My Wife Had One Rule About Lingerie Photography

I've been sourcing products from China for about ten years now. One category we never really got into was lingerie, which is slightly strange considering how attractive I've always thought the economics looked. Although, as you'll discover, there was a reason.

Take toys, for example. Almost anyone can go to China, find a factory, put their own brand name on the box, and start selling in Europe. And that's exactly what a lot of European distributors and companies like ours do. The competition is enormous, often with essentially the same products sold under different names.

Lingerie is a little different. The prices we can buy at in China compared with European distributor prices are considerably more attractive, at least from what I've seen. And there aren't nearly as many European distributors doing the same private-label thing.

Try selling lingerie in Sweden using a poor factory photograph from China, featuring a Chinese model wearing the product. The product itself might be perfectly good, but the photographs rarely fit the Scandinavian market or the brand you're trying to build. You really need your own models and photography.

And that gets expensive.

We learned this the hard way some years ago when we tried to jump on the women's fitness leggings trend. So many new Swedish fitness brands were doing extremely well. Great sales, great margins, making lots of money. We wanted in.

We were too late, though. And to be fair, we didn't exactly go all in either. It was a pretty half-assed entry into the market.

It didn't work out particularly well, but we did hire a very good Swedish photographer to shoot the collection, including video for social media.

The whole production cost us around SEK 300,000. For one collection.

The photographs were excellent. The business wasn't.

Now, being a photographer myself, the obvious solution would have been to shoot future collections myself. I have the equipment, I know how to use it, and I rather enjoy photography.

So back to lingerie. Why didn't we start sourcing it from China years ago?

Well, there was just one small problem.

My wife.

She drew a very clear line at me photographing half-naked women in lingerie. Even fitness clothing was a hard no. Apparently, that's where she draws the line. And I suppose I can understand that. :)

So there we were. A product category with attractive margins and somewhat higher barriers to entry, and one of those barriers was my own wife.

We never really pursued it.

Fast forward to April this year.

I was in Shanghai for the annual trade fair I attend for our business. It's not a lingerie fair, although there are always a few exhibitors selling that sort of thing.

One saleswoman was particularly persistent. She knew our company and really wanted me to sit down and look at what they had. I wasn't especially interested, but she wasn't giving up.

Long story short, I eventually sat down, had a conversation, and came home with some samples.

In June, I decided to place a small initial test order. Nothing particularly complicated. Panties, stockings, and a few other basic items. I deliberately picked products where the factory already had reasonably good product photographs, so we wouldn't need to arrange a photo shoot ourselves.

The first delivery arrived in September.

Naturally, I started experimenting with generating our own product photographs using AI. Because, of course I did.

Most of it worked surprisingly well. But some products were an absolute no-go. The AI models simply refused to generate the images. Even Elon Musk's Grok, which is usually a little less restrictive, accepted most of them but still refused a few.

Apparently, a close-up of a woman wearing an ordinary pair of lace panties is too much to ask. It's a pair of panties. Nothing particularly revealing, nothing unusual. Just a product photograph.

But no.

That got me thinking about something I'd tried six or eight months earlier with ComfyUI to solve the same problem. I'd experimented with taking a photograph of an actual lingerie product and putting it onto an AI-generated model.

The idea was simple enough. The results weren't particularly good. It was complicated, required a lot of fiddling, and certainly wasn't something I could see us using for actual product collections.

But six or eight months is a very long time in AI development these days.

So I tried again with a local model running on my DGX Spark without restrictions. And this time it just worked.

I can now take a photograph of the actual product, generate a Swedish or Scandinavian-looking model, and have the AI put the lingerie onto her. The results look remarkably realistic. What required a complicated ComfyUI workflow and produced mediocre results just a few months ago has become almost ridiculously easy.

Of course, I still have to make sure the generated garment actually matches what we're selling. But the results are so good now that I'm actually creating our own lingerie brand as I write this. We've already placed a second order, this time with our own branding.

I couldn't help thinking about that SEK 300,000 photography and video production. And all the collections we never pursued because the costs were too difficult to justify.

And then it occurred to me that I'd finally solved the wife problem.

I'm now "shooting" half-naked women in lingerie. I just do it on my computer instead of in a studio.

They're still half-naked, but they're synthetic women, not real ones. I'm just not standing in the same room with a camera. So technically, I'm following the rules.

My wife doesn't seem entirely convinced, though. Sometimes she looks over at my screen and says:

"You have way too much fun doing this, don't you?"

"No, no. It's just work," I tell her.

"Someone has to do it. And the margins are very good." :)