Quality Presets

Recaster provides six quality presets that configure the entire processing pipeline with a single click. Each preset is optimized for a specific use case.

Overview

Quality presets are the fastest way to configure the upscaling pipeline. When you select a preset, it automatically sets the model, scale factor, denoise strength, sharpen strength, multi-pass mode, and color correction to values optimized for that use case. You can always fine-tune individual settings after selecting a preset.

Recommended for most users

The Balanced preset works well for the majority of projects. It uses SwinIR 2x with multi-pass upscaling, light denoising, and medium sharpening to deliver high-quality results in a reasonable time. Start here and adjust individual settings if needed.

Preset Comparison

The table below shows the full parameter set for each preset:

Preset	Model	Multi-Pass	Denoise	Sharpen	Color Correct
Fast	Real-ESRGAN 2x	No	Off	Off	No
Balanced	SwinIR 2x	Yes (2x → 2x)	Light (2.0)	Medium (0.5)	Yes
Quality	SwinIR 2x	Yes (2x → 2x)	Heavy (3.0)	Strong (1.0)	Yes
Anime	SwinIR 2x	Yes (2x → 2x)	Light (1.0)	Off	Yes
Restoration	SwinIR 2x	Yes (2x → 2x)	Very Heavy (5.0)	Strong (1.5)	Yes
Temporal Studio	BasicVSR++ 4x	No	Light (2.0)	Medium (0.5)	Yes

Fast

The Fast preset is designed for speed above all else. It uses Real-ESRGAN 2x with no post-processing stages, making it the quickest way to get an upscaled result. Use this preset when you need a rough preview, when processing time is critical, or when the source material is already clean and does not need denoising or sharpening.

Best for

Quick previews and test runs
Clean source material that does not need denoising
macOS users processing locally (CoreML accelerated)
Batch processing where throughput matters

Balanced

The Balanced preset is the recommended default for most projects. It uses SwinIR 2x with multi-pass upscaling (two 2x passes for 4x total), light denoising at strength 2.0, medium sharpening at 0.5, and color correction enabled. This combination produces high-quality results without excessive processing time.

Best for

General-purpose upscaling with excellent results
Video content with typical compression artifacts
Projects where you want good quality without manual tuning
Social media and web delivery

Quality

The Quality preset maximizes visual fidelity at the expense of processing time. It uses the same SwinIR 2x multi-pass approach as Balanced but with heavier denoising (strength 3.0) and stronger sharpening (1.0). This produces the cleanest, sharpest output and is ideal for final delivery or professional work.

Best for

Final delivery and professional output
Film and broadcast quality requirements
Source material with moderate noise or grain
When processing time is not a constraint

Anime

The Anime preset is optimized for animated content. It uses light denoising (strength 1.0) to remove compression banding without blurring line art, and intentionally disables sharpening to preserve the clean, smooth lines that characterize animation. Color correction keeps the vibrant palette intact.

Best for

Anime and animated content
Cartoon-style artwork with flat colors
Content where preserving clean line art is critical
Upscaling older animation for modern displays

Restoration

The Restoration preset is built for recovering detail from degraded source material. It applies very heavy denoising (strength 5.0) to remove grain, noise, and scanning artifacts, followed by strong sharpening (1.5) to restore edge definition lost during denoising. This is the most aggressive processing configuration.

Best for

Old or degraded footage with heavy grain
VHS or film scan restoration
Heavily compressed source material
Archival and preservation projects

Denoising trade-off

Heavy denoising removes noise effectively but can also soften fine detail. The Restoration preset compensates with strong sharpening, but some texture detail may be lost. For source material that is only mildly degraded, consider using the Balanced or Quality preset instead.

Temporal Studio

The Temporal preset uses BasicVSR++ 4x, a video-specific model that maintains temporal consistency across frames. Unlike the other presets that process each frame independently, BasicVSR++ considers neighboring frames to produce smooth, flicker-free results. This preset is available exclusively on the Studio tier.

Best for

Video content where frame-to-frame consistency matters
Scenes with slow camera movement or static backgrounds
Reducing temporal flickering artifacts
Professional video production workflows

Customizing After Selection

Selecting a preset sets all pipeline parameters at once, but you can adjust any individual setting afterward. For example, you might select the Balanced preset and then increase the denoise strength from 2.0 to 3.0 for a particularly noisy source. Your customized settings persist for the current session.

For full control over every pipeline parameter, see the Pipeline Settings page.

Quick Selection Guide

Not sure which preset to start with? Use this decision tree:

Need results fast? → Use Fast
General project, good quality? → Use Balanced
Final delivery or professional output? → Use Quality
Animated or cartoon content? → Use Anime
Old, grainy, or damaged footage? → Use Restoration
Flickering between frames? → Use Temporal (Studio)

Was this page helpful?