← back to blog

Your Client Sent an AI Render of a Display. Now What?

November 9, 2026··9 min read

The email has one line of text and one attachment. "Can you make this for our spring launch? Around 200 stores." The attachment is a floor display that looks finished. A lit header with the logo crisp across it, four shelves of product glowing under soft light, a curved clear side panel, a base that seems to hover just off the floor.

It came out of an AI image tool in an afternoon, and the brand manager already loves it.

You can see the problems before you finish scrolling. Nothing in it has a size. The shelves hold more product than the unit could carry. Nobody has decided what any of it is made of. Saying so is the fastest way to lose the job, and quoting it as drawn is the fastest way to lose money on it.

I've worked in point-of-purchase for close to four years and I build AI POP Displays. This post is the manufacturer's side of that email: what the client actually sent, what to do with it, and what to send back. The brand's side of the same email is in how a brand gets an AI concept built.

Search for advice on this and you land with sign and print shops. They have pages on the AI artwork customers send in, and they're about print files: resolution, vectors, garbled lettering, who owns the image. A display has those problems too, plus a structure, a load and a format, and that half nobody covers.

What did the client actually send you?

A picture of what they want the shopper to see. Read it that way and it's useful.

The image answers questions a written brief usually leaves open. The format family they have in mind, floor or counter, open shelves or a hook wall. The material feel, warm wood or clear acrylic or printed board. Where the brand sits and how big. The mood of the light. A brand that used to send a paragraph and a competitor's photo now sends this, and it saves a round of guessing.

What it doesn't carry is anything you measure. A raster image has no millimeters in it. It has no material per part, no thickness, no joints, no idea how the base carries a full load, whether the unit ships flat, or what the retailer will let into the store. The proportions can be completely believable and still be wrong for the product, because nothing in the picture knows how deep the pack is.

So it's a brief with pictures. It isn't a specification, and it isn't an insult to your design team either.

What usually goes wrong in a client's AI render?

The same things every time, because a general image model draws what looks plausible and measures nothing.

The format reads as a style. The client typed "floor display" and got the look of one, with proportions that belong to no unit anybody builds. Headers come out oversized. Shelves float with nothing under them.

Materials come back as adjectives. Acrylic painted like glass, with no edge thickness. Chrome where powder-coated steel would go. Wood grain on something that would be printed board.

The load is composed, not merchandised. The model places as many packs as balance the picture, at whatever size balances it. Sometimes it isn't even the client's product, just a pack in roughly the right colors, and the client hasn't noticed because everything around it looks so finished.

Light comes from nowhere. Glowing headers and backlit panels with no LEDs, no diffuser and no route for a cable.

It never says when it's wrong. A general model has no failure state, so an unbuildable base comes back at the same confidence as a buildable one. I went through how to read a render for scale in how accurate are AI display renders, and the same checks work on a client's.

None of that makes the image useless. It makes it a first draft somebody else made, of a piece you're going to have to stand behind.

Should you tell the client it can't be built?

No. Show them the version that can.

The client spent an afternoon on that image, and it's their idea. A reply that lists what's wrong with it reads as "we don't want this job", even when every point is right. A shop that answers with a buildable version of the same idea, in the store it's going to, reads as the shop that understood the brief.

The opposite mistake is quoting the picture as drawn. Every shop reads an open image its own way. One assumes thick acrylic and a bonded joint, another sizes the shelves to the packs and another to the picture, and the quotes end up pricing different displays. Whoever reads it cheapest wins the job and then finds the difference at the sample stage.

So the work is two steps. First, take the client's image seriously enough to work on it. Then answer it with your own concept, one that keeps what they liked and fixes what can't stand.

How do you work on the client's own image?

Upload it in Apps, the page for renders made outside AI POP Displays. It's free. You pick the sector and the display type, add the brand and the material if you know them, and it opens in the editor as the original upload, version one, the root of its own edit tree inside a project. The client's image is never overwritten. Every operation lands as a new version next to it, and credits are spent only on the operations you run.

Two of them are worth running on almost every client render.

Change background, on the free plan. Client renders tend to arrive on a white void or on a set that doesn't exist. Put the display in the store it's going to, from presets like supermarket, pharmacy, retail aisle, boutique or tech store, or from your own description. A display seen in its aisle is where the scale conversation starts, because the gondola next to it is a ruler everybody already knows.

Generate angles, on Pro. The client sent one view, usually a flattering three-quarter. Generate the front, the side and a top-down view of the same display, in perspective or as flat elevations on a clean background, as a photo or as a line drawing. The side is where depth goes missing, because a three-quarter view hides how shallow or deep the unit really is. The flat front is where you read the shelf spacing and the header against the whole.

Compare then puts any two versions side by side, at no cost in credits, so you can check the display held through the change.

Anything that changes the display itself, the format, the materials, the product on the shelves, belongs in your own concept, not in edits on theirs.

If a client's image is sitting in your inbox right now, upload it for free and look at it in the aisle it's meant for.

How do you turn it into your own proposal?

Start a new concept from the brief, and use their image as a reference rather than as the subject.

In AI POP Displays you paste the client's email, and the briefing step reads it against a display taxonomy, so format, material, environment and product load land in named fields and the gaps show up as gaps. The client's AI image goes in as a POP reference, so the tool reads it for its look and its structure rather than as something to copy. Then you add what their render never had.

A format with proportions. The display type comes from a taxonomy that carries its proportions, so a floor unit starts out taller than it is deep, with a base under its load, and a counter unit leaves the cashier room to work.

A material that behaves. Acrylic with edges, powder-coated steel that isn't chrome, corrugated with fold lines.

The real packs. Upload photos of the client's products and they're used as the product, at scale, facing forward, with each variant kept separate. That's often the first difference a client notices between the two images, because the pack in their render was never quite theirs.

The logo in its own slot, as a mark to reproduce rather than a mood to borrow from.

You end up with two images answering the same brief. Theirs shows what they had in mind. Yours shows the same idea as a display your shop would build, in the store, with their product on it. If the brief and the packs are already on your desk, make that second image on the free plan before you reply to the email.

What should you send back?

Three images and a sentence.

Their render in their store, from the background edit, so they know you started from their idea. Your concept, from the brief, with their real packs on it. And the views that show the difference, front and side, where the depth and the header are easiest to read.

The sentence says what each image settles. The look, the proportions and the capacity are agreed in the images. The dimensions, the material per part and the joints follow in the drawing. Clients take that well when they hear it up front. What they don't take well is hearing it after they approved.

Then ask what the picture couldn't tell you. Which part of their image they're attached to, the light, the curve, the header. How many stores, how many facings per shelf, which retailer, what date. Those answers are the brief their image was standing in for.

If the client later wants to reuse your concept in their own deck, the ownership question comes with it, and I went through that in who owns an AI-generated display render.

Where does a quotable version come from?

The client's image settles the look. Your concept settles the format, the material and the load. A quote needs what neither image has: real dimensions derived from the product, a material and thickness for every part, the joints, drawings and cut files. That's geometry, and it's the half we're building now.

Bellto is an AI agent for brands and agencies that takes a POP campaign from brief to shop drawings. The brand tells it what it's launching (brand, product, channel, stores, budget, date) and it asks for what's missing. It proposes the campaign mix and materials, floor stand, glorifier, shelf strip, stopper, with the budget split per store, and designs every piece at real scale in parametric 3D, with proportions derived from the product and the facings. A dimension change recomputes the whole piece and re-runs the manufacturing checks in seconds. Every approved piece comes out as a 3D model, a part-by-part cutlist, dimensioned drawings, a STEP of the assembly and a cut DXF per part, a package a workshop can quote without redrawing, and it suggests manufacturers that fit by material, format and volume. The quote comes from the manufacturer. For a shop, that's the email at the top of this post arriving as a package instead of a picture.

We're building it now. The waitlist is open to any brand, and we're contacting the first ones soon to run the first real campaigns end to end. It opens by invitation, in small groups. Pricing goes first to the people on the list. No card. POP manufacturers can sign up too.

For the client render in your inbox today, signup is free with 15 one-time credits and no card required, and uploading the client's image costs nothing. Pro is $49 a month for 150 concept generations at the founding price, with the $69 list price stated openly. Nothing you upload or render trains AI models, on any plan, which matters when the image in that email is a launch nobody has announced yet.

Frequently asked

What should a display manufacturer do when a client sends an AI-generated design?

Treat it as a brief with pictures. The image shows the look the client wants, the format family, the material feel and where the brand sits, and it carries no dimensions, no material per part and no real product load. Don't turn it down and don't quote it as drawn. Upload it to AI POP Displays for free, put it in the store it's going to, and answer it with your own concept built from the client's brief and real packs, so the client sees their idea as a display your shop would build.

Can you quote a POP display from a client's AI render?

Not as drawn. A raster image has no millimeters, no thickness per part and no joints, so every shop that quotes it is pricing its own reading of the picture, and the quotes end up describing different displays. The render settles the look. A quote needs real dimensions derived from the product, a material for every part, drawings and cut files. Bellto, an AI agent for POP campaigns that we are building, designs each piece at real scale in parametric 3D and delivers that package.

Can I edit a render made in another tool with AI POP Displays?

Yes. In Apps you upload a PNG, JPG or WEBP image of the render, pick the sector and display type, and it opens in the editor as the original upload, the root of its own edit tree. Uploading is free and the operations you run on it use credits. Change background is on the free plan and Generate angles is on Pro, and every operation lands as a new version next to the original.

What should I send back to a client who made their own AI concept?

Their image placed in the store it's going to, so they know you started from their idea, your own concept of the same idea built from the brief with their real product on it, and front and side views where the depth and the header are easiest to read. Add one sentence on what the images settle, the look, the proportions and the capacity, and what follows in the drawing, the dimensions, materials per part and joints.


Start generating Get on the Bellto waitlist