AI images look generic because of what they are anchored to, not which model made them. A prompt describing a sneaker invents a plausible sneaker. A render built from a photo of your sneaker shows yours, with its real stitching, logo placement, and materials. Anchor the generation to an actual product photo and most of the generic look goes away before you touch a single setting.
You have seen the output. The lighting is beautiful, the composition is clean, and the product is subtly wrong. The logo sits a few millimetres off. The stitching runs the wrong way. The cap is the right colour but the wrong shape. Nobody on the team can say exactly what is off, and everyone agrees it cannot ship.
That failure is usually blamed on the model. It is a briefing problem. The model did exactly what you asked: it drew a convincing example of the category you described. You wanted your product, and you gave it a sentence.
This covers the tells that give AI imagery away, the real cause under each one, and the fixes that hold up on a product detail page or in a paid ad.
Key Takeaways
- Anchoring beats prompting. A description generates a category-average product. A real product photo as input generates your product. This one input decides most of the outcome.
- The tells are catalogued, not mysterious. Peer-reviewed research groups them into anatomy, texture, function, physics, and cultural plausibility, and human reviewers notice anatomy most (arxiv.org, July 2026).
- People are better at spotting it than teams assume. Study participants correctly identified AI-generated images 76% of the time, and accuracy climbed to 82% when they looked for 20 seconds (arxiv.org, July 2026).
- Visible AI costs trust roughly 4 to 1. 31% of consumers trust a brand less when they notice AI-generated marketing content, against 7% who trust it more (emarketer.com, July 2026).
- Fix texture and light before you fix the prompt. Waxy surfaces, doubled shadows, and cinematic over-saturation are the most common giveaways after anatomy.
- Editing passes keep resolution. Retouching, relighting, changing angles, and swapping backgrounds each preserve detail, so you correct rather than reroll and lose the product.
Why do AI images look fake?
Most generic output comes from the input, not the model. A text prompt asks the model to produce a statistically typical member of a category, so it returns the average of everything it has seen labelled “sneaker on a plinth”. Average is exactly what reads as generic. When the render starts from a photograph of a specific object, the model has to respect that object’s real geometry, materials, and markings.
This is why swapping models rarely fixes a generic result. The same brief that produced a plausible-but-wrong bottle in one model produces a plausible-but-wrong bottle in the next one, because plausible-but-wrong is what a description asks for. You are not fighting image quality. You are fighting the absence of a reference.
The second-order effect matters more for brands. A category-average product is on-brand for nobody. Every asset drifts a little further from the real thing, and the drift compounds across a campaign until the ads and the product page look like two different companies.
What are the tells that give an AI image away?
Researchers have catalogued the giveaways rather than guessed at them. A CHI 2025 study of diffusion-generated images sorted artifacts into five groups: anatomical implausibilities, stylistic artifacts, functional implausibilities, violations of physics, and sociocultural implausibilities. When participants explained their reasoning, 61% of comments pointed at anatomy and 30% at style (arxiv.org, July 2026).
The same study is a useful reality check on how much you can get away with. Participants correctly identified AI-generated images 76% of the time and real photographs 74% of the time. Detection improved with attention: 72% accuracy after one second of viewing, rising to 82% after twenty seconds (arxiv.org, July 2026). A shopper comparing two listings is closer to twenty seconds than one.
| The tell | What actually causes it | The fix |
|---|---|---|
| Hands, fingers, and grip | Anatomical implausibility, the most-noticed artifact class | Composite from a real hand reference, or keep hands out of frame |
| Garbled text on packaging | Text is drawn as shapes, not typed | Anchor to a photo of your real packaging, or pick a model with native text rendering |
| Waxy, plastic, over-smoothed skin | Stylistic artifact from aggressive smoothing | Retouch the real texture instead of regenerating the surface |
| Shadows falling two directions | Physics violation, no single committed light source | Relight from the source photo and commit to one key light |
| Uncanny symmetry and spotless surfaces | The model averages toward the mean of its training data | Art-direct asymmetry, wear, and a real scene reference |
| Cinematic gloss and over-saturation | Stylistic artifact, the house look of the default settings | Pull saturation back to your actual brand palette |
| Product detail that is subtly wrong | The render was built from a description, not your product | Anchor the generation to your real product photo |
The last row is the one that matters. The first six are craft problems you can retouch your way out of. The seventh is structural, and no amount of retouching saves an image of a product you do not sell.
Does anchoring to a real product photo actually fix it?
It fixes the category of error that retouching cannot reach. Anchoring means the generation starts from a photograph of the real object, so the shape, finish, logo placement, and proportions come from your product rather than from the model’s idea of the product. Craft tells still need craft fixes. Product accuracy is not a craft fix, and it is the one people notice on a product page.
In practice the input hierarchy runs like this. One clean photo of the real product is the floor. Multiple references are better, because they pin down what the object does from more than one angle. Kontext Multi exists for exactly this: combining several references so character and style stay consistent across a set, rather than drifting shot to shot.
Every asset in DesignerBox builds on your actual product photo, which is the reason nothing comes out looking generic AI. The gallery is the honest test of that claim, because every image in it started from one product photo or one line of text and you can judge the result yourself.
Be aware of the tier. Importing your own photos, writing custom prompts, and remixing are Pro features at $35 a month. Hyper realism sits on Premium at $75. The free plan gives you 112 credits to see whether the output holds up, but the anchoring workflow this article describes starts at Pro. That is a real constraint and worth knowing before you plan around it.
How do you fix texture and lighting without losing the product?
Correct in passes, not in rerolls. A reroll throws away the product accuracy you just established and hands you a fresh set of artifacts to fix. An editing pass keeps the source and changes one thing. In DesignerBox, retouching, relighting, changing angles, and swapping backgrounds each preserve the detail and resolution of the original, so the final asset is ready to ship.
That distinction is the whole workflow. When the skin looks waxy, retouch the skin. When the shadows disagree, relight. When the background is wrong for the placement, swap the background. Each of those is a 5-credit edit, and each one leaves the product untouched.
Reroll culture is what produces the plastic look. Teams generate forty variants, pick the prettiest, and ship an image whose product accuracy nobody checked, because checking forty variants is nobody’s job. Generating two and correcting one beats generating forty and hoping. It is also cheaper: two generations and three edits is 25 credits, which is a fifth of what the forty-variant habit costs.
Lighting deserves a specific note. Doubled shadows and mismatched reflections are the physics-violation class from the research, and they are the tell that survives casual review and then fails a creative director. Commit to one key light direction in the source photo and carry it through every downstream asset. Consistency across a campaign is a separate discipline, covered in the guide to keeping every asset on brand.
Do you have to tell people the image is AI?
Sometimes, and the rules are platform-specific. Meta’s help documentation states that ads created or significantly edited with Meta’s own generative AI features carry an “AI info” label, and that minor changes such as resizing or colour correction do not trigger it. Meta also notes the label is rolling out gradually, so coverage is not uniform (meta.com, July 2026). Check the current policy of every platform you run on before a campaign ships, because these rules have moved repeatedly.
Disclosure is a separate question from accuracy, and the second one carries more commercial risk. An image that misrepresents the shape, colour, or material of what arrives in the box is a returns problem and a reviews problem regardless of how it was made. Anchoring to the real product is what keeps a generated image a truthful depiction, and there is a set of accuracy checks worth running before a model earns your catalogue.
The trust data argues for restraint rather than concealment. 31% of consumers say they trust a brand less when they notice AI-generated marketing content, against 7% who trust it more, from a December 2025 survey of 8,000 consumers across eight markets, published in Klaviyo’s 2026 AI Consumer Trends report (emarketer.com, July 2026). Getty Images research found nearly 90% of consumers want to know whether an image was created with AI, though that study’s fieldwork ran between 2022 and 2023 and predates the current model generation (gettyimages.com, April 2024).
Read those numbers precisely. The penalty attaches to content people notice as AI-generated. That is a quality threshold, and it is the same threshold this whole article is about.
What does a working method look like?
Five steps, in order, each one closing a specific failure mode.
- Shoot one accurate source photo. Real product, honest colour, one committed light direction. Everything downstream inherits this. Skipping it is what the rest of the process cannot recover from, and it is why generation shifts the shape of a photography budget rather than deleting it.
- Anchor, do not describe. Load the photo as the input. Use the prompt to direct the scene, the light, and the framing, never to describe the product itself.
- Add references for anything repeated. Multiple references pin down a look across a set. This is what stops the third shot in a campaign quietly becoming a different product.
- Correct in passes. Retouch texture, relight, swap the background, change the angle. One change per pass, resolution intact.
- Review at 100% against the real thing. Put the render next to the product photo. Check logo placement, stitching, material, proportion, text. Twenty seconds of attention is what your customer gives it, and per the research that is when detection peaks.
Model choice comes last, not first, and it is a per-shot decision rather than a subscription decision. Native text rendering matters for packaging, multi-reference editing matters for consistency, and different jobs want different models. The 13 image and video models sit behind one subscription for that reason: you swap per shot instead of per bill. Plans start free at 112 credits, and an image generation or edit is 5 credits.
For on-model work the anchoring rule holds and the stakes rise, because a garment that renders as “AI clothes” fails on fabric texture before anyone checks the face. The comparison of AI fashion model generators goes deeper on that.
FAQ
Why do my AI images look fake even on the best model?
Because model quality is not the constraint you are hitting. A text description asks any model to produce a category-typical object, and category-typical is what reads as generic. Switching models changes the aesthetic and leaves the underlying problem in place. Anchor the generation to a photograph of your real product and the same model returns a usable result.
What is the most common giveaway in an AI image?
Anatomy. In a CHI 2025 study, 61% of participant comments identifying AI-generated images pointed at anatomical problems such as distorted hands and facial irregularities, and 30% pointed at stylistic artifacts like waxy or plastic textures (arxiv.org, July 2026). Physics violations, particularly shadows falling in two directions, are the next most reliable tell.
Can people actually tell that an image is AI-generated?
More often than most teams assume. Study participants correctly identified AI-generated images 76% of the time and real photographs 74% of the time, and accuracy rose from 72% at one second of viewing to 82% at twenty seconds (arxiv.org, July 2026). Accuracy varied widely by image, so a careful shot can pass where a careless one does not.
Does editing an AI image reduce its quality or resolution?
No. In DesignerBox you can retouch, relight, change angles, and swap backgrounds, and each pass keeps the detail and resolution of the source. That is what makes correcting in passes better than rerolling: you fix the specific artifact without discarding the product accuracy you already established.
Do I have to label AI-generated images in ads?
It depends on the platform. Meta applies an “AI info” label to ads created or significantly edited with its own generative AI features, and states that minor changes like resizing or colour correction do not trigger it, with the label rolling out gradually (meta.com, July 2026). Rules differ by platform and change often, so verify current policy before each campaign.
Does using AI imagery damage brand trust?
It depends on whether people notice. 31% of consumers report trusting a brand less when they notice AI-generated marketing content, against 7% who trust it more (emarketer.com, July 2026). The penalty attaches to visible AI, which makes output quality and product accuracy the thing to control, rather than the decision to generate at all.
How many credits does this workflow cost?
An image generation or an edit is 5 credits each. A realistic loop of two generations plus three correction passes is 25 credits. The free plan includes 112 credits, Basic is $15 a month for 500, and Pro is $35 for 1,000. Importing your own photos and custom prompts require Pro, which is the tier where the anchoring workflow starts.
Artifact categories, detection rates and consumer trust figures verified from the CHI 2025 study “Characterizing Photorealism and Artifacts in Diffusion Model-Generated Images” (arxiv.org), eMarketer’s reporting of the Klaviyo and Datalily 2026 AI Consumer Trends survey (emarketer.com), Getty Images VisualGPS research (gettyimages.com), and Meta’s own AI labelling documentation (meta.com), as of July 2026. DesignerBox credit costs, feature gating and model catalog verified against the live product configuration, July 2026. Platform disclosure rules change frequently. Individual results vary.
