Real estate
Imagen eliminates hours of busywork from real estate post-production. Photographers upload listing photos and receive fully edited JPEGs back, complete with sky replacement, perspective correction, window pull, and an automatically applied HDR merge. The real estate API supports two editing modes:
- Profile-based editing: Apply a trained AI Profile to your photos. Choose one of Imagen’s built-in presets or your own Personal or Talent AI Profile.
- Smart Editing: Use Imagen’s image-to-image model. No AI profile required.
Both modes share the same setup: sign-up, API key, and supported file formats. They use different endpoint namespaces. The setup is covered in Onboarding, including the supported file formats.
Choose an editing mode
| Profile-based editing | Smart Editing (I2I) | |
|---|---|---|
| Endpoints | /v1/projects/... | /v1/i2i/projects/... |
| Needs an AI profile | Yes. Choose one of Imagen’s built-in presets or your own profile. | No. It uses Imagen’s image-to-image model. |
| Edit body | Full tool set, including profile_key, photography_type, and export. | Includes hdr_merge, perspective_correction, sky_replacement, and callback_url. |
| Upload | Single PUT per file. | Single PUT per file, or multipart upload for large files. |
| Status polling | Two steps: poll /edit/status, then /export/status. | One step: poll the project. |
| Output | Full-resolution edited JPEGs, exported automatically. | Edited JPEGs. No separate export step. |
Use profile-based editing when you want consistent, full-resolution results that match a preset or your own profile. Use Smart Editing when you want Imagen to detect the best editing style for your photos. Smart Editing uses the image-to-image model, which has no AI Profile and a simpler edit body.
Profile-based editing
The profile-based flow uses the same /v1/projects/... endpoints as the
Quickstart: create a project, request upload links, upload,
edit, poll status, then download. This guide covers what’s specific to real
estate. For the create, upload, poll, and download mechanics, follow
Quickstart.
Built-in presets
For real estate editing, choose exactly one of two categories for profile_key:
a built-in preset, or a Personal or Talent AI Profile from the Imagen desktop app.
To list your own profiles, see Quickstart.
The three built-in presets:
| Preset | JPEG profile_key | RAW profile_key |
|---|---|---|
| Elegant Home | 178014 | 178011 |
| Modern Home | 472033 | JPEG uploads only |
| Natural Home | 333757 | 333755 |
Pick the JPEG variant when all uploaded images are JPEGs. Pick the RAW variant when any uploaded image is a RAW. Modern Home works with JPEG uploads only. Use Elegant Home or Natural Home for RAW shoots.
The edit body for real estate
For real estate, explicitly send every tool flag. Three fields must be set to fixed values for the flow to work end-to-end:
photography_type: Set to"REAL_ESTATE".window_pull: Set totrueto apply window pull. It’s optional and defaults tofalse, so include it explicitly for the real estate flow.export: Set totrue. This schedules a JPEG export automatically after the edit completes.
curl -X POST 'https://api.imagen-ai.com/v1/projects/$PROJECT_UUID/edit' \
--header 'x-api-key: $IMAGEN_API_KEY' \
--header 'Content-Type: application/json' \
--data '{
"profile_key": 333757,
"hdr_merge": true,
"perspective_correction": true,
"sky_replacement": true,
"window_pull": true,
"photography_type": "REAL_ESTATE",
"export": true,
"crop": false,
"straighten": false,
"portrait_crop": false,
"headshot_crop": false,
"subject_mask": false,
"smooth_skin": false
}'The flags crop, straighten, portrait_crop, headshot_crop, subject_mask,
and smooth_skin are part of the broader Imagen API but the real estate flow does
not use them. Send them as false to match the web app, or omit them. The server treats unset tool flags as false. The cURL example above sends them explicitly. The other languages omit them.
When sky_replacement=true, pass
sky_replacement_template_id to choose a specific sky.
List the
available templates with
GET /v1/projects/sky_replacement/templates.
Omit sky_replacement_template_id to apply your
account’s default template.
When sky_replacement=false, make sure to omit sky_replacement_template_id. See
Quickstart for more information.
In the real estate flow, the export field must be set as export: true.
What each tool does
| Field | Effect on a real estate shoot |
|---|---|
hdr_merge | Merges bracketed exposures into a single DNG. Set to true for bracketed shoots. |
perspective_correction | Straightens vertical and horizontal lines in walls and doorways. Recommended for all interiors. |
window_pull | Pulls highlight detail through interior windows so the exterior view is visible. Optional (defaults to false); set to true to apply it. |
sky_replacement | Replaces blown-out sky with your account’s default sky template. Use it for exteriors. See the callout above for choosing or omitting a sky replacement template. |
For full per-tool detail across all genres, see the AI tools reference.
Status and timing
Because export=true schedules an export after the edit, the profile-based flow
has two statuses to wait on: poll /edit/status until Completed, then poll
/export/status until Completed. The export status endpoint returns an error
until the export has been scheduled (which happens automatically once the edit
completes), so poll edit status first, then switch to export status. The polling
mechanics and code samples are documented in Quickstart.
You can skip polling entirely by passing a callback_url on /edit. See
Callbacks.
A 42-image project with 3-frame brackets takes 30-40 minutes end-to-end. 6-frame brackets roughly double that.
Smart Editing alternative
If you’d rather skip profiles and use the image-to-image model, the
Smart Editing flow takes a minimal edit
body and returns JPEGs as soon as the project reports Completed. There’s no separate export step. HDR-merged output is downsized in Smart Editing. Use the
profile-based flow above for full-resolution HDR.
Refine with Copilot
After a project reaches Completed in either mode, you can apply prompt-based
enhancements to individual photos with the Copilot API. It’s the same capability as the AI Assistant in the real estate web app.