7 Checks Before AI Product Photos Go Into a Catalog

Before AI product photos go into a catalog, check 7 things first: product proportions, shadow direction, material authenticity, edge flaws, background consistency, whether the color has been over-enhanced, and the scale relationship among multiple products on the same page. The MINDS Printing (MS, mid- to high-end fully custom commercial printing) three-stage print submission check breaks these 7 items into: ① layout credibility, ② print feasibility, and ③ sales accuracy without misleading the buyer

I often see clients drop AI-retouched product photos straight into a catalog. They look clean on screen, but after proofing, the edges look cut out, the shadow directions fight each other, or a cup on the same page looks larger than a thermos bottle

E-commerce images can use brightness, sharpness, and background mood to attract clicks. A printed catalog needs buyers to believe, "This is roughly what the product will look like when I receive it."

・① Layout credibility: product size, angle, shadow, and same-page scale need to make sense, especially when 10 items are arranged on one page

・② Print feasibility: cutout edges, resolution, shadow detail, and highlight detail must hold up when inspected on paper. An A4 page is much more honest than a phone screen

・③ Sales accuracy without misleading: color, material, gloss, and texture cannot be retouched until they no longer resemble the real product, especially for fabric, wood grain, metal, and food packaging

"Good-looking" can only come second in a catalog. "Credible" comes first

I have said this many times on prepress floors, because reprinting does not become cheaper just because an AI image looks beautiful

E-commerce images are usually skimmed quickly on phones. The image size is small, viewing time is short, and consumers often judge by the main image, selling points, and reviews together. A printed catalog, however, is used by sales teams for in-person explanations, and customers may stay on the same page comparing 3, 5, or even 10 items

That is the difference

E-commerce images allow a little drama. Catalog images need stable presentation

・Acceptable for e-commerce: a slightly brighter background, a softer shadow, or product edges that are not obvious in a thumbnail. The goal is usually to increase clicks

・Needs caution in a catalog: once products on the same page are compared, shadow angles, proportion differences, and color saturation all become exposed. The goal shifts to helping purchasing teams make decisions

・Common e-commerce scenario: one image stands alone, with one product in one frame

・Common catalog scenario: products in the same series appear on the same page, so bottle height, packaging thickness, and material gloss are compared against each other

A very common case: 3 bath and body products from the same series. AI makes the shoulder line of one bottle sharper and the glossy surface stronger. On its own it looks refined, but once placed into the catalog, it looks like a different capacity or different packaging material

When purchasing teams see a page like this, their first question is usually not about price. It is, "Are these three really from the same series?"

For product proportions, set a reference point first, then compare within the layout. When multiple products appear on the same page, do not use the visual size output by AI directly for layout. At minimum, manually correct it once based on capacity, actual dimensions, or the relationship within the series

I ask designers to check 3 points first: height, width, and visible area

These 3 points are more reliable than "it looks about right."

・Products in the same series: use actual height or capacity as the proportional reference first. For example, 250ml, 500ml, and 1000ml should not be laid out at nearly the same height

・Different product categories: decide the purpose of comparison on the page first. Are you comparing size, comparing appearance, or simply creating a lifestyle pairing?

・Products with packaging boxes and bare items: boxes, bottles, and accessories should be evaluated separately. Do not let an AI composite make accessories more visually dominant than the main product

・Handheld or scene images: hands, tabletops, cups, plates, and background props all imply size. If those cues are wrong, the product proportions will be wrong too

The biggest problem in catalogs is when "the proportions look good, but sales cannot explain them clearly."

For hardware parts, food gift boxes, skincare sets, and similar products, customers often use catalogs for initial purchasing decisions. Wrong proportions can complicate quotations, inventory discussions, and on-site communication all at once

When the MINDS Knowledge Academy consulting team helps clients organize catalog assets, we usually recommend making one "proportion proof sheet" first, placing products from the same series on a single A4 or A3 layout for review

This step does not take long, but it catches the scale errors that AI images most easily hide

For shadows, check direction, density, and whether the product feels grounded. For edges, check cutout fringing, semi-transparent ghosts, and tiny holes. For backgrounds, check whether color temperature, brightness, and horizon lines are consistent

Together, these 3 areas are usually the key to whether AI product photos look as if they truly belong on the same page

I use 100% view to inspect edges first, then zoom back out to the full page to evaluate the overall mood

That is because edge issues need magnification, while shadow and background issues only become clear at full-page scale

・Shadow direction: when 5 products appear on the same page, the light source cannot come from the upper left for some and the lower right for others, unless the layout is clearly built as different scenes

・Shadow density: the shadows of a transparent bottle, metal box, and matte paper box will not be identical. AI often retouches them into the same soft-focus gray shadow

・Grounding: the bottom of the product cannot look like it is floating, especially for products that need to stand firmly, such as bottles, jars, boxed goods, and small home appliances

・Cutout edges: check gray edges on white products, bright edges on dark products, and broken edges on wool, bristles, and transparent plastic, which are especially prone to edge artifacts

・Background consistency: the same page should not contain 3 kinds of white. Once printed on paper, yellowish, bluish, and grayish casts become more obvious than on screen

Many AI retouching workflows make product edges too clean

Real products have thickness, bevels, and tiny reflections. When an edge looks perfectly knife-cut, it instead looks pasted on

If this catalog will be produced as mid- to high-end commercial printing, a fully custom printing workflow such as MINDS Printing will pay closer attention to edge and background consistency during proofing

Paper, coating, matte lamination, and spot gloss effects can all amplify subtle differences in image detail

AI retouching most easily makes products brighter, more saturated, and smoother than the real item. But the job of catalog color is to help people buy the right product, not make the picture outperform the physical object

This is especially true for 4 categories: apparel fabrics, wooden furniture, metal hardware, and food packaging. Once the material is over-retouched, trust drops quickly

First, clarify the core term: color management is the process of keeping the monitor, file, proof, and final printed piece as consistent as possible. Common tasks include monitor calibration, assigning an ICC Profile, checking CMYK conversion, and comparing against proofs. The goal is to reduce color variation

Before AI product photos go into a catalog, I require at least one "screen to paper" imagination check

Do not only ask whether it looks beautiful. Ask whether it will still look like the actual product when printed

・Fabric: texture cannot be polished into a plastic feel. Weave, nap direction, and thickness need to remain visible

・Wood grain: grain patterns cannot become so regular that they look like a texture map. Light and dark layers should stay close to the real object

・Metal: highlights cannot blow out into large white areas, and reflection positions need to match the product shape

・Transparent materials: bottles, acrylic, and glass need to retain refraction and thickness, not just a bright outline

・Food packaging: brand colors cannot drift away from the real product just to look more vivid. Red, orange, and green are the most common problem colors

Printed catalogs usually enter a CMYK workflow. Much of the fluorescent feel, bright blue-green, and highly saturated orange-red seen on RGB screens becomes more restrained on paper

If the AI image is already pushed too far, color conversion often produces one of two outcomes: it either becomes dull, or the hue shifts

My method is simple, but effective: place the actual product, proof, and screen on the same table

Keep the lighting controlled. Look at the main color first, then the shadows, and finally the material

This is much closer to print reality than spending half a day moving sliders on screen

・Before AI product photos go into a catalog, check credibility first, then aesthetics

・E-commerce images aim for clicks. Printed catalogs carry the responsibility of supporting purchasing decisions

・When multiple products appear on the same page, proportions must be corrected. The visual size output by AI cannot be taken at face value

・Cutout edges, shadow direction, and background whites are the weak points in AI retouching that proofing most often exposes

・Do not retouch color beyond the real product. A catalog sells trust, not a filter

When bringing AI into the catalog workflow, the most valuable step is not handing off all retouching. It is standardizing the checkpoints: the design side creates proportion proof sheets, prepress checks resolution, edges, and CMYK conversion, sales confirms whether product color and material might mislead customers, and a SaaS system can turn these fields into required checks before asset submission

If a small or medium-sized business does not yet have a complete workflow, it can start with the MINDS Printing (MS) three-stage print submission check, sorting each batch of AI product photos into three groups: ready for layout, needs retouching correction, or needs reshooting or rebuilding. Clean assets give layout, proofing, and printing a real basis for quality discussion