| Setting | Value | Purpose | |--------|-------|---------| | | Auto | Works for 90% of images. | | Overall NR | 20–40 | Lower = retain grain for aesthetic. | | Sharpen | 30–50 | Don’t overdo – do sharpening later in workflow. | | Recover Detail | 40–70 | Essential – stops plastic look. | | Color NR | 60–80 | Separate from luminance NR. |
Here’s a deep write-up on , focusing primarily on Topaz Photo AI (their current flagship) and noting key differences for Gigapixel AI and DeNoise AI where relevant. Understanding Topaz’s Core Philosophy Topaz software uses adaptive AI models – not static sliders. This means the “best” settings depend heavily on your source image (noise level, subject matter, resolution goal). The goal isn’t to max every slider, but to balance detail recovery against artifacts . 1. Topaz Photo AI – Recommended Settings by Scenario A. Noise Reduction (Standard/RAW images) | Setting | Recommended Value | Notes | |--------|------------------|-------| | Model | Standard (v2) | Best for general photos. Low Light for extreme noise, Severe Noise for very high ISO (>6400). | | Remove Noise | 30–55 | Start at 40. Increase if grain is distracting; decrease if losing fine texture (skin, fur). | | Recover Original Detail | 60–80 | Critical – restores texture erased by NR. Lower = smoother, higher = grainier. | | Color Noise Reduction | 60–80 | Default 70 works for most. Bump to 90+ for high ISO in shadows. | Pro tip: Enable “Preview Differences” (split view) to see exactly what NR removes vs. detail recovery adds. B. Face Recovery | Setting | When to Use | |--------|-------------| | Face Recovery Strength | 40–60 for portraits (natural), 80–100 for very small faces or old low-res photos. | | Face Refine | Check if eyes/mouth look AI-generated or waxy. | Warning: Do not use Face Recovery on non-face subjects (statues, paintings, animals) – it invents fake features. C. Sharpening (Standard) | Setting | Value | Explanation | |--------|-------|-------------| | Model | Standard | For natural subjects. Lens Blur for motion/defocus, Out of Focus for heavy blur. | | Sharpen Strength | 35–55 | Start 45. Below 30 is subtle, above 70 can cause halos. | | Remove Blur | 40–70 | Higher = more aggressive edge sharpening. Tune based on visible softness. | | Recover Fine Details | 50–70 | Prevents oversharpening of grain/noise. Lower if image becomes crunchy. | D. Upscaling (Resolution Increase) | Factor | Model | Additional Settings | |--------|-------|----------------------| | 2x | Standard v2 | Best for general photos. | | 4x | High Fidelity | Slower but preserves textures like fabric/foliage. | | 6x+ | Low Resolution v2 | Designed for tiny images (< 0.5 MP). Use Auto settings first. | Key setting: Keep Remove Blur under 40 when upscaling more than 2x – over-sharpening amplifies artifacts. 2. Topaz Gigapixel AI – Standalone Upscaling Settings | Scenario | Model | Settings | |----------|-------|----------| | Faces / portraits | Faces | Strength 60–80, Suppress Noise 40 | | Architecture / lines | Art & CG | Remove Blur 30, Suppress Noise 20 | | Natural landscapes | Nature | Detail Recovery 70, Noise Reduction 50 | | Heavy JPEG artifacts | Compressed | Artifact Removal 80, then use Standard model as second pass | Golden rule: Never use “Very Compressed” model unless image is under 300 KB – it makes clean images waxy. 3. Topaz DeNoise AI – Legacy but Still Useful DeNoise AI gives more manual control than Photo AI’s NR. topaz settings for edits