This used to be a nearly impossible task. The client sends over a handful of product shots. Clothes on hangers. Maybe a flat lay or two.
Then the brief lands: “We need something really dynamic for the campaign. Lifestyle feel, you know what I mean.”
Yup. They basically just asked you to do a photoshoot without a camera.
So that’s exactly what we’re going to do.
I’m working inside Freepik Spaces, which lets you build a visual pipeline using connected nodes. Each step feeds the next, kind of like an assembly line for your images.
I start with two inputs. First, a model. In this case, AI generated. There’s plenty of model stock photos out there, but finding a clean full-length shot that isn’t already cropped or part of a lifestyle scene? That’s a different story. The AI gave me exactly what I needed in about thirty seconds.
Second, the clothing the client sent. A flat lay of a summer outfit. Tank top, denim shorts, hat, sandals, sunglasses.
The first generator node takes both images and combines them. The prompt is pretty straightforward: dress the model in the outfit, include all the accessories, keep the exact proportions of the garments.
That last part is not optional. Without it, the AI will quietly “improve” things nobody asked it to improve. It’ll adjust a skirt length, shift a waistline, change a proportion. More on that in a minute.
And just like that, she’s dressed. One prompt. The outfit’s on, the hat made it, the sandals are there.
The AI did put the sunglasses on her face, and then gave her a second pair to hold. I guess it thought it was really bright outside.
I’m not thinking the sunglasses work, so a quick edit to remove those, and we’re ready to move on.
So we’ve got a model in the client’s clothing. Now let’s put her somewhere other than a plain background.
I connected four image generator nodes to create environments. Beach, outdoor cafe, city sidewalk, and park. Each prompt is basically one line: the setting and “full body visible.”
I’m keeping these loose on purpose. I don’t want four identical shots with a swapped background. I want the AI to read each environment and adjust the lighting and mood on its own. That’s what makes it feel like a real shoot instead of something assembled in a rush. Which, technically… it was.
This is one of those times giving AI creative freedom can work to your benefit. I said “can.” Doesn’t always.
Beach. Cafe. City. Park. Four shots that actually feel like different moments in a campaign.
One small thing: say “city sidewalk,” not “city street.” Unless you want your model doing a lifestyle shoot in the middle of traffic. The AI takes directions literally.
Now let’s see how easy it is to apply this same workflow to a different model and outfit.
This time the client gave us even less. One pink dress. No shoes. No accessories. Just a dress.
Different model. Different clothes. Same workflow.
The prompt tells the AI to dress her in the dress and add appropriate summer sandals, because the client didn’t send any shoes. So we’re letting the AI make a styling call. Used to be you’d spend twenty minutes searching stock libraries for the right pair. Now we just ask.
First try, the dress came out too long. The AI lengthened it. This is that proportion thing I mentioned. A flat lay sits differently than fabric on a person, and the AI decided to interpret the length its own way.
So I used the edit tool to shorten the dress. And the AI gave me… her back.
The dress is on backwards. For some reason the AI thinks her back is her front.
This is a new one.
Two more tries. Still backwards. I ended up adding “model facing camera” to the prompt, which is not a sentence I ever expected to type.
That helped, but I then had to tell it the waist was sitting too high. Took several rounds to land on a version that actually looked right.
But we got there. Almost. The fit looks natural. The sandals work. The proportions are where they should be.
But at full resolution, the AI quietly dropped the three buttons from the neckline. At web sizes and social, you’d never spot it. But if this goes to print or a close-up, that’s a different conversation. It’s the full-res file that tells you what the AI actually got right and what it missed.
For the lifestyle shots, I ran the same workflow. Beach. Cafe. City. Park. Different model, different dress, different energy, but every shot still feels like it belongs in the same campaign. That’s the system doing its job.
And then, because the brief said “dynamic,” I gave a couple of them motion for social and YouTube. I connected the beach and park stills to video generators running Kling 3 and generated five-second clips.
Walking on the beach. Hair moving in the breeze. And yeah, I checked: she’s leaving footprints in the sand. This is the small stuff that keeps it believable. The kind of detail your eye catches even if your brain doesn’t name it.
Park scene, same thing. Natural stride, soft light. Like someone actually planned it.
One thing worth knowing. I ran this whole thing in stages. Model comp first. Check it. Fix it if you need to. Then environments. Check those. Then video.
You could hit go on everything at once. But if the first step comes back wrong, you’ve just burned credits regenerating six visuals downstream that were built on a bad foundation. That’s not a workflow, that’s an expensive way to be disappointed.
So. Two models. Two outfits. Eight lifestyle shots. Four video clips.
All from a couple of flat lays and a brief that said “make it dynamic.” No camera. No crew. No awkward small talk with the client on set.
Done in about 30 minutes.
The images went from “clothes on a table” to “campaign that tells a story.” That’s the shift. Not just putting an outfit on a model, but building the scenes, the lighting, the mood that make a viewer feel something about the product. That’s what the client was actually asking for when they said “lifestyle feel.” They just didn’t know how to say it.
