AI product images for Shopify are generated or edited images that fill your product galleries without a photoshoot. Shopify sets the scope before any tool does: 2,048 variants per product, three product options, and a hard cap of 250 media files per product. Price the job in those units, then pick a tool.
Most guides on this topic quote a cost per product. Shopify does not sell in products. It sells in variants, and a store with 40 products, six colours and five sizes is not a 40-image job. It is a 1,200-variant catalog sitting behind a media limit that was never raised when the variant limit was.
This guide covers the Shopify limits that decide the size of the job, the free AI editing already sitting in your admin, where that editor stops being enough, what to generate in which order, and what the disclosure rules that took effect on 2 August 2026 ask of you.
Key Takeaways
Shopify raised variants to 2,048 and left media at 250. The variant limit went from 100 to 2,048 for all merchants on 15 October 2025 (shopify.dev changelog, accessed August 2026). The per-product media cap stayed at 250 images, videos and 3D models combined (help.shopify.com, accessed August 2026). Three options is still the hard wall. A product can carry up to three options, so colour plus size plus length works and a fourth dimension does not (help.shopify.com, accessed August 2026). Shopify already gives you AI image editing at no extra cost. Media generation in the file editor removes and recolours backgrounds, rewrites a scene from a text prompt, and extends the frame, on Basic and above (help.shopify.com, accessed August 2026). It outputs 1 megapixel. Shopify’s media generation processes at 1 MP by default, while Shopify’s own guidance says 2048 x 2048 px displays best for square product images. That is a quarter of the recommended pixel count. Shopify marks its own AI output. Generated images carry an invisible metadata watermark, and Shopify states it claims no ownership and places no restriction on where you use them (help.shopify.com, accessed August 2026). Article 50 of the EU AI Act applied from 2 August 2026. Penalties reach 15 million euro or 3% of global annual turnover (artificialintelligenceact.eu, accessed August 2026). Generating or editing an image costs 5 credits in DesignerBox. Basic is $15 a month for 500 credits, which is 100 images.
What are Shopify’s product image requirements?
Shopify accepts product images up to 5000 x 5000 px or 25 megapixels, under 20 MB per file. PNG and JPEG are preferred, with PSD, TIFF, BMP, GIF, SVG, HEIC and WebP also accepted. For square product images Shopify recommends 2048 x 2048 px. A single product holds a maximum of 250 media files, counting images, videos and 3D models together.
If you also sell on marketplaces, note that Shopify is the loosest of the five on slot count. The per-platform slot and main-image rules set a much tighter ceiling on Amazon and Walmart.
Those numbers are worth writing down before you shop for a tool, because two of them are the ones that bite. The 20 MB ceiling almost never matters. The 2048 x 2048 recommendation and the 250-file cap decide whether a given tool’s output is usable and whether your plan for the catalog is even possible.
The aspect ratio matters more than the resolution on collection pages. Shopify advises keeping a consistent aspect ratio across featured images so the grid does not go ragged. Mixed ratios are the most common thing that makes an AI-filled catalog look assembled rather than shot, and what Shopify actually serves after upload explains why a correct square still arrives cropped.
| Spec | Shopify limit | What it means for generated images |
|---|---|---|
| Max resolution | 5000 x 5000 px, 25 MP | No generation tool gets close, ignore it |
| Max file size | 20 MB | Not a constraint for JPEG or WebP output |
| Recommended square size | 2048 x 2048 px | 4.2 MP. This is the number to hit |
| Media per product | 250 files, all types | The real ceiling on a wide variant matrix |
| Options per product | 3 | Caps how deep the matrix can go |
| Variants per product | 2,048 | Raised from 100 in October 2025 |
Why your variant matrix sets the image count
Product count is the wrong unit. Shopify stores images against products, and shoppers browse variants, so the number of images a catalog needs tracks the variant matrix rather than the SKU list.
Take a jacket in eight colours and five sizes. That is one product, 40 variants, and one option slot still free. The size axis needs no separate photography because a size 10 and a size 14 in the same colour look the same in a packshot. The colour axis needs eight distinct images minimum, and realistically eight galleries.
So the multiplier you care about is not variants. It is the number of visually distinct variants, which is usually the colour or finish axis alone. Run that number across the catalog before you price anything:
- List your products.
- For each, count only the options that change what the product looks like. Colour and material usually count. Size and length usually do not.
- Multiply that by the number of gallery slots you want per look. Four is a working baseline: packshot, second angle, detail crop, in-context shot.
- That total is your image job.
A 40-product store with an average of six colourways and four slots per look is 960 images. At a studio rate that is a budget conversation. At a per-image generation cost it is a review-time conversation, which is the actual shift, and it changes which SKUs deserve full coverage first.
The 250 media cap is the ceiling nobody prices for
Shopify caps a single product at 250 media files. Videos and 3D models draw from the same pool, so a product carrying 240 images has room for ten videos and no more.
Compare that with the 2,048 variants the same product can hold. The two numbers no longer line up. A product at the top of the variant range cannot give each variant even one dedicated image, let alone a gallery, so the assumption that every variant gets its own visual quietly breaks somewhere above 250.
Where does it break in practice? Divide 250 by the gallery depth you want:
| Gallery slots per look | Distinct looks you can cover | Typical shape |
|---|---|---|
| 8 | 31 | Deep gallery, small colour range |
| 6 | 41 | Apparel with detail crops |
| 4 | 62 | Standard PDP gallery |
| 2 | 125 | Packshot plus one context shot |
| 1 | 250 | Swatch-level coverage only |
Most catalogs never touch this. A store with 12 colourways and a six-image gallery uses 72 of 250. The stores that hit it are the ones with a genuinely wide finish range: print-on-demand, made-to-order furniture, personalised goods, anything where the colour axis runs past 40. The fabric-by-SKU multiplication problem specific to large-format goods is worked through in furniture product photography.
If you are in that group, the fix is structural rather than photographic. Split the product into several products grouped by navigation, which is also how merchants get around the three-option limit. Deciding that before you generate 400 images is considerably cheaper than deciding it after.
Shopify already ships AI image editing at no extra cost
Media generation lives in the Shopify file editor. Open Content, then Files, select an image, or reach it from any product’s media. It removes or recolours a background with automatic subject detection, rewrites the scene from a text prompt, and extends the frame through outpainting (help.shopify.com, accessed August 2026).
It is included at no additional cost for a limited time, on the Basic, Grow, Advanced and Plus plans. There is no app to install and no separate subscription. For a store whose entire problem is 200 supplier photos shot against a messy warehouse wall, this covers the job.
Two things it settles that paid tools often leave open. Shopify states it claims no ownership over images created with media generation and places no restriction on where you use them. And it applies an invisible watermark as metadata, which is the machine-readable marking the new EU rules ask providers to supply.
Start here. The honest sequence for most Shopify stores is to exhaust the free editor first, measure what it cannot do, and buy against that gap rather than against a feature list.
Where the free editor runs out
Shopify’s media generation processes at 1 megapixel by default. Larger inputs are scaled down and smaller ones scaled up (help.shopify.com, accessed August 2026). Shopify’s own product image guidance recommends 2048 x 2048 px for square images, which is 4.2 megapixels.
That is the gap in one line: the built-in tool returns roughly a quarter of the pixels the platform recommends. It is fine on a collection grid thumbnail. It is thin on a zoom interaction, on a retina hero, and on anything headed for print or a lookbook. Legacy assets have the same problem from the other direction, and which low-resolution photos are worth restoring is the same decision run backwards.
Three other jobs sit outside what the file editor does:
- On-model imagery. Putting a garment on a person is a different operation from replacing a background, and it needs a model that handles garment fidelity rather than scene fill.
- Consistency across a batch. Prompt-driven scene generation, run image by image, drifts. Forty products edited one at a time will not share a look, which is the whole reason brand consistency needs a locked input rather than a repeated prompt.
- Video. Shopify’s editor is an image tool. Product video for a listing or a paid placement is a separate pipeline.
The upgrade case is these four things: output resolution, on-model shots, batch consistency, and video. If none of them applies to your catalog, you do not need a second tool. When resolution is the blocker, which model handles product photography is the thing to compare, not which app has the nicest interface.
What to generate first
Order matters more than volume. The first images to generate are the ones a shopper needs to decide, not the ones that look best in a case study.
- Clean packshot, consistent background. Every variant that ships with a supplier photo needs this. It is the highest-volume, lowest-risk operation and the free editor handles it.
- Second and third angle. The most common gap on tail products. A single front-facing image asks the shopper to guess at the rest.
- Scale or detail reference. Texture, closure, finish. This is where returns get prevented.
- In-context or on-model. Highest production value, highest cost, and the one to reserve for products that already sell.
- Video. Last, and only where the placement justifies it. Shopify listings accept video, and product video generated from a Shopify listing runs from the same source photo.
Run steps one through three across the whole catalog before you run step four on anything. Coverage beats polish when half your products carry one image, and pushing a catalog through in one pass is what a bulk catalog processing workflow is for.
Before any of it goes live, run the output past the checks in what to verify on an AI product photo. Colour shift on a product where colour is the variant is the failure that generates returns.
AI disclosure changed on 2 August 2026
Article 50 of the EU AI Act became applicable on 2 August 2026. It sets transparency duties in two directions. Providers of systems that generate synthetic image content must mark outputs in a machine-readable format and make a detection method available. Deployers who generate or manipulate image content that constitutes a deepfake must disclose that it was artificially generated, through a label a person can understand without a detection tool (artificialintelligenceact.eu, accessed August 2026).
Penalties for breaches reach 15 million euro or 3% of global annual turnover, and they apply regardless of where the company is headquartered (artificialintelligenceact.eu, accessed August 2026).
The machine-readable half is your tool’s responsibility, and Shopify already does it with the metadata watermark on media generation output. The visible-label half is yours, and the risk is concentrated in on-model imagery where a generated person appears, not in a packshot of your own product on a clean background.
This sits on top of platform policy rather than replacing it. Marketplaces and ad networks each run their own rules on what has to be labelled and what cannot be generated at all, which the marketplace rules on AI product photos covers listing by listing. Check both before a catalog-wide rollout, not after.
Shopify itself sets none of this. If the question you are really asking is whether you are allowed to use AI photos on Shopify at all, the answer sits in four other rulebooks, and Google Shopping holds the strictest one.
What this costs against a shoot
A studio day for on-model work runs roughly $2,500 to $8,000 and returns 40 to 80 finished images, which prices out per shot at somewhere between $25 and $130. That number is the benchmark, and it is a fair one, because a campaign shoot buys art direction that a prompt does not.
Generation prices differently. In DesignerBox, generating or editing an image costs 5 credits, and the plan sets the monthly ceiling:
| Plan | Price | Credits | Images per month |
|---|---|---|---|
| Free | $0 | 112 | 22 |
| Basic | $15 | 500 | 100 |
| Pro | $35 | 1,000 | 200 |
| Premium | $75 | 2,500 | 500 |
| Ultra | $200 | 8,000 | 1,600 |
Run the 960-image example from earlier through both. At the low end of studio rates it is a five-figure production programme. At $75 a month it is two months of Premium, and the constraint moves from budget to how fast someone on your team can review 960 images.
That review step is the part the cost comparison usually drops, and it is the part that decides whether the catalog looks shot or looks assembled. Budget time for it. Packshots, styled scenes and on-model work all run from the same source photo inside Photo Studio, and DesignerBox pricing lists what each tier includes.
FAQ
Can I use AI-generated product images on Shopify?
Yes. Shopify places no restriction on AI-generated product imagery, and for its own media generation feature it states it claims no ownership and does not restrict where merchants use the output (help.shopify.com, accessed August 2026). Separate rules still apply from marketplaces, ad platforms, and the EU AI Act’s transparency duties that took effect on 2 August 2026.
What size should Shopify product images be?
Shopify recommends 2048 x 2048 px for square product images (help.shopify.com, accessed August 2026). That is the upload spec only. Shopify re-encodes every file on its CDN and the theme decides the crop, which the full breakdown of Shopify product image size works through slot by slot.
How many images can one Shopify product have?
- That cap covers images, videos and 3D models combined, so a product with 240 images has room for ten videos (help.shopify.com, accessed August 2026). The limit was not raised when Shopify increased the variant limit to 2,048 in October 2025.
Does Shopify have a built-in AI image generator?
Shopify’s file editor includes media generation, which removes and recolours backgrounds, rewrites a scene from a text prompt, and extends the frame. It is available on Basic, Grow, Advanced and Plus at no additional cost for a limited time (help.shopify.com, accessed August 2026). It processes at 1 megapixel by default.
Do I need a separate tool if Shopify Magic is free?
Only for four jobs: output above 1 megapixel, on-model or try-on imagery, consistency across a batch of products, and video. If your catalog needs none of those, the built-in editor is enough. Test it on your ten hardest products before paying for anything.
How many product images does my catalog actually need?
Count visually distinct variants, not SKUs and not total variants. Size and length rarely need separate photography. Colour, material and finish do. Multiply distinct looks by gallery depth, with four slots as a working baseline: packshot, second angle, detail crop, in-context shot.
Will AI product images hurt my conversion rate?
The evidence points the other way for warm traffic, but the test needs more sessions than most stores get. Before you attribute any change to imagery, work through the traffic math an image test requires, because underpowered tests read as noise.
Shopify platform limits verified from Shopify’s Help Center and developer changelog, and EU AI Act Article 50 obligations from artificialintelligenceact.eu, as of August 2026. DesignerBox pricing and credit costs from the product brief. Individual results vary.
