AI lingerie model photoshoot in department-store catalog style

Why lingerie stalls where other clothing doesn’t

Most garments go from packshot to on-model photo without drama. Lingerie is different, and the first obstacle comes before any image exists. The image models behind Blurbly run their own safety checks on the input, and a lace bralette on a white background can fail them even though nothing about the photo is explicit.

That is what happened in our test. Nano Banana Pro and Nano Banana 2 both declined the Marenvale packshot as an input, reporting a failed safety check, and generated nothing. Seedream 4.5 took the same photo and the same briefs and produced all three shots. Blurbly doesn’t swap models for you when this happens: you choose a different image model in the studio and run the brief again. Treat a decline as one model’s limit rather than a judgment on your product, and respond by making the brief more neutral, never more explicit.

Brief it like a department store

The images that get through model checks and ad review look like the intimates pages of a mainstream department store. A neutral standing pose with arms at the sides. Soft, diffused light. A warm off-white backdrop. The whole set in frame, at mid or full-body distance. Our studio brief used exactly those words.

The platform rules point the same way. Meta lists lingerie and undergarments as allowed as long as an ad stays within its Adult Nudity and Sexual Activity policy. TikTok’s ad policy lists lingerie that exposes or implies exposure of intimate areas, emphasis on sensitive body parts, sexualized positions and undressing as not allowed. So leave out close crops on the chest or hips, bedroom posing, and sheer fabric that shows what it covers. Our lingerie ad-policy guide quotes the sources.

For lifestyle images, layering does a lot of the work. The set worn under an open knit cardigan by a window, holding a mug, reads as an everyday product photo and gives you something usable beyond the product page.

Showing a size range

Size range is the most common reason lingerie brands look at AI models in the first place. Booking a model for every size is expensive, and shoppers want to see the set on a body like theirs. The size-inclusive shot on this page used the same catalog styling with one extra line: keep her natural body type.

Check that shot for realism, not flattery. Proportions should look like a real body, with straps and band sitting where they would on the real garment. If the image reshapes the body, or makes the band look tighter or looser than the product is, regenerate it. And keep fit advice out of the picture. A generated photo can’t tell anyone whether a 34C runs small; your size chart and fit notes do that.

Checking lace, straps and trims

Lace is where generated images go wrong most visibly, because the pattern is intricate and the model tends to embellish it. One of our three Seedream shots added a small bow and a keyhole that the real bralette does not have. A buyer who receives the plain version would be right to feel misled.

So review each image against the packshot at full size: the lace pattern, strap width, band height, and the colour under warm light, where cream can drift toward ivory or beige. If you sell the same bralette in cream and black, generate each colour from its own packshot rather than recolouring one, and give the black version a closer look at the pattern, since dark detail is easy to lose.

Where these images belong

On marketplaces, keep the white-background packshot as the main image and use on-model shots in the secondary slots. On your own store, the catalog shot can lead the gallery with the lifestyle and size-inclusive shots after it. For paid social, stay with catalog framing and check each platform’s current rules for your category and targeting.

Some work this approach is not for. Blurbly does not make nude or sexualised images, and it prohibits putting real people into lingerie without their consent or anything involving minors. If your brand is built on real customers wearing the product, or you sell bridal and couture lace where the exact pattern justifies the price, a real shoot with signed releases is the better investment. AI catalog shots are strongest on the everyday range: core colours, size coverage, and new drops that need images this week.