Stock vs AI vs Custom Photography: Advice From a Photographer Who Uses All Three
Stock vs AI vs custom photography comes down to one question for every image in your marketing: does it need to be true? If it's decoration, stock is fine. If it's a concept or a product mockup, AI is shockingly capable now. But if the image is your actual team, your actual space, or the reason someone should trust you, only real photography does the job.
I should disclose something before we go further. I have a fine art photography degree and decades of commercial photography behind the camera and out in the field, and right now I'm designing a T-shirt line using AI mockups, with no photoshoot involved. So this isn't a photographer telling you AI is cheating. It's a photographer telling you honestly where each tool wins, including the times my own camera stays in the bag.
What Does This Image Need to Do?
Think of a specific image your website or marketing needs, then answer for that image.
Stock vs AI vs Custom Photography: Where Each Wins
Supporting imagery, abstract backgrounds, generic scenes nobody's identity depends on. Cheap, instant, and perfectly acceptable when the image is wallpaper, not evidence. The risk is sameness: your "diverse team laughing at laptop" is on ten thousand other sites.
Product mockups, placement composites, concept visualization, testing ideas before spending real budget. I'm using it right now to put T-shirt designs on models without a shoot. In 2026, for imagery that illustrates rather than documents, AI is genuinely good, and dismissing it is nostalgia, not strategy.
Your actual people, your actual space, your product in real hands. AI can generate a plausible optometrist; it cannot photograph your optometrist. When the image's job is trust, real is the only tool, and in a web flooding with synthetic imagery, verifiably real photos are becoming more valuable, not less.
Three Projects Where Only Custom Photography Would Do
Remember, I'm the photographer currently using AI mockups for my own T-shirt line. So when I tell you these three client projects all required real photography, it's not reflex, it's the principle in action: concepts and mockups can be generated; people, places, and crafted products can't.
VitaliTi Wellness
Ti Caudron is the brand. Her programs sell because people trust her specifically, and no generated face builds that. AI wouldn't have done her justice, and honestly, it couldn't have done her at all. Personal brands photograph the person or they photograph a lie.
SAGE Veterinary Centers
An on-site shoot of a real facility, real staff, and real animals, used to recruit specialists and reassure pet owners. People deciding where to work or where to bring a sick pet can tell the difference between a hospital and a rendering. On-site is the point.
Haute Home Linens
A luxury linens catalog where the product is the detail: hand-embroidered stitching, fabric texture, the way light sits on real material. That's exactly where AI compositing is weakest. Generate this catalog and the embroidery goes generic, which misrepresents the one thing customers are paying for. When texture and craft are the product, the camera is non-negotiable.
Where Real Photography and AI-Generated Imagery Blur the Lines
Real client websites are rarely purist. Here's how the mix actually plays out across four current projects, including one where AI is a bridge and one where all three approaches share a single site.
Pergolas by Julie: R-WEAVE™
This site ran on 100% real photography for years, until the manufacturer launched a brand-new product, R-WEAVE, and supplied mockups that were obviously AI-generated. The product is so new that no real installations exist to photograph yet. So we used them, and built our own, refining the imagery to look as true to the product as possible until the first real-world installation gets a proper photoshoot. That's the honest playbook for launching something that doesn't physically exist yet: AI as a bridge, with real photography scheduled the moment reality catches up.
Shyft
The Shyft site is mostly stock imagery, deliberately, with one exception: the product boxes are real product photography. That's the split done right. Lifestyle mood can be borrowed; the thing a customer will actually receive in the mail cannot. Photograph what you ship, borrow the atmosphere around it.
Pleasanton Optometry
One website, all three tools, each doing what it's best at. Real photography of the actual practice and team carries about 90% of the site, because patients choosing an eye doctor are choosing people and a place, and credibility demands the real thing. Stock fills supporting moments. And AI illustrates the complex service explainers, like what myopia actually does inside the eye, where a generated visual teaches better than any photo could. Nothing pretends to be something it isn't, which is the entire rule.
The image above is one page doing all three: stock (the large eye), real photography (the patient in red light therapy), and AI-generated mockups for the other services.
Elaine Clark VO Newsletters
I design voice coach Elaine Clark's monthly newsletter every month, and the illustrative imagery is AI-generated. It's the perfect fit: the graphics illustrate each issue's theme rather than document anything real, the cost of stock subscriptions disappears, and every image matches the topic exactly instead of settling for the closest stock result. Here's the detail I love: Elaine's own newsletters argue that AI voices can never replace a human's lived experience, and she's right. AI illustration works for her newsletter precisely because the images aren't the product. Her voice is, and that stays human.
Why AI-Generated Images Often Look Fake (and How a Photographer Fixes That)
Let me be direct about where I stand: I embrace AI imagery, because used in the right places it saves my clients real time and real money. No stock subscriptions, no waiting on a shoot for an illustrative graphic, no settling for the closest available image when the exact right one can be generated. The R-WEAVE launch and Elaine's newsletters above are that savings in action.
But here's the part that decades of commercial photography changes: AI imagery is only as believable as the eye directing it. Most AI-generated marketing images fail on the things a working photographer checks by reflex. Light coming from two directions at once. Shadows that don't match their objects. Materials that reflect like plastic when they should absorb like linen. Depth of field no lens would produce. Skin, wood, and metal all rendered with the same texture.
The average viewer can't name what's wrong, but they feel it, and "feels off" is fatal for marketing.
When I build or refine AI imagery, I'm applying the same judgment I'd use on set: where the key light lives, how shadow falls off, what the lens would compress, how this specific material behaves. That's why the R-WEAVE imagery holds up as product marketing and why Elaine's newsletter graphics feel intentional rather than generated. AI didn't replace the photography experience. It became another place to spend it.
When Everyone's Marketing Imagery Is AI-Generated, Real Photography Becomes the Luxury
Here's the twist in the AI imagery story. Stock photography's weakness was always sameness, and AI is now producing sameness at infinite scale: the same lighting, the same plastic-perfect faces, the same nowhere-rooms, on every website that took the cheap path. Which means the market position of real photography didn't shrink. It moved upmarket. A photo of your actual storefront with your actual name on the glass, your team's actual faces, your product in an actual customer's hands: these are becoming the visual equivalent of a verified checkmark. The businesses that stay real in the next few years will look noticeably, valuably different from the flood. I say this as someone happily using AI where it belongs: the camera isn't dying, it's becoming the premium tool.
Real when it matters.
AI when it's smart.
Frequently Asked Questions
Is AI-generated imagery good enough for marketing in 2026?
For concepts, product mockups, illustrative graphics, and explainer visuals, often yes, and it saves real money on stock subscriptions and shoots. For images representing your actual people, place, or products in use, no: audiences increasingly sense generated imagery, and trust drops when they do.
Why do AI-generated images often look fake?
Usually lighting and material errors: light from two directions at once, shadows that don't match objects, fabric reflecting like plastic, depth of field no lens would produce. Viewers can't always name the problem, but they feel it. A photographer's eye directing the generation is what closes that gap.
When is stock photography okay to use?
For supporting visuals where the image is decoration rather than evidence: backgrounds, mood, generic scenes. It becomes a problem when stock stands in for your team, workspace, or products, because visitors can tell.
How much does custom photography cost for a website?
A half-day shoot typically runs $1,000 to $1,800 and covers most small business website needs. Full brand shoots with multiple locations and styling run $2,500 to $4,500.
Can AI replace product photography?
For placements, mockups, and products too new to photograph, increasingly yes, and it can bridge a launch until real installations exist. For hero images where texture, color accuracy, and material honesty matter, real photography still wins, and misrepresenting products carries real risk.
Should I mix stock, AI, and custom photography?
Most smart imagery plans do exactly that: custom photography for trust-critical images, AI for mockups and explainers, stock for filler. The skill is knowing which image is which, and that's a strategy conversation, not a purchase.
