Adobe’s experimental Indigo camera app is getting a generative AI upgrade—and the way it’s being rolled out suggests Adobe is trying to thread a needle between two competing ideas: making photos look more “natural” and giving users tools that can fundamentally change what a photo contains.
Indigo first arrived as Project Indigo, an iPhone camera experiment aimed at producing a more SLR-like look. The pitch was familiar to anyone who has ever compared phone photography to a traditional camera: phones are fast and convenient, but their computational photography can sometimes feel too processed—too smooth, too sharpened, too “perfect.” Indigo’s promise was different. Instead of chasing a stylized aesthetic, it tried to make images feel closer to what people expect from a real camera: natural contrast, believable color, and a sense of depth that doesn’t scream “algorithm.”
Now, according to reporting from The Verge, Indigo is being updated with a suite of generative AI tools called “AI Playground.” This is not just another filter pack or a minor enhancement to existing editing controls. Generative AI implies a shift in capability: rather than only adjusting exposure, tone, or clarity, the app can create or transform content in ways that go beyond conventional image processing. In other words, the app is moving from “computational photography that looks like a camera” toward “computational photography that can invent.”
What makes this update especially notable is that Adobe is positioning it as an experiment, and it’s offering an opt-out path for users who don’t want the new behavior. Adobe also says the AI Playground suite isn’t dependent on its own Firefly AI models. And during an initial testing phase, Adobe is reportedly allowing free access to the suite without requiring sign-on, but only for a small percentage of Indigo users over the next few weeks.
That combination—experimentation, opt-out, and selective rollout—matters because it reveals how Adobe may be thinking about the next phase of consumer imaging. The company isn’t simply adding AI features and hoping users accept them. It’s trying to learn how people actually use these tools, how they react when the app can do more than enhance, and where the boundary is between “helpful” and “unsettling.”
A camera app that asks “what is a photo?”
The most interesting part of Indigo’s evolution isn’t the existence of generative AI. It’s the context. Indigo’s original identity was built around realism—making phone photos feel less synthetic. Generative AI, by contrast, is often associated with transformation: changing the scene, altering details, or producing results that may not correspond to what the camera captured.
That tension creates a question many creators are already debating: what counts as a photo? If a tool can reshape the image so thoroughly that the final output no longer matches the original capture, is it still a photograph—or is it closer to a digital artwork?
Adobe’s approach appears to acknowledge that discomfort rather than ignoring it. By providing an opt-out button, the company is effectively telling users: you can keep using Indigo in its earlier mode, and you can decide whether you want the generative layer. That’s a meaningful design choice in a market where many AI features arrive silently, embedded into default workflows.
Opt-out isn’t just a courtesy—it’s a signal
In consumer apps, opt-out options are often treated as a checkbox for compliance or user preference. Here, the opt-out seems to function as a product philosophy. Indigo is being updated, but Adobe is not forcing everyone into the same experience at once.
This matters because camera apps are deeply personal. People don’t just use them to edit; they use them to document. Even when users apply filters, there’s usually an implicit understanding that the image remains anchored to the original moment. Generative AI can blur that anchor. If the app can “fix” something by inventing plausible details, users may worry about authenticity—especially when images are shared publicly.
By letting users opt out, Adobe is giving them control over how much the app participates in the creative process. Some users will want the convenience of AI assistance. Others will want the app to behave like a camera pipeline: capture, enhance, and preserve the integrity of the scene.
The opt-out also gives Adobe a cleaner learning loop. If some users stay in the older mode while others try AI Playground, Adobe can compare engagement patterns, satisfaction, and retention. It can also observe whether users who opt out later return to the AI features—or whether they remain firmly committed to the non-generative workflow. That kind of behavioral data is far more valuable than surveys alone.
Why “not dependent on Firefly” is a bigger deal than it sounds
Adobe’s claim that AI Playground doesn’t rely on Firefly models is easy to gloss over, but it hints at something important: Adobe may be experimenting with different model providers, architectures, or inference strategies.
Firefly is Adobe’s well-known generative AI ecosystem, designed to integrate with Adobe’s broader creative tools. If Indigo’s AI Playground isn’t dependent on Firefly, then Indigo could be using a different set of models optimized specifically for mobile performance, latency, or particular editing tasks.
Mobile imaging has constraints that desktop creative suites don’t. A camera app needs responsiveness. Users expect near-instant feedback when they adjust settings or apply edits. Generative AI can be computationally expensive, so the app may be using models that are smaller, specialized, or deployed differently than Firefly’s typical workflows.
It’s also possible that Adobe is separating “consumer camera experiments” from “creative suite ecosystems.” In other words, Indigo might be treated as a sandbox for generative capabilities that can later inform broader product decisions—without necessarily tying the experiment to the same model stack used elsewhere.
The result is that Indigo’s AI Playground could represent a distinct technical track: generative tools designed for the camera context, not for the full creative suite context.
Free access without sign-on: a deliberate adoption strategy
Adobe is reportedly testing free access to AI Playground with a small percentage of Indigo users, and notably without requiring sign-on. That’s a strong adoption lever. Sign-in requirements can reduce participation, especially for experimental features. By removing friction, Adobe increases the likelihood that users will actually try the tools and provide feedback—whether through explicit ratings or implicit usage signals.
But there’s another reason this matters: if the feature is truly experimental, Adobe likely wants a wide range of real-world inputs. Camera apps see everything: different lighting conditions, skin tones, motion blur, low-light noise, and the messy unpredictability of everyday life. A generative AI tool that works well in controlled demos can fail in the wild. Free, low-friction access helps Adobe gather the messy data needed to improve reliability.
And because it’s limited to a small percentage of users, Adobe can manage risk. If the AI Playground produces occasional artifacts or unexpected transformations, the impact is contained. That’s crucial for a camera app, where trust is fragile. People will tolerate a lot from a filter app, but they expect a camera app to be dependable.
What “AI Playground” likely means in practice
The name “AI Playground” suggests a set of tools that encourage experimentation rather than strict, one-click “beautify” outcomes. In camera apps, generative AI can take several forms, and the label implies multiple options rather than a single feature.
Even without seeing the full list of tools, it’s reasonable to expect that AI Playground includes controls that let users explore transformations—possibly adjusting scenes, enhancing details, or changing elements in ways that feel more like interactive editing than traditional post-processing.
The key point is that generative AI changes the editing relationship. Traditional camera enhancements are constrained by the original pixels: denoise, sharpen, adjust color, correct exposure, and so on. Generative tools can go further by reinterpreting parts of the image. That can be powerful—especially for fixing issues like harsh lighting, distracting backgrounds, or imperfect focus—but it can also introduce uncertainty. Users may wonder which parts are “real” and which parts are “suggested.”
This is why the opt-out matters again. If users can choose whether to engage with generative behavior, they can decide how much they want the app to act as a creative collaborator versus a faithful imaging pipeline.
The “natural look” goal meets generative reality
Indigo’s original mission was to deliver a more natural, SLR-like look. That’s a subtle goal, and it’s not just about aesthetics. It’s about how images feel when you look at them for longer than a thumbnail.
SLR-like imagery often carries cues that computational pipelines struggle to replicate: gentle highlight roll-off, realistic texture, and a depth-of-field impression that doesn’t look like a software effect. Phone cameras have improved dramatically, but the “natural look” promise still resonates because many users want their photos to feel like they came from a device with a different optical and processing philosophy.
Generative AI could either help or undermine that mission. On one hand, generative tools can potentially improve realism by reconstructing details that were lost to noise or blur. They can also help unify color and texture so the image looks coherent rather than artificially enhanced.
On the other hand, generative AI can produce a kind of realism that’s convincing but not necessarily accurate. It can fill in missing information with plausible guesses. That might look great, but it can also lead to subtle inconsistencies—textures that don’t match the scene, edges that don’t align with physical reality, or details that appear “too perfect” in a way that feels uncanny.
Adobe’s decision to call it an experiment suggests the company is aware of these risks. The opt-out and limited rollout imply that Adobe expects iteration—tuning the tools, refining guardrails, and learning from user behavior.
A unique take: Adobe is treating the camera as a dialogue, not a pipeline
There’s a broader shift happening across consumer imaging: apps are moving from “capture and process” toward “capture and negotiate.” The camera becomes a system that interprets your intent. It decides what matters, what to enhance, and what