Adobe's Project Indigo — the experimental iPhone camera app launched to deliver a more natural, SLR-like look — is being updated with a suite of generative AI tools that don't use Adobe's own Firefly models, a notable departure from the company's usual AI stack.
Key takeaways
Adobe's Project Indigo camera app is gaining generative AI features including background removal, object erasure, and style transfer for iPhone photographers.
None of the new generative AI tools in Indigo use Adobe Firefly — Adobe is drawing on third-party models for this experimental playground.
A new AI critique feature analyzes your photos and provides compositional and technical feedback directly inside the app.
Indigo was originally launched to give iPhone photos a more natural, SLR-like rendering — the AI additions represent a significant pivot for the project.
The update positions Indigo as a testing ground for AI-assisted photography workflows outside Adobe's flagship Creative Cloud suite.
What the Indigo update actually adds
The new feature set, which Adobe is calling an "AI Playground," includes background removal, object removal, and style-transfer effects that can restyle a photo as a sketch or painting — and then, in a striking reversal, convert illustrations back into photorealistic images. According to The Verge, Adobe describes the tools as an experimental space rather than a polished product release, which explains the choice to reach outside Firefly rather than wait for those capabilities to mature inside Adobe's own model family.
Before and after shots showing Adobe Indigo's new AI object removal in action.
For anyone already using AI image tools, the object and background removal features are familiar territory — but the style-transfer round-trip is more interesting. Being able to push a photograph toward a painted or sketched look, then pull an illustration back toward photorealism, opens up a quick iteration loop that normally requires jumping between separate tools.
Adobe Indigo's style transfer can convert painted illustrations into photorealistic images.
TechCrunch reports that Indigo can now analyze a photo you've just taken and return written feedback on composition, lighting, and technical quality. That's a different kind of AI assistance than generation — it's closer to having a second opinion baked into the capture workflow. For photographers learning to compose shots, or creators who shoot reference images for AI generation work, immediate in-app critique cuts out the loop of exporting and asking a separate tool.
Adobe Indigo's new AI critique feature analyzes composition and technical quality after capture.
Adobe has been careful, in its commercial products, to position Firefly as a commercially safe model trained on licensed content. Indigo's AI Playground sidesteps that entirely, which tells you something about where Adobe sees the project sitting — it's a sandbox for capability exploration, not a product with IP-clearance guarantees. Creators who use AI-generated images commercially should treat outputs from Indigo's generative features with the same scrutiny they'd apply to any third-party model whose training data isn't fully disclosed.
For AI-art creators who shoot reference photography to feed into generation pipelines — using real photos as style anchors, ControlNet inputs, or img2img sources — a capture-to-edit app with built-in background removal and style transfer has obvious utility. The question is whether Indigo's image quality holds up as a source asset. Adobe hasn't released the underlying model names for the generative features, so there's no way yet to benchmark output quality against dedicated tools.
If you want to experiment with the style-transfer and generation features in your own workflow, the Charmloop image generator and model catalog offer a range of styles worth comparing against what Indigo produces from a real photograph.
Indigo remains an experimental app — Adobe hasn't committed to a timeline for bringing any of these features into Lightroom or Photoshop. But the fact that Adobe is testing non-Firefly generative tools in a shipping app, even a small one, suggests the company is watching closely to see which capabilities creators actually reach for before deciding what graduates to the flagship suite.
