Topaz for Web — 11/3 Release
Big one this month. A brand-new flagship image model, HDR for your photos, and a batch of quality-of-life improvements. Here’s everything new.
Wonder 4 — Our Smartest Image Model Yet
Wonder 3.5 set the bar for generative upscaling. Wonder 4 goes further — it’s an all-in-one model that upscales and repairs your image in a single pass.
Where older tools fix the whole photo the same way, Wonder 4 looks at your image the way a retoucher would. It runs on a new mixture-of-experts architecture — breaking your image into thousands of regions, with each region calling in only the enhancements it actually needs. Your blurry subject and your noisy background get completely different treatment, automatically.
In one pass, with zero settings, Wonder 4 combines:
- Upscaling
- Denoising
- Deblurring & Sharpening
- Compression-artifact removal
- Face recovery
The result is higher quality than Wonder 3.5 and more cohesive than chaining separate tools — with natural textures fully intact. Upload, enhance, done.
Wonder 4 is the new standard.
Wonder 4 Sharpen Example
Wonder 4 Denoising Example
Wonder 4 Deblur Example
Image HDR — Bring Your Photos to Life on HDR Displays
Image HDR automatically transforms standard (SDR) images into HDR — with brighter highlights, expanded dynamic range, and more lifelike contrast on HDR-capable monitors and mobile devices.
Using AI, it predicts how your photo would appear in HDR and generates a gain map to expand its dynamic range. This non-destructive approach preserves your original image while adding HDR information for greater brightness and tonal depth — automatically, with no parameters to adjust.
- Automatic SDR-to-HDR in a single step
- Non-destructive — adds HDR via an AI gain map without altering original content
- JPEG HDR output with embedded gain maps (ISO 21496-1), compatible with Google Ultra HDR and Apple Adaptive HDR
Note: HDR enhancements are only visible on HDR-capable displays. On standard SDR screens, the image will appear unchanged.
Quality-of-Life Improvements
- Uncapped batch downloads — TAR downloads no longer have a file-count limit; grab your whole batch at once
- JPEG stays JPEG — output now keeps your input’s format instead of defaulting to PNG
- Smoother progress — upload and batch progress are now more accurate, including large files and split renders
- Better ETAs — smarter estimates for queued work
- Canceled videos now show as canceled (not failed) in Recents and keep their actions
Learn More
For guides and support, visit docs.topazlabs.com and click Support in the bottom right.
Get started at app.topazlabs.com.
— The Topaz Team
