Upscale Image

The Upscale Image node increases the resolution of a single image using AI super-resolution engines (Topaz, AuraSR, Creative Upscaler, ESRGAN).

What does the Upscale Image node do?

The Upscale Image node increases the resolution of a single image using AI super-resolution. It dispatches to one of four engines — Topaz, AuraSR, Creative Upscaler, or ESRGAN — selected through the LLM model picker. Each engine exposes its own sub-models and parameters tuned for different content types (photos, AI art, anime, documents).

Common use cases:

• Enhance low-resolution product photos for e-commerce listings or marketplaces.
• Upscale AI-generated images to print-ready or retina-display resolution.
• Restore old, scanned, or heavily compressed photos with reconstructed detail.
• Sharpen documents, screenshots, and infographics before publication.

Quick setup

Follow these steps to add and configure the Upscale Image node in your workflow:

Configuration parameters

The node ships with one common section plus four engine-specific sections. Only the section matching the selected Model is honored at runtime — values in other sections are ignored.

Required fields

Optional fields — Common

Optional fields — Topaz

Professional-grade upscaler with 10 specialized sub-models split into two families.

Core sub-models: Standard V2, Low Resolution V2, CGI, High Fidelity V2, Text Refine. Generative sub-models: Recovery, Recovery V2, Redefine, Standard MAX, Wonder.

Some fields below are only honored by a subset of sub-models (visibility matrix lives in the node settings UI).

Optional fields — AuraSR

Optional fields — Creative Upscaler

Optional fields — ESRGAN

What does the node output?

The node outputs an array of upscaled images in the standard ImageGallery format used across the AI image nodes. Even though Upscale Image processes a single source image, the output is wrapped as an array so it plugs directly into nodes that expect an image gallery (Loop, Static Image, exporters).

[
  {
    "url": "https://cdn.fal.media/files/upscaled-image-...",
    "format": "jpeg"
  }
]

How to use the output

In Draft & Goal:

1. Draw a connection from the Upscale Image output port.
2. Connect it to the next node’s input.
3. In that next node, create and name your own variable (for example, upscaled_image). The image array is injected automatically.

Usage examples

Example 1: 2x photo upscale with Topaz High Fidelity V2

Use Topaz with the High Fidelity V2 sub-model at 2x to upscale product photos while preserving fine grain and texture. Enable Sharpen and Fix Compression for images sourced from the web or e-commerce catalogs.

Configuration:

• Model name: Topaz Upscale
• Upscale Factor: 2
• Topaz Sub-model: High Fidelity V2
• Sharpen: 0.3
• Fix Compression: 0.5
• Output Format: jpeg

Example 2: AI-art enhancement with Creative Upscaler (SDXL)

After generating an image with Text to Image, connect it to Upscale Image with the Creative Upscaler engine on the SDXL backbone. Use a moderate creativity and a strict negative prompt to keep the result clean.

Configuration:

• Model name: Creative Upscaler
• Upscale Factor: 4
• Architecture (Creative): SDXL
• Creativity (Creative): 0.5
• Detail (Creative): 1.5
• Guidance Scale: 7.5
• Inference Steps: 30
• Negative Prompt: blurry, artifacts, noise, lowres, jpeg

Example 3: Batch upscale inside a Loop with ESRGAN

Place Upscale Image inside a Loop to process a list of images. Use ESRGAN with RealESRGAN_x4plus for a good speed-quality balance across large batches; switch to RealESRGAN_x4_wdn_v3 if the source images are noisy.

Configuration:

• Model name: ESRGAN
• Upscale Factor: 4
• Variant (ESRGAN): RealESRGAN_x4plus
• Face Upscaling (ESRGAN): true
• Tile Size: 512 (lower this if you hit GPU memory limits)

Common issues

The connection from my image source is rejected

Cause: Upscale Image processes one image at a time. The source node is producing an array (Static Image with multiple URLs, File / Google Drive with “Multiple files” enabled, Google Drive Reader with multiple files).

Solution: Disable the “Multiple files” toggle on the source node, or wrap Upscale Image in a Loop node and feed it one image per iteration.

The node fails with requires a model to be selected

Cause: No engine has been picked in the Model name selector. The validator blocks the workflow before it reaches the runner.

Solution: Open the node settings and select Topaz, AuraSR, Creative Upscaler, or ESRGAN. Make sure your plan grants access to the chosen engine.

The output looks blurry or oversmoothed

Cause: The chosen sub-model or variant doesn’t suit the input. Topaz Core sub-models can flatten textures on AI art; ESRGAN base variants can struggle on noisy sources.

Solution: For Topaz, switch from a Core sub-model to a Generative one (Recovery V2, Redefine, Wonder). For ESRGAN, try RealESRGAN_x4_wdn_v3 for noise or RealESRGAN_x4plus_anime_6B for illustration. Bump up the Sharpen value if the engine supports it.

Faces are distorted or hallucinated

Cause: Generative sub-models can invent facial detail at high scale factors. Generic upscaling without a face pass leaves faces soft.

Solution: On Topaz Standard V2 or Recovery V2, enable Face Enhancement and tune the strength / sensitivity sliders. On ESRGAN, enable Face Upscaling. Reduce the upscale factor (try 2x instead of 4x) when faces still drift.

The job times out or runs out of memory

Cause: The source image is too large for the chosen engine, or the inference budget is exhausted (Creative Upscaler with very high Inference Steps).

Solution: Resize the source image before upscaling, lower the Upscale Factor, drop ESRGAN’s Tile Size to chunk the work, or reduce Creative Upscaler’s Inference Steps. Set Error Handling to Skip if the workflow must keep running on failures.

Best practices and pitfalls

How does it fit into a workflow?

Upscale Image typically sits between an image-producing node and an export or post-processing step.

graph LR
    Source[Static Image / File / Web Scraper] --> Upscale[Upscale Image]
    TextToImage[Text to Image] --> Upscale
    Upscale --> AspectRatio[Change Aspect Ratio]
    AspectRatio --> WP[WordPress Post Create]

Related nodes