AI POP Displays vs Midjourney for Retail Display Concepts

I've spent close to four years in point-of-purchase, and a good stretch of that trying to make general image models produce something I could put in front of a customer. Midjourney got more of those hours than anything else, because in 2023 and 2024 it was the model everyone was talking about, and it seemed obvious that it would produce the best displays.

It didn't. On display briefs it kept coming up short in the same places, and those places are the reason I build AI POP Displays.

For the comparison with ChatGPT, see AI POP Displays vs ChatGPT. For the workflow context, the AI POP display generator guide.

Why isn't taste the constraint?

Midjourney is a general image model. It makes pictures of people, landscapes, atmosphere and concept art without knowing what any of the objects in them are. That's fine for a picture. It's the whole problem for a display.

A display brief is a fixture with a pallet footprint, a header card at a printable proportion, a shelf count that matches the packs, and a client's actual logo on it. None of those are questions of taste. They're questions of knowing the object, and a general model has no way to know it. A pleasing render of a display-shaped thing isn't a display concept, and the gap between the two is where every one of the limits below comes from.

Where does Midjourney stop on a POP brief?

Some of these come from Midjourney's own documentation, and the rest from running real briefs through it.

There is no way to place your client's logo on the fixture. This is the one that ends it. Midjourney's ways of pointing at an existing image don't insert an asset unchanged. Style reference transfers color and texture, and the documentation says plainly that it doesn't copy objects. Omni-reference takes one image per prompt, and Midjourney's own guidance says fine details such as logos may not match your reference.

One image per prompt means a logo and a pack shot can't both go in as objects. On a real brief you have a logo, three SKUs and a mood reference. The tool has one slot.

It can't put the brand name on the header. Midjourney's own documentation lists "a logo with the company name" among its examples of prompts that don't work well. That's the vendor telling you not to expect the single most common requirement in POP.

Named formats come back as style. Ask for an FSDU and you get something with the visual language of a floor display at proportions no floor display has ever had. Counter glorifiers come back too tall and too theatrical. Endcaps are rarely at planogram height. The model is drawing from a vague visual prior, because "FSDU" is a word to it and a set of dimensional constraints to us.

Materials come back as adjectives. Acrylic, PET and lacquered MDF are three different objects under store lighting, with different edges, different reflections and different weight on the shelf. In a general model they're three flavors of "glossy". A concept a manufacturer can quote against needs the material to behave like itself.

Revisions fight you. Editing an HD image in Midjourney drops it back to standard resolution, so the edited render no longer matches the one the client saw. When the client asks for a taller header and a material change, the usual route is to regenerate, and regenerating moves everything you'd already approved. Revision rounds are where the hours in this business go.

Your work is public by default and trains the model. Midjourney's published training summary says it trains on user prompts and images, and the license you grant is perpetual and survives cancellation. Stealth mode only controls whether renders show up in the public feed. For an NDA'd launch that's a conversation with your client, not a footnote.

What has to sit on top of an image model?

An image model draws a picture. The work that decides a display happens before the request ever reaches it, and that work is what AI POP Displays is. It runs on Google's Nano Banana Pro. Everything else is the layer between a customer's email and the request the model receives, and that layer is the product.

It's close to four years of knowing which questions decide a display concept and which ones don't. Format names carry real geometry instead of being adjectives, so an FSDU arrives with a footprint, a header proportion and a shelf pitch. The materials list behaves, so acrylic reads as acrylic and lacquered MDF reads as lacquered MDF rather than as a mood. Product load is built from the packs you uploaded rather than from what balances the picture, so it's merchandised, not composed. References carry a role, so a pack shot is used as the product, a logo as the brand mark and a sketch as the structure, and sketch fidelity (faithful, balanced or free) sets how closely the render follows the customer's drawing. Retail environments are options rather than descriptions, so the aisle looks like the aisle your client sells in. A briefing step reads the customer's email and turns it into that structured request, so the first render answers the brief.

Then the revisions. Every edit lands as a new version next to the one it came from, so the image the client approved is never overwritten, and on Pro, apply graphic places the real artwork file on the branded surfaces of a finished render.

That's the job, built in once. In a general model there's nowhere to put any of it, which is why a brief pasted into Midjourney comes back as a picture of a display rather than as a display. If you have a brief on your desk, I'd rather you run it through AI POP Displays than take my word for the layer.

Where does each one belong?

Midjourney. Mood boards and exploration before a brief exists, and scenes built around people.

AI POP Displays. From the brief onward. Client-facing concepts, brand approval, anything a manufacturer will quote against, anything where a real logo has to be a real logo.

Bellto. When the campaign behind the concept has to reach a workshop as drawings, by invitation from the waitlist.

What it costs

AI POP Displays is free to try with 15 one-time credits and no card, then $49 a month for 150 concept generations at the founding price, with the $69 list price stated up front. Nothing you upload or render trains AI models, on any plan. The free credits cover the brief you're holding, so start with that one and judge the fit on your own SKUs.

What we're building next: Bellto

Both tools in this comparison end at an image. For a POP campaign the image is the start, and everything after it, the campaign mix, the sizing of each piece, the drawings a workshop quotes from, still gets done by hand across several companies.

Bellto is the new entrant, and the only one in this comparison that takes a campaign to a manufacturable package. It's an AI agent for brand and shopper marketing teams and their agencies. You tell it what you're 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. It designs every piece with you at real scale in parametric 3D, with proportions derived from the product dimensions and facings, and 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. It then suggests manufacturers that fit by material, format and volume. The quote still comes from the manufacturer, and POP manufacturers can sign up too.

We're building it now. The waitlist is open to any brand and agency, 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.

Where to go from here

The comparison with ChatGPT is AI POP Displays vs ChatGPT, where the argument is what the layer has to do on top of a current image model. The underlying case is in why generic image AI fails at POP displays.

If you have a brief on your desk right now, that's the fastest way to test any of this. Start here, the first render lands in under a minute. And if the campaign after it has to end up in a workshop, put it on the Bellto list.