Photo editingCAN I VIBECODE
Imagen AI?
BUILD FIRST
- A non-sensitive or read-only version of the core experience
- A manual review step wherever correctness or trust matters
- A narrow companion tool that removes a real piece of work
LEAVE FOR LATER
- Regulated decisions, payments, or irreversible actions
- Claims of correctness, compliance, or uptime that the MVP cannot support
- Sensitive integrations until security, support, and auditability are ready
RESEARCH NOTES
What this tool actually does.
Imagen AI is a desktop app plus API platform for professional photographers. It learns a photographer's editing style from past Lightroom edits to build a personal AI Profile, then applies that style to full galleries. Core features include AI photo culling (rating, grouping, sorting by quality), automatic masking, color correction, cropping and perspective correction, horizon straightening, skin retouching, HDR merge, sky replacement, window pull, batch editing, and video color correction. It integrates with Adobe Lightroom catalogs and exports edited results as XMP sidecars or JPEGs. The platform also offers pre-built 'Talent' profiles from named photographers, a REST API with Python/Node/Go SDKs, a command-line tool, webhook callbacks, and scale-out processing for thousands of photos. Pricing is subscription-based with a free trial.
RESEARCHED BUILD PROMPT
Give this to your coding agent.
Build a focused replacement for Imagen AI's core post-production workflow for professional event and portrait photographers. The workflow: a photographer imports a shoot into Lightroom, creates an 'AI profile' from a set of manually edited reference photos (theirs or a talent style), then applies that profile to a full gallery to auto-edit, cull, and export. User-facing requirements: (1) a desktop or web app that connects to a Lightroom catalog or accepts a folder of RAW/JPEG images; (2) a 'Train Profile' flow where the user selects 20–50 edited reference images and the system learns tone, color, contrast, and crop preferences; (3) a 'Cull' step that rates/groups images by sharpness, composition, and exposure; (4) a 'Batch Edit' step that applies the profile with adjustable intensity, preview, and per-image overrides; (5) export as edited JPEGs plus XMP sidecars for Lightroom. Technical requirements: profile learning via image-pair modeling (edited vs. unedited), a per-image edit pipeline covering exposure, white balance, tone curve, hue/saturation/luminance, crop/aspect ratio, and noise reduction; a queueing system for batch jobs; REST API plus a Python CLI that accepts a folder, profile ID, and export settings; webhook or callback on job completion; storage for original/edited files and user profiles. Suggested implementation: backend in Python or Node.js with an SQLite/Postgres database, job queue using Redis or a task worker, and use a neural style-transfer or tone-mapping model trained per profile; if building without ML expertise, implement interpretable edit scalars (temperature, exposure, vibrance, etc.) derived statistically from reference edits. Do NOT copy Imagen AI's brand, visual design, website, pricing pages, or proprietary Talent marketplace. Out of scope: video editing, generative image synthesis, sky/HDR/skin micro-services (include only if trivial), mobile apps, cloud sync, and any real-estate-specific features. Ship a README covering setup, how to train a profile, how to run a batch edit via API/CLI, storage layout, and known limitations (e.g., profile drift, RAW handling, Lightroom integration via XMP only, no DAM).