Nero AI
AIUpscale, restore, edit backgrounds, and animate images with Nero AI. Agents prepare visual assets for listings and campaigns without a manual editing pass.
What This Integration Enables
Nothing in this connector creates an image from nothing. Every operation starts from a file that already exists and is damaged, and that is the whole argument for it. Upscale Image enlarges by 2, 4 or 8 times with the model chosen for the subject matter, so a landscape, a camera photo, a piece of artwork, a blurry face, a fur-and-eyes close-up and a high-resolution portrait each get a different treatment rather than one generic enlargement. Unblur Image reverses motion blur. Fix Scratches detects and removes scratches, creases and surface marks from scanned photographs, and returns the mask it used alongside the repaired file. Restore Photo bundles scratch repair, colorization, face enhancement and denoising into a single pass for scans that need several kinds of recovery at once.
Beyond restoration, the connector covers the ordinary asset work that sits next to it. Change Background composites a cut-out subject onto a replacement image. Generate Background produces up to four new backdrops from a description, though it requires an alpha-channel cutout as input, so Remove Background runs first. Compress Image reduces file size and reports the compression ratio with input and output byte counts, which is the sort of number a page performance workflow can act on. Generate Video From Image animates a still into a short clip. Every one of these is asynchronous: the submit operation returns a task identifier and Get Task Result is polled until the status reports done. Video generation takes noticeably longer than image work, so the polling interval should be set accordingly rather than left at the image default.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Digitizing a corporate photo archive
A company preparing an anniversary retrospective has boxes of prints scanned but not restored. The agent reads each scan, classifies the dominant defect, and routes accordingly: heavy creasing and surface damage to Fix Scratches, soft focus to Unblur Image, low-resolution group shots to Upscale Image with the model tuned for high-resolution portraits and text, and the worst cases to Restore Photo for a combined pass. Finished files land in the asset record in Notion next to the untouched scan. The archivist never opens an editor. They open a review queue.
Legacy catalog assets brought up to current standards
A retailer has thousands of product images shot years ago at resolutions that no longer meet the storefront's requirements. The agent runs Upscale Image with the Reconstruct model, which is tuned for recovering fine detail in high-resolution portraits, text and product shots, then Denoise Image at a moderate strength to clear compression artifacts left by an earlier round of processing, then Compress Image so the finished file meets the page weight budget. The compression result reports the ratio and the input and output byte sizes, so the flow can hold back any file that came in over budget and flag it as an automation exception rather than publishing something that slows the page.
Seasonal variants from one product cutout
A campaign needs the same product shown against four seasonal backdrops. The agent calls Remove Background once to produce the alpha-channel cutout, then Generate Background with a description per season, which returns up to four variations in a single call. Where the brand already owns approved backdrop photography, Change Background composites onto those files instead of generating anything. The variants go to the campaign record for a designer to choose from, and the original product photograph is never modified.
Human-in-Loop Highlight
Restore Face improves facial clarity and recovers missing features such as eyes and contours in low-quality portraits. Read that carefully. When a scan is degraded enough that the eyes are not in the file, the model is not sharpening a face, it is proposing one. In an archive, the restored file becomes the record copy: the version that goes in the retrospective, the version the family receives, the version everyone downstream treats as what the person looked like. So the agent runs the whole batch, files the surface repairs on its own, and stops on every scan where Restore Face or the face-enhancement stage of Restore Photo actually ran. The message carries the original and the restoration side by side and asks a specific question: "Scan 0042, staff photograph 1974. The eyes and jawline in this restoration were reconstructed, not recovered. Does this read as a plausible likeness to someone who knew them? File as the record copy, keep the unrestored scan as the record copy, or discard." The original scan is retained either way. The agent moved 400 images. It stopped on the eleven where a person's face was at stake, which is what a digital andon cord is for.
Agent Capabilities
15 actionsEnhancement
5- Upscale Image Submits an upscaling task that enlarges an image by 2, 4 or 8 times using a model tuned for the subject matter. Standard suits landscapes and snapshots, Photograph gives softer results for camera photos, Anime And Illustration suits artwork, Face Enhancement restores blurry faces, Fine Detail sharpens eyes and fur, and Reconstruct recovers fine detail in high-resolution portraits, text and product shots.
- Unblur Image Submits a deblurring task that reverses motion blur and restores focus, producing a crisp version of a blurry photo.
- Sharpen Image Submits a sharpening task that improves clarity and detail. Sharp maximises detail visibility, Sharp V2 produces the crispest results, Balanced suits everyday photos, Gentle stays closest to the original, and Anime And Illustration is tuned for artwork.
- Denoise Image Submits a noise removal task that clears visual noise and compression artefacts, with an adjustable strength so grain can be softened rather than removed entirely.
- Compress Image Submits a compression task that reduces file size while preserving visual quality. High compression produces the smallest files with some possible quality loss, and moderate compression keeps more detail at a larger size. The completed result reports the compression ratio with input and output byte sizes.
Restoration
4- Fix Scratches Submits a scratch repair task that detects and removes scratches, creases and surface marks from scanned or damaged photos. A hand-drawn mask can target specific areas, and the result includes both the repaired image and the mask that was used.
- Restore Face Submits a face restoration task that improves facial clarity and recovers missing features such as eyes and contours in low-quality portraits. Because the recovered features are proposed rather than read from the file, this is the operation that belongs behind a human decision on any image of a real, identifiable person.
- Colorize Photo Submits a colourisation task that adds natural colour to a black and white or sepia photograph.
- Restore Photo Submits an advanced restoration task combining scratch repair, colourisation, face enhancement and denoising in a single pass. Used for old or badly damaged photographs that need several kinds of recovery at once.
Backgrounds
3- Remove Background Submits a background removal task that cuts the subject out of a photo and returns it on a transparent background. Also the required first step before Generate Background, which needs an alpha channel.
- Change Background Submits a task that removes the original background and composites the subject onto a replacement background image you supply. Used when the brand already owns approved backdrop photography.
- Generate Background Submits a task that generates new backgrounds behind a cut-out subject from a text description, producing up to four variations in one call. The input must already be a foreground cutout with an alpha channel.
Creative
2- Apply Image Style Submits a style transfer task that recreates a photo in a chosen artistic style such as cartoon, comic or 3D figure, optionally producing several variations and an HD result. Style identifiers come from the live Nero AI style catalogue.
- Generate Video From Image Submits a task that animates a still image into a short video, optionally guided by a text description of the motion. Video generation takes noticeably longer than image tasks, so set the polling interval accordingly.
Tasks
1- Get Task Result Retrieves the status and result of a previously submitted task. Every Nero AI operation is asynchronous, so poll this until the status reports done, at which point the result carries the output URL, or a list of output URLs for operations that produce several images.
Frequently Asked Questions
What can FlowRunner do with Nero AI?
FlowRunner agents can run Upscale Image, Unblur Image, and Sharpen Image in Nero AI, plus 12 more actions.
Does connecting Nero AI to FlowRunner require OAuth?
No. Nero AI connects to FlowRunner with an API key, no OAuth flow required.
Can Nero AI trigger a FlowRunner workflow automatically?
Nero AI doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with Nero AI
$100 in credits. No card required. Connect in minutes.