AI product photos and video: studio quality without the studio

For most small e-commerce and consumer brands, product photography is one of the largest recurring marketing costs. A proper shoot is $2,000–$10,000 for a day's output. A new shoot is needed every time you launch a product, change positioning, or want to refresh creative.

AI image and video generation has gotten remarkably good in the past year. Off-the-shelf models from companies like Midjourney, Flux, Recraft, Stable Diffusion variants, and the major image APIs can produce studio-quality product and lifestyle imagery from a few sample photos and a written prompt. Video models from Runway, Luma, and similar are quickly closing the gap on short product video.

The result for a small brand: same imagery quality, a fraction of the cost, ten times the speed.

What it can do well

Three categories of imagery where current AI is genuinely production-ready:

1. Product on neutral backgrounds. White-background catalog shots, gradient backgrounds for ads, hero shots — clean, sharp, lit consistently. Indistinguishable from studio in most cases.
2. Product in lifestyle settings. Your candle on a styled bedside table. Your skincare bottle in a marble bathroom. Your apparel on a model walking down a street. As long as the product itself looks right, the surrounding scene can be entirely generated.
3. Short product video. 4–8 second clips of a product rotating, being used, or appearing in a scene. Good enough for social ads, hero animations, and email creative.

What it still does badly

A few categories where AI is not yet good enough to ship without significant human cleanup:

• Text on packaging. Image models still produce gibberish text on labels. Real packaging shots usually beat fully generated ones.
• Human hands holding the product. Getting better fast, but still inconsistent.
• Faithful reproduction of specific product details. If the exact stitching on a bag or the exact pattern on a fabric matters, you need a hybrid workflow — real product photo, AI-generated background.
• Long-form video. Anything past 8–10 seconds with consistent characters is still hit or miss.

What we typically build

A Genesee AI image and video deployment for a brand usually includes:

1. Product reference setup. A small library of real product photos used as references so the AI generates the actual product, not a generated approximation.
2. Brand style training. Documented prompts and references that produce on-brand output — color palette, lighting style, mood, model styling.
3. A generation pipeline. From a written brief to a batch of options, ready for human selection.
4. A review and approval interface. Where the marketing lead picks the keepers and rejects the rest. AI proposes, humans approve.
5. Integration with where the imagery lives. Shopify product pages, ad platforms, social schedulers, your DAM. New imagery flows where it needs to without manual upload.

What it costs

For most small e-commerce brands, the ongoing usage cost is $50–$300 per month in image and video generation fees. The build is project-based.

Compared to even one $3,000 photo shoot, the math closes immediately. Brands typically use the savings to test more creative variants and refresh imagery more often, rather than to spend less overall.

Where it pays back fastest

The brands that get the biggest lift:

• High SKU count. Brands with 20+ products need lots of imagery. The volume math is what makes AI generation transformative instead of incremental.
• Frequent creative refresh. Brands running paid social ads constantly need new creative. AI lets them test 20 variants for the cost of 1 shoot.
• Seasonal or trend-sensitive merchandising. Brands that need imagery to react to seasons, holidays, or cultural moments quickly cannot afford to schedule shoots for every cycle.
• Limited budget for photography. Brands too early-stage to afford proper studio work can produce launch-ready imagery in days.

The legal and ethical considerations

A few things to be careful about:

• Model rights. If your imagery features an AI-generated person, that person is fictional, which sidesteps model release issues. But if you train a model on real-person reference imagery without rights, that is a problem.
• Style imitation. Generating imagery in the explicit style of a named living artist is murky legal territory. We avoid it.
• Disclosure norms. Some platforms and some markets are starting to require disclosure of AI-generated content in advertising. The EU is ahead here. The US is patchwork. We help clients comply.
• Authenticity. Customers can sometimes tell. Heavily AI-generated imagery in categories where authenticity matters (food, beauty before-and-afters) can backfire. Use AI where it is a true substitute for studio, not where it would be a misrepresentation.

A note on the broader market

This category is moving fast. The capabilities described here are accurate as of mid-2026 and will be better by quarter's end. The pattern we have seen hold: brands that built AI imagery workflows early are pulling ahead of brands that are still scheduling quarterly shoots.

If you want help setting up an AI imagery workflow for your brand, book a free consultation.