Choosing the Right AI Preset: Photo vs. Art vs. Anime vs. Text

Why Different Presets Exist

Every AI upscaling model encodes assumptions about what high-resolution content looks like. These assumptions come from the training data — the model learns by example. A model trained on millions of photographs learns that high-frequency detail looks like skin texture, fabric weave, and leaf edges. When it encounters a cartoon with solid color fills and crisp outlines, it doesn't know how to handle it and tries to add photographic texture to the flat areas, producing a muddy result.

Content-specific presets solve this by training separate model checkpoints on domain-appropriate data. The Photo preset has never seen anime. The Anime preset has never been trained to generate photographic grain. Each model is optimized for exactly what you're going to feed it.

Photo Preset: Real-World Photography

Use this for: portraits, landscapes, product photography, real estate images, event photos, family photos, wildlife photography.

The Photo preset is trained on high-resolution photographic images and their degraded counterparts. It excels at:

• Skin texture reconstruction — pores, hair strands, and the subtle variation of real skin
• Preserving natural color gradients in skies, skin tones, and surfaces
• Removing camera noise while maintaining sharpness
• Recovering detail from compressed JPEG files

This is the correct choice for roughly 70% of upscaling tasks. When in doubt, start here for anything that came from a camera.

Art & Illustrations Preset: Creative and Digital Work

Use this for: digital paintings, concept art, graphic design assets, illustrations, book covers, poster art, scanned traditional paintings, logos.

The Art preset is trained to respect artistic intent. It preserves:

• Brush stroke direction and texture
• Clean vector-like edges in digital designs
• Color saturation and vibrance without oversaturation
• Flat color regions in graphic design work

Importantly, this preset does not add photographic texture to artwork. A smooth gradient in a painting comes back as a smooth gradient, not as a noise-textured surface.

Anime & Cartoons Preset: Animated Content

Use this for: anime screenshots, manga panels, Western cartoons, game sprites, character art, cel-shaded 3D renders, fan art.

Anime and cartoon content has specific characteristics: clean line art, flat color fills, subtle shading gradients, and a stylized look that should not look photographic. The Anime preset is trained specifically on this type of content and:

• Keeps line art crisp and clean without adding grain
• Maintains flat color regions instead of adding texture
• Respects cel-shading gradients
• Reduces the blocky compression artifacts common in low-bitrate anime video

Text & Documents Preset: Legibility First

Use this for: scanned documents, screenshots with text, presentation slides, UI mockups, infographics, invoices, charts, whiteboards.

This preset is optimized for a single priority: maximum text sharpness. It:

• Maximizes edge contrast around letterforms
• Removes blur from scanned documents
• Reduces JPEG compression artifacts around text edges
• Maintains high contrast for OCR preprocessing

Do not use this preset for photographs or artwork — it will over-sharpen natural content and produce an artificial, over-processed look.