stable Diffusion v3. This training helps it create high-quality images from text prompts. The model excels in producing various styles, including realism, artistic, anime, and surrealism.
Recommended Settings
VAE: Baked in
Sampling Method: DPM++ SDE, Euler a
Sampling Steps: 20 - 60
CFG Scale: 3 - 10
Hires Steps: 10 - 20
Denoising Strength: 0.3 - 0.5
