Detail that was not in the file
Drag to compare. Every pair is one source image and its enlargement.
Catalogue still
A web-size product shot taken to 4×.
Portrait
Face enhancement on the advanced engine, held short of plastic.
Scanned artwork
Compression artefacts cleared as part of the enlargement, not after it.
Upload, pick a factor, compare the result
Upload the image
Drop a still into the app. It is uploaded to your workspace rather than fetched from a link, up to 200 MB.
Choose a factor and a model
1×, 2×, 4× or 8× on the simple engine. The advanced one adds a choice of ten Topaz models and shows only the controls that model takes.
Render and compare
Runs in roughly two to five minutes. The result step opens with the before and after behind a drag handle, and the enlargement lands in your library at full resolution.
What Upscale Image takes, and what it returns
What Upscale Image needs from you
- Input
- One image, uploaded to your workspace
- File size
- Up to 200 MB per upload
- Mode
- Image — the same app also runs on video
- Typical run time
- 2–5 minutes
Export specifications
- Simple tiers
- 2K 245 credits · 4K 490 · 8K 980
- Advanced tiers
- 24MP 245 credits · 48MP 490 · 96MP 980 · 512MP 4,165
- Comparison
- Before and after behind a drag handle on the result step
- Failed runs
- Refunded
- Delivery
- Rendered file in your workspace library
How far to enlarge, which model does it, and how much to fine-tune
The simple engine takes one decision. Advanced opens ten Topaz models, and the controls change with the model: pick one that does not take a setting and the setting is not rendered rather than quietly ignored.
-
Upscale factor
1×, 2×, 4× or 8× simple; 1× to 4× advanced
Multiplies both dimensions. 2× is the default; 8× is the simple app's ceiling, and the advanced engine reaches further by megapixel tier instead.
-
Model
Ten Topaz image models
Standard V2, Low Res V2, CGI, High Fidelity V2, Text Refine, Recovery, Recovery V2, Redefine, Standard MAX and Wonder.
-
Mode
Image or video
One app, two pipelines. Stills are priced by the output size they land in; simple video is a flat charge per job.
-
Face enhancement
On or off, with strength and creativity
You decide how much reconstruction a face receives rather than accepting a fixed amount.
-
Model-conditional controls
Rendered per model
Redefine adds creativity, texture, an auto-prompt and a written direction; Recovery V2 adds detail; Text Refine adds strength.
-
Restoration controls
Sharpen, denoise, compression repair, subject detection
The general cleanup pass on the advanced engine, applied alongside the enlargement rather than after it.
Engine choice
Two ways to run it. Pick the one that matches your source.
-
Standard
DefaultOne factor, sensible defaults, priced by the size the output lands in: 245 credits at 2K, 490 at 4K and 980 at 8K.
Most stills
-
Advanced
Fine controlTen Topaz models with per-parameter tuning, at factors of 1× to 4× and megapixel tiers of 245 credits at 24MP, 490 at 48MP, 980 at 96MP and 4,165 at 512MP.
Print and difficult sources
Built for images smaller than the surface they have to fill
-
Ecommerce teams
Take supplier images that arrived at thumbnail size up to the resolution a product page zoom needs.
-
Designers and print buyers
Push a still past its native pixel count for large-format output, using the advanced engine's megapixel tiers.
-
Photographers
Recover detail in older or heavily compressed frames rather than reshooting them.
Often used with
Built for Growth at Every Stage
Every plan includes every model and every feature. Plans only change how many credits you get and how many generations run at once.
- Credits refresh monthly
- Top-up additional credits anytime
- Unused credits don't roll over
Not ready for the commitment?
FAQ
Frequently asked questions
Everything you need to know about Upscale Image on BeHooked.
How large can an image get?
On the simple app, 8× is the top factor and 8K the top tier, at 980 credits. The advanced engine works at 1× to 4×, but its output tiers run 24, 48, 96 and 512 megapixels at 245, 490, 980 and 4,165 credits, so it is the one to reach for at print sizes.
Which model should I pick?
Ten are offered: Standard V2 and Standard MAX for general work, Low Res V2 for small sources, CGI for rendered images, High Fidelity V2 and Wonder for detail, Text Refine when lettering has to survive, Recovery and Recovery V2 for badly degraded frames, and Redefine when the model has to invent texture rather than restore it.
Why do the controls change when I change model?
Because each model takes a different set. Redefine adds creativity, texture, an auto-prompt and a written direction; Recovery V2 adds detail; Text Refine adds strength. A control the request would drop is not rendered at all.
Is there a grain control?
Not on stills. Grain is a video-only control on the advanced engine.
How long does an image take?
Roughly two to five minutes. Video runs on the same app take longer, five to ten minutes.
Will faces look artificial?
The advanced engine has a dedicated face enhancement control with its own strength and creativity settings, so you decide how much reconstruction is applied rather than accepting a fixed amount.
What happens if a run fails?
Failed runs are refunded.
