# Topaz Image Upscale (https://docs.fuser.studio/docs/nodes/image/topaz-image-upscale)

_Provided by [Fal](https://fal.ai) — [Learn More](https://fal.ai/models/topaz/upscale/image/precision)._

## Inputs [#inputs]

- **Image** (`image`) — The image to upscale.

- **Mode** (`dropdown`) — Precision enlarges photos faithfully. Generative rebuilds detail in small or blurry sources. Creative reinvents detail in AI-generated images. Transparent preserves the alpha channel and always outputs PNG.

   Optional: `true` • Default: `precision` • Accepts: `Precision`, `Generative`, `Creative`, `Transparent`

- **Model** (`dropdown`) — Precision: Standard V2 fits most photos, High Fidelity V3/V2 preserve detail in professional shots, Low Resolution V2 recovers compressed sources, CGI targets art and renders, Text Refine keeps text crisp. Generative: Wonder 3.5 is the latest with finer detail, Wonder 3 is the most advanced realism model, Recover 3 rebuilds natural detail, Standard MAX is precision-leaning, Redefine adds prompt-guided detail, Recovery and Recovery V2 rebuild extreme low-resolution images. Creative: Bloom 2 is the most capable, Bloom transforms AI-generated images, Bloom Realism biases toward photorealism.

   Optional: `true` • Default: `Standard V2` • Accepts: `Standard V2`, `High Fidelity V3`, `High Fidelity V2`, `Low Resolution V2`, `CGI`, `Text Refine`, `Wonder 3.5`, `Wonder 3`, `Wonder 2`, `Wonder`, `Recover 3`, `Standard MAX`, `Redefine`, `Recovery V2`, `Recovery`, `Bloom 2`, `Bloom`, `Bloom Realism`

- **Scale Factor** (`slider`) — How much to enlarge the image. 2 doubles the width and height; 4 quadruples them.

   Optional: `true` • Default: `2` • Minimum: `1` • Maximum: `4`

- **Crop to Fill** (`toggle`) — Crop the result to fill the target dimensions instead of padding it.

   Optional: `true` • Default: `false`

- **Output Format** (`dropdown`) — File format of the upscaled image. Transparent mode always outputs PNG.

   Optional: `true` • Default: `png` • Accepts: `JPEG`, `PNG`

- **Subject Detection** (`dropdown`) — Which part of the image the enhancement targets: the whole image, the foreground subject, or the background.

   Optional: `true` • Default: `All` • Accepts: `All`, `Foreground`, `Background`

- **Face Enhancement** (`toggle`) — Detect and enhance faces in the image.

   Optional: `true` • Default: `true`

- **Face Creativity** (`slider`) — How much the face enhancement may invent detail. 0 stays faithful to the source; 1 is maximum creativity.

   Optional: `true` • Default: `0` • Minimum: `0` • Maximum: `1`

- **Face Strength** (`slider`) — How strongly faces are enhanced. 0 applies no enhancement; 1 is maximum strength.

   Optional: `true` • Default: `0.8` • Minimum: `0` • Maximum: `1`

- **Auto Adjust** (`toggle`) — Let Topaz analyze the image and pick sharpening, noise reduction, compression repair, and model strength automatically. Turn off to set them yourself.

   Optional: `true` • Default: `true`

- **Sharpen** (`slider`) — How much to sharpen edges and fine detail. 0 leaves the image as is; 1 is the strongest sharpening.

   Optional: `true` • Default: `0.5` • Minimum: `0` • Maximum: `1`

- **Denoise** (`slider`) — How much grain and noise to remove. 0 keeps all noise; 1 is the strongest noise reduction.

   Optional: `true` • Default: `0.5` • Minimum: `0` • Maximum: `1`

- **Fix Compression** (`slider`) — How aggressively to remove JPEG blocking and compression artifacts. 0 does nothing; 1 is the strongest repair. Not available for the CGI model.

   Optional: `true` • Default: `0.5` • Minimum: `0` • Maximum: `1`

- **Model Strength** (`slider`) — How strongly the Text Refine model reshapes text and edges. Higher values are crisper but can look unrealistic.

   Optional: `true` • Default: `0.5` • Minimum: `0.05` • Maximum: `1`

- **Enhancement Strength** (`dropdown`) — How aggressively Wonder 3 and Wonder 3.5 rebuild detail. Auto lets Topaz choose.

   Optional: `true` • Default: `auto` • Accepts: `Auto`, `Low`, `Medium`, `High`

- **Prompt** (`textarea`) — Optional text guiding the detail the Redefine model adds. Maximum 1024 characters.

   Optional: `true`

- **Auto Prompt** (`toggle`) — Let Topaz describe the image automatically to guide the added detail. Turn off to rely on your own prompt.

   Optional: `true` • Default: `true`

- **Creativity** (`slider`) — How much new detail the Redefine model may invent. 1 stays closest to the source; 6 takes the most liberties.

   Optional: `true` • Default: `3` • Minimum: `1` • Maximum: `6`

- **Texture** (`slider`) — How much surface texture the Redefine model adds. Topaz recommends 1 at low creativity and 3 at higher creativity.

   Optional: `true` • Default: `3` • Minimum: `1` • Maximum: `5`

- **Detail** (`slider`) — How much fine detail the Recovery V2 model rebuilds. 0 adds none; 1 adds the most.

   Optional: `true` • Default: `0.5` • Minimum: `0` • Maximum: `1`

- **Creativity** (`slider`) — How much the Bloom 2 model reinterprets the image. 1 is restrained enhancement; 9 is bold reinvention with newly generated detail.

   Optional: `true` • Default: `3` • Minimum: `1` • Maximum: `9`

- **Preserve Colors** (`toggle`) — Keep the source image's colors in the Bloom 2 result instead of letting the model reinterpret them.

   Optional: `true` • Default: `true`

## Outputs [#outputs]

- **Upscaled Image** (`image`) — The resulting upscaled image.