AI photo editing software

Virtual staging and AI photo editing software, built for your platform

Want virtual staging inside your portal, an AI photo editor for your customers or product photography at scale? We built and run exactly this for real estate and e-commerce, with a style that stays consistent across a whole listing or catalogue.

  • Declutter, stage and restyle while walls, windows and camera angle stay put
  • One house style across every photo of a property or product line
  • Visible AI disclosure and metadata on every export
Twelve Lumens editor: a living room photo split into the original and a Scandinavian-style render, with version history and an edit prompt
What we can build

What we can build for your photos

As a standalone product, a white-label tool for your customers or a feature inside your existing platform.

Virtual staging

Furnish empty rooms in preset or custom styles. The architecture is locked in the prompt, so the room stays the room and a kitchen stays a kitchen.

Declutter and clean-up

Empty a lived-in room, or tidy exteriors by removing cars, bins and clutter while neighbouring houses stay untouched.

Brand and house styles

Create a style from a description and reference photos, turn it into a moodboard and apply it consistently to every photo in a listing or catalogue.

Batch processing

Upload a whole shoot. Photos are sorted by room or product type automatically and rendered in one go, with live progress.

Editor and version history

Before and after comparison, edits by prompt, sliders and locks for walls, floors and windows, and a version timeline that never loses the original.

Exports, labels and showcases

Portal-ready crops in any aspect ratio, disclosure labels in five languages, IPTC metadata, and shareable showcase pages with a before and after slider.

Proof, not promises

Our own staging and product-photo engine

Twelve Lumens is our AI virtual staging product for estate agents, photographers and homeowners. Upload a full shoot and it sorts the photos into 21 room types, declutters them and stages every room in one house style, in minutes.

Salestreak runs the same engine on products: a phone snapshot becomes a clean packshot or a styled scene, with captions and alt text written in five languages.

  • A vision pass checks every upload first: room type, how furnished it is, collages and existing AI labels
  • Six preset styles plus custom styles built from your own reference photos
  • 1K and 2K output in nine aspect ratios, including the formats property portals expect
  • A disclosure bar and IPTC metadata on every render, so buyers always know what's AI
  • Originals are always kept; every render is a new version
The same room staged in a warm luxury style, with beams, windows and mantel unchanged
A cluttered canal-house room with moving boxes and stacks of books
Before After
Declutter and stage in one pass. Example images, generated for illustration.
One empty living room in six versions: empty, mid-century, Scandinavian, luxury, industrial and coastal
One empty room, five styles. Architecture and camera angle stay identical.
A mug photographed on a cluttered counter, the same mug as a clean packshot, and in a styled breakfast scene
The same engine for products: phone photo, packshot, styled scene (Salestreak).
room types recognised automatically
21
preset styles, plus custom styles
6
aspect ratios, in 1K or 2K
9
languages for labels and descriptions
5
Who it's for

Who we build this for

Property portals and proptech

Offer virtual staging inside your platform, under your brand, with usage limits per subscription and labels that match your listing rules.

Estate agencies and franchises

One house style across every listing and every branch, with branded showcase pages to send to buyers.

Photographers and media studios

Batch pipelines per client, fixed styles per agency and exports in exactly the formats your clients upload.

E-commerce brands and marketplaces

Packshots and lifestyle scenes generated from simple product photos, consistent across your catalogue.

How it works

How the pipeline works

The quality comes from what happens before and after the image model, not just from the model itself.

  1. 1

    Ingest

    Uploads are converted (HEIC included), stripped of location and device metadata and stored privately.

  2. 2

    Analyse

    A vision model labels each photo: room or product type, how full it is, collages, existing AI labels and a description per language.

  3. 3

    Build the prompt

    Room rules, architecture locks, the chosen style, sliders and the user's own guidance are combined in a fixed order, so results stay predictable.

  4. 4

    Render in the background

    Jobs run in a queue with live progress in the interface. Stalled jobs are detected and their usage refunded.

  5. 5

    Review and refine

    Compare before and after, edit by prompt, switch styles or roll back. Every change is a new version.

  6. 6

    Export and share

    Crops per aspect ratio, labels and metadata applied, a ZIP to download or a showcase page for buyers.

Image tooling we use

Image models from OpenAI and others for rendering, a vision model for analysis, libvips and exiftool for processing and labels, and Cloudflare for private storage and fast delivery.

  • OpenAI image models
  • Vision models
  • libvips
  • exiftool
  • Cloudflare R2
  • Cloudflare Workers
  • Ruby on Rails
  • Hotwire
  • Solid Queue
  • PostgreSQL
Cost & timeline

What does AI photo editing software cost?

The build and the running costs both depend on a handful of choices. These matter most:

Edit types
Staging, decluttering, restyling, sketch to photo, product scenes: each needs its own prompt rules and testing.
Consistency requirements
One-off images are easier than a fixed house style across hundreds of photos. Consistency is where the pipeline work goes.
Volume and resolution
Higher resolutions and bigger batches raise the cost per image and the need for queues and limits.
Integration
A standalone app, a white-label tool or an API inside your platform: the integration shapes the scope.
Labelling and compliance
Disclosure bars, metadata, custom disclaimer templates and translations for each market you serve.
FAQ

Questions about AI photo editing and virtual staging software

Can you build a white-label virtual staging tool?

Yes. We built Twelve Lumens, our own virtual staging product, and can build the same capabilities under your brand: your styles, your plans and your disclosure rules, as a standalone app or inside your platform.

How much does it cost to build AI photo editing software?

It depends on the edit types, how consistent results must be, the volume and how it integrates with your platform. After a free intro call you get a written scope and estimate, including the expected running cost per image.

How do you keep the architecture of a room unchanged?

With a fixed prompt structure: hard rules that walls, windows, built-ins and camera angle stay as photographed, rules per room type, and locks the user can switch on or off. A vision pass checks the photo before anything is rendered.

Do virtually staged photos need a label?

Often, yes. Many portals and agencies expect virtual staging to be disclosed, and the EU AI Act asks for AI-generated images to be recognisable as such. We add a visible disclosure and IPTC metadata to every export.

Can it process a whole photo shoot at once?

Yes. A full shoot is uploaded in one go, sorted by room automatically and rendered as a batch, with live progress and a ZIP export organised by style, format and room.

Does it work for product photos too?

Yes. The same pipeline turns a simple product photo into a clean packshot or a styled lifestyle scene, as we do in Salestreak.

Which image models do you use?

Currently OpenAI's image models for rendering and a vision model for analysis. We've also worked with Flux, Seedream and Google's Nano Banana models, and choose per project on quality, speed and cost.

Want AI photo editing in your product?

Tell us what kind of photos, which edits and at what volume. You'll get an honest view on quality, running costs and how to integrate it.

martin@balk.dev

Typically a reply within one business day.

Useful to mention

  • What you want to build, in a few sentences
  • Who is going to use it
  • Any deadline or launch date
  • Tools and systems it should connect to