Vete ┃ Nightfall Anime ILL is a refined anime-focused model built around contrast, control, and visual discipline. It is designed to deliver clean, high-accuracy anime rendering with a strong emphasis on lighting logic, shadow separation, and emotional clarity rather than visual noise or exaggerated effects.
This model balances softness and darkness with precision. Highlights are deliberate, shadows are shaped, and forms remain readable even in low-light scenes. Nightfall Anime ILL excels at preserving facial structure, eye alignment, and line cleanliness while allowing light to define mood instead of overpowering the image.
Rather than pushing extreme stylization, the model prioritizes stability and consistency. Skin tones remain smooth without plastic shine, hair retains structure under directional light, and monochrome or low-saturation outputs remain crisp and intentional. It performs reliably across bright interiors, dim environments, backlit scenes, and high-contrast compositions.
Nightfall Anime ILL is suitable for direct use as a standalone model and also functions well as a clean base for merging when controlled lighting behavior and structural accuracy are required. It is not experimental and not chaotic. It is tuned for artists who value readable compositions, controlled highlights, and dependable anime aesthetics.
This model reflects a focus on quality over quantity and is part of a broader direction centered on high-end, disciplined anime rendering rather than trend-driven excess.
Description
Version Negatives: fused_digits, 3digits, 4digits, (6digits:1.5), missing_digits, bad_hands, extra_digits, extra_limbs, bad_anatomy, bad_proportions, deformed, poorly_drawn, lowres, blurry, overexposed, underexposed, color_banding, chromatic_aberration, compression_artifacts, watermark, text, logo, signature,
FAQ
Comments (2)
Hi, I really like the design philosophy behind this model — the emphasis on "visual discipline" and controlled lighting is exactly what I'm looking for.
However, I'm experiencing severe color channel issues when generating images. The output gets dominated by a single color (yellow/green), with other channels appearing dead or extremely muted.
What I've tried:
- Used your exact prompts and settings from the preview images
- Same resolution as your examples
- Switched VAE to sdxl_vae.safetensors (external, not baked-in)
- Verified file hash matches (E090E8EB90)
Hardware: RTX 4080 12GB, ComfyUI
The symptom looks like a latent space color mapping issue — possibly something went wrong during the merge process?
Would love to use this model once the issue is resolved. Let me know if you need more details for debugging.
It is a hires issue. Next time please try using hires when using comfyui.






