A product mockup is a realistic image of your product before it even exists: your tea box sitting on a stone countertop, your logo printed on a t-shirt worn out on the street, your app displayed on a laptop. Until recently, you needed either a photographer, a Photoshop template, or a stock-image subscription. AI image generators now produce the same kind of result in one to two minutes, starting from a simple written description and, if you want, your own artwork to drop in.
This guide walks you through three methods, from the most beginner-friendly to the most precise for client-ready work. You'll find the exact prompts to copy for packaging, apparel, and a screen, the four flaws that make a fake obvious in two seconds, and the exact point where AI alone stops being enough. No retouching skills required to get started.
Product mockups: what AI does well, and what it still gets wrong
AI builds a believable photo scene around a product in seconds, but it stays unreliable as soon as it has to reproduce your logo or your text character for character.
A mockup (or presentation render) shows your product in context. You use it for a product page, a sales page, an ad, an investor deck, or an Instagram post, without having manufactured anything or organized a photo shoot.
There are two technical approaches, and the difference in wording matters because the results are nothing alike:
- Full generation: you describe the scene in text, and the AI invents the product, the setting, and the lighting. It's fast and very flexible, but the text on the label will be approximate.
- Compositing, also called guided editing: you supply your artwork (logo, label, screenshot) and ask the model to apply it to an existing object. It's more faithful, and it's what you should aim for whenever your brand has to be legible.
| What AI does well | What it still often gets wrong |
|---|---|
| Settings, lighting, materials (glass, kraft, cotton) | Long text and fine print |
| Ten different moods in five minutes | Exact reproduction of a complex logo |
| Believable fabric folds and cardboard volumes | Consistent reflections on a screen panel |
| Format variations (square, vertical, banner) | Strict brand-guideline compliance (Pantone, custom typeface) |
The rule to remember is simple: the more text that has to be reproduced faithfully, the less AI can handle on its own. In that case, you generate the setting with AI and add your real artwork on top, as explained below.
Method 1: create your mockup in Skilzy's AI Lab
This is the most direct route for a complete beginner: you learn to write the prompt and generate the image in the same place, with the tool credits already included.
Skilzy is a French e-learning platform for learning and using AI. Its built-in AI Lab lets you generate images without stacking up subscriptions, and the programs explain what to write to get a usable result instead of one more blurry image. For product mockups, the most useful track is the AI ad creation program: it covers the product scene, lighting, format variations, and the hooks that go with the visual.
Here's the concrete process, in five steps:
- Pick your medium (packaging, apparel, screen) and prepare your artwork as a PNG with a transparent background if you have one.
- Open the AI Lab and paste in a structured prompt (the three templates in the last section work as is).
- Generate three variations and keep the best composition, not the prettiest lighting.
- Ask for a targeted edit: swap the label for this artwork, brighten the background, recrop to vertical 9:16.
- Export, then upscale the image if it's headed for print.
If you're using these visuals professionally, two certifications registered in France's répertoire spécifique are eligible for funding: RS7439 (AI content marketing) and RS6792 (AI and sales). You can check their registration directly in the official France Compétences directory.
Method 2: consumer AI image generators
A standard image generator will produce a decent mockup from a single prompt, as long as you accept that the label text will be reinvented.
All the major image models can produce a product scene. The most interesting ones for a mockup are those that accept a reference image as input: you upload your logo or your screenshot, then ask for it to be applied to the object. Gemini's image editing feature, for example, is publicly documented in the Gemini image generation documentation, along with the accepted formats and resolution limits.
What you should know before diving in:
- Free quotas run out fast. Two or three images a day depending on the tool, then it's wait or subscribe.
- Watermarks vary. Some free versions stamp the image, which makes it unusable on a product page.
- Commercial usage rights aren't automatic. Read the tool's terms of use before you publish a paid ad.
- One subscription per tool adds up. Between an image generator, a video tool, and an editor, you quickly end up with three monthly charges.
If you want to compare the options available without paying, our test of tools for generating images with AI for free breaks down the quotas and the actual quality of each one.
Method 3: non-AI tools, when your artwork has to stay intact
When your logo or label has to appear exactly as you designed it, a template-based tool is more reliable than generative AI.
Ready-made mockup libraries
Smartmockups (built into Canva), Placeit, Mediamodifier, and Mockey all work on the same principle: you upload your file, and the tool automatically maps it onto a real photo (t-shirt, mug, box, iPhone). Your artwork isn't reinterpreted, just warped to follow the perspective.
Upsides: your logo stays sharp and legible, the perspective is correct, and the result is ready in thirty seconds. Limits: the catalog is finite, the same photos circulate across thousands of shops, and you can't adapt the setting to your brand world. Free versions often add a watermark or cap the resolution; paid plans generally start around €10 to €20 per month depending on the tool (check current pricing on their site, it changes often).
This is the right call for a print-on-demand shop, an Etsy catalog, or a Shopify listing where the legibility of the design matters more than the originality of the setting.
A Photoshop or Figma template, for an exact result
The original method is still the most precise. You open a mockup PSD file, double-click the smart object layer, and paste in your artwork: Photoshop repositions it in perspective on its own, with the original folds and shadows. Figma offers equivalent plugins for screens, and Blender lets you model packaging in 3D if you need an angle that's impossible to photograph.
Budget thirty minutes to an hour of learning, plus a software subscription for Photoshop. In exchange, you get an editable file at the right resolution, with perfectly crisp text.
The best compromise is a hybrid approach, and it's what most professionals do: you generate the setting and the mood with AI (asking for a blank label), then composite your real artwork in Photoshop or Figma. You keep the creative freedom of AI and the precision of a template. If the final file is going to print, run it through an upscaler: our comparison on how to enlarge an image without pixelating it explains how to increase resolution without damaging the edges.
Prompts that work for packaging, a t-shirt, or a screen
A good mockup prompt describes five things in this order: the product, the material, the setting, the lighting, and the camera lens.
Most bad results come from a description that's too short. Here are three proven starting points to copy and then adapt to your product.
Packaging:
Product photo of a matte kraft cardboard tea box, 10 cm cube format, sitting on a light stone countertop. Two eucalyptus leaves placed to the right. Natural side light coming from a window, soft shadows and short cast shadows. 50 mm lens, f/2.8 aperture, slightly blurred background. Square crop, blank label, empty space at the top for text.
Apparel:
Photo of a person seen from behind wearing a thick off-white cotton t-shirt, natural folds across the shoulders. Cobbled street in late afternoon, blurred background. Warm, low-angle light. Vertical 4:5 crop, torso centered, back area clearly visible and flat so artwork can be added.
Screen:
Mockup of a silver laptop at a three-quarter angle on a light wood desk, screen on and showing a clean, empty web interface. Notebook and coffee cup to the left, outside the focal plane. Soft studio lighting, no reflections on the panel. 16:9 crop, screen perfectly sharp.
The four flaws that give away an AI mockup
- Warped text. Made-up letters or half-erased words on the packaging. Fix: ask for a blank label, then add your text on top in Canva or Photoshop.
- Impossible reflections. A window reflected on a screen even though the scene is outdoors. Fix: specify no reflections on the screen in the prompt.
- Inconsistent shadows. The product lit from the left but its shadow also falling to the left. Fix: specify a single light source and which side it's on.
- The plastic look. Materials that are too smooth, with no imperfections at all. Fix: add photorealistic, slight film grain, micro-imperfections in the material.
Once your mockup is approved, it becomes the foundation for everything else: product page visual, thumbnail, banner, or opening shot of an ad created with AI.
Key takeaways
For a clean first mockup without installing anything, start with method 1: an environment where you learn the prompt and generate the image in the same place. If you just want to put together a one-off image, a consumer generator will do the job, watermarks and quotas included. And as soon as your logo has to be reproduced down to the millimeter, go through a smart-object template, ideally over an AI-generated setting. The real time savings aren't in the tool, they're in the prompt: write a precise one, and you get a usable image on the first try.
