Adobe Firefly
The next evolution of creative AI is here for all your ideas, with image, video, audio and vector tools.
JUMP TO SECTION
How to create AI product photography with Adobe Firefly.
Why AI product photography matters now.
The full Firefly product photography workflow.
How to write product photography prompts that work.
How to keep AI product photos visually consistent.
Where AI product photography works – and where it still doesn’t.
How to create AI product photography with Adobe Firefly.
AI product photography uses generative AI to transform a basic product image into a polished, on-brand campaign visual without a studio shoot, professional camera, or photographer. A single product photo, a written prompt, and a few minutes inside Adobe Firefly are now enough to produce a full campaign’s worth of imagery.
The category has matured fast. Two years ago, AI-generated product photos looked noticeably artificial with strange shadows, wrong reflections or products that didn’t quite match the original. Today, outputs from models such as Firefly Image 5 close enough to studio work that the person browsing your store is more likely to think “I want one of these” instead of wondering if it’s even real.
This guide walks through the full Firefly workflow – from one product shot to a coherent campaign – then goes further than the workflow itself: how to write prompts that produce what you actually want, how to keep multiple product photos visually consistent at catalogue scale, and where AI product photography still struggles enough that traditional shoots still matter.
Why AI product photography matters now.
The shift in AI product photography is no longer experimental. A growing share of catalogue listings, ad creatives and social campaign imagery is being generated rather than photographed. Consumers now notice the product itself instead of the image and its origins.
For freelance designers, that means client briefs that used to require a half-day studio rental can now be turned around without that expense. For volume creators producing Reels, Shorts and brand collaborations, it means positioning the same product across ten different scenes without re-shooting any of them. For small business founders and those selling direct to consumers, a campaign-ready set of visuals can be produced from a single phone-shot photo of a product on a kitchen table.
And because Adobe’s own Firefly models are trained on licensed Adobe Stock and public domain content, eligible outputs from those models are covered by Adobe’s commercial-use indemnification policies. That added layer of protection can matter when you’re creating assets for product pages, advertising or client work.
The promise of this new reality: a full campaign from a single product shot, generated inside Adobe Firefly, in the time it takes to brief a photographer.
The full Firefly product photography workflow.
Check out the video for a visual walkthrough and then follow the step-by-step breakdown when creating your first product photo.
1. Open Quick Actions
Sign in at firefly.adobe.com and open the Quick Actions panel. This is the entry point for the workflow. Every step that follows runs from inside Quick Actions or links out from it.
2. Select Generate Background
Generate Background is the core capability for this workflow. Upload one clean, high-resolution product photo straight from your device or from Adobe Cloud storage. One good source image consistently produces better results than several lower-quality alternatives, so spend a minute picking the right one before you start prompting.
The toolbar on the left gives you access to editing tools such as Remove background, Fill or Expand.
3. Remove the existing background
If the source image still has its original background, use Remove Background to isolate the product automatically before generating a new scene. The cleaner the cut-out, the more believable the final composite. Unless the product is clearly isolated, unwanted, colourful halos or remnant background colour can be carried into every generation that follows.
4. Set your prompt parameters
Back in the prompt settings, choose your AI model. Firefly Image 5 is a good start – it’s Adobe’s latest image model, designed for photorealistic detail, anatomic accuracy and multi-layered compositions at native 4MP resolution. Set your output ratio and image quality, then write your background prompt.
5. Generate and experiment
Click Generate to create your first product scene. The first output rarely lands exactly where you want it. Small prompt adjustments, such as changing the lighting direction, swapping the surface material or tightening the composition note, can produce significantly different results. Treat the first generation as a starting position, not the final image.
6. Add typography and branding details
If you’re happy with the composition and layout but the product still needs a label, a price tag or campaign message: generate again with a targeted prompt that describes the text placement and packaging style. For exact brand fonts, regulated copy or anything where the text really has to be correct, refine the output in Photoshop rather than trying to push the AI further inside Firefly.
7. Resize for different formats with Generative Expand
You’ll almost always need the same composition in multiple formats: square for the product grid, portrait for stories and landscape for an ad placement. Use Generative Expand to extend the canvas. Firefly fills the new space with composition that matches the original instead of re-generating from scratch.
8. Build campaign variations from the same image
This is where AI product photography earns its place: generating multiple compositions from one source photo by varying camera angles, crop and composition notes in the prompt. The discipline that turns a set of variations into a real campaign (rather than a scattered group of one-offs) is keeping lighting, colour palette and props consistent across every shot.
That consistency problem is the hardest part of the workflow and Firefly has three distinct tools for solving it. They’re covered later in the article.
How to write product photography prompts that work.
Structured prompts beat verbose ones. Generic phrases like “cinematic lighting, 4k, ultra detailed” are too vague for a good prompt. What works best is naming each element of the scene with the specificity a real photographer would use on a brief.
A reliable structure for product photography prompts has six slots:
- Subject: what the product is and where it sits in the frame.
- Surface: the material the product is resting on or against (polished marble, brushed concrete, raw linen, weathered wood).
- Lighting: direction and quality (soft key from camera left, gentle fill, subtle rim light separating the product from the background).
- Composition: camera angle, framing, depth of field.
- Mood: minimalist, luxurious, editorial, lifestyle.
- Style: commercial product photography, editorial campaign, magazine cover, lifestyle still.
Put together, one instruction can become a full prompt the AI model can realistically work with:
Luxury glass perfume bottle on a polished black marble surface, soft studio key light from camera left with gentle fill, shallow depth of field, minimalist editorial composition, commercial product photography style.
The thing to avoid is a string of adjectives, e.g., “cinematic, dramatic, beautiful, premium, high-end”, and vague catch-alls that don’t tell the model anything specific. If a word doesn’t change what the image looks like, cut it.
How to keep AI product photos visually consistent.
Generating one good product photo is straightforward. Keeping the next twenty looking like they belong to the same brand is where most AI product photography workflows fall apart.
Firefly helps keep product photos consistent with style matching for generated images and your own photos. Firefly Custom Models takes this further with catalogue-scale brand training. Most general-purpose AI generators offer one approach. Firefly gives you the right tool for the scale you’re working at.
Style match for generated images
Upload a reference image, and Firefly matches its visual style across new generations – lighting, colour, mood and finish. This is the right tool when you have a hero image whose look you want every subsequent shot to share. Style-based Generative Match works on Firefly Image 5; structure-based reference is currently available on Firefly Image 4 only, so if you need to lock the spatial structure of a reference (not just its style), pick Image 4 for that step.
Style match for your own images
The Gemini models in Firefly can be used to match your own photos to a preferred style by uploading a style reference and a content source. Firefly restyles the second image to match the first’s style, including gradients, lighting and colour tone. A prompt pattern that consistently works:
Using the first image as a style reference and the second image as the content source, restyle the second image to match the artistic style of the first, including gradients, lighting and colour tone.
Firefly Custom Models for catalogue-scale brand work
If your brand has a specific visual signature you want every product shot to inherit, Firefly Custom Models (public beta) lets you train a model on a set of your own images. Once trained, the model carries your brand’s style through every generation without you having to upload a reference image each time. This is the right tool when you’re producing a catalogue, not a one-off campaign.
Where AI product photography works – and where it still doesn’t.
AI product photography handles a lot of work brilliantly. Scene composition, lifestyle backgrounds, lighting and mood iteration, and multi-format resizing can be done quickly by AI. It also has clear limits that are worth knowing before you commit a full campaign to it. The strongest workflows pair AI with traditional photography making use of the strengths of each.
Where it works: generating lifestyle scenes around a product, iterating quickly through lighting and mood variations, resizing one composition into every format the campaign needs, prototyping a creative direction before booking a real shoot and producing catalogue-scale variation from a single source photo.
Where it still struggles:
- Label and packaging text. AI models frequently makes mistakes with small print, brand names and nutrition tables. Always start with a source photo where the text is already visible and let the AI build the scene around it rather than asking it to generate the text from scratch.
- Complex reflections. Glass bottles, chrome surfaces and translucent materials sometimes show reflections that don’t match the light sources in the generated scene.
- Precise colour matching. Cosmetics, paint and any product where exact colour fidelity matters often need manual correction in Lightroom or Photoshop after generation.
- Fine product details. Stitching patterns, hardware placement, specific design details on the product itself can shift between generations – start with a real product photo and modify the context, not the product.
Keep in mind that AI is a tool for product photographers, not a replacement, and there’s still a place for skilled photography. The strongest creative workflows pair both: AI for speed, iteration and scale; traditional photography for the moments where craft really has to be there.
Also, AI product photography is not appropriate for some sectors. Pharmaceuticals, food labels with mandatory information, safety equipment and anything subject to compliance review should still use traditional photography.
Going further: refining your Firefly output across Adobe.
A Firefly generation doesn’t have to be the end of the workflow. The all-in-one creative AI studio means moving from generation to refinement to publication without leaving the Adobe ecosystem.
- Lightroom: colour grading at scale across a full set of campaign generations, AI-powered photo enhancements, and the masking and tone tools that bring a generated set into a single visual signature.
- Adobe Express: quick resizing for social, ads, and marketplace listings; templates that turn a Firefly output into a finished post in two clicks.
That’s the point of the integration. You don’t switch between a dozen sites on a dozen tabs – the same work moves through ideation, generation, refinement and publication inside one creative workflow.
Bringing it together.
The Firefly AI workflow is genuinely accessible: one product photo, a structured prompt, a few iterations, and a coherent campaign set from a single source image. The mechanisms for keeping it consistent at scale exist. The limits are knowable.
Open Adobe Firefly and try Generate Background on a product photo you already have. Most of the workflow becomes obvious once you’ve made the first generation.
Frequently asked questions about AI product photography.
You upload a clean product image, write a prompt describing the scene you want around it, and the AI generates the result. In Adobe Firefly the workflow runs through the Quick Actions panel: upload your product, remove the existing background if needed, write a structured prompt and click Generate.
No. A clean, well-lit product photo from a phone is enough to start. Higher-resolution source images produce better outputs, but you don’t need a studio camera, professional lighting or photo editing software to begin.
Adobe Firefly offers a free tier with limited generative credits, which is enough to try the workflow. Paid Firefly plans start at the Standard tier and unlock unlimited standard generations plus a monthly allowance of premium credits.
For many situations – lifestyle scenes, format variations, catalogue-scale imagery and creative prototyping – yes. For products with tactile materials, complex reflections or precise colour requirements consider a more traditional approach.
No. AI is a tool for product photographers. It can help with opening new creative directions and absorbing the repetitive work. But it’s not a replacement for the craft of photography itself.
Fashion, beauty, food, furniture, accessories and tech work well. Categories that depend on text rendering on labels, complex reflections, precise colour matching or regulatory compliance still benefit from traditional photography or careful manual refinement after generation.
