The same character in every scene.




These frames are from the reference set for the No Plans Today character. Whenever I make a new scene, I compare the face, hair, styling and light with this set.


I direct AI to make UGC characters, reels, campaign films, social posts and product visuals for brands.

I build a character for your brand and check it against its references in every new scene.

Short videos I cut to catch attention in the first second.

Films I build scene by scene, like the knitwear film for Karaca.

Posts I plan as one set for your feed.

Your product, shown in scenes it was never photographed in.






Real products, AI scenes.




These frames are from the reference set for the No Plans Today character. Whenever I make a new scene, I compare the face, hair, styling and light with this set.

The No Plans Today bottle among the clouds. Before I deliver, I compare its shape, label and colour with the real bottle.
YOS posts from October and November. I made the carousels and single posts in the sizes each channel uses.

It starts in black and white. Each thing the character touches turns to colour: a drawing, a school bag with a cat on it, a water bottle. I cut it for vertical screens so each scene reads in a few seconds.

Karaca's knitwear film and the YOS characters both went through these steps.
First I pin down what the piece has to do: where it will run, who will see it, what must stay the same. I write that down before anything is generated.
Example direction Vertical film for a knitwear launch. Warm, quiet, the stitch has to read.
I gather the brand's own material: product photos, past posts, colours, light. I also note what shouldn't be repeated. From then on I check every output against this set.
Example direction Take colour from the autumn product shots. Stay away from last season's studio light.
I write the directions and generate variations in rounds. After each round I go back to the brief to see what's working.
Example direction Same character, café terrace, late afternoon, soft side light, handheld feel.
I throw away most of what comes out. I keep the frames where the product, the character and the light are right, then go through the details one by one.
Example direction Keep the frames where the label reads clearly. Drop any hand with odd fingers.
I finish the chosen pieces as a designer: retouching, type, the edit and the sizes each channel needs. You get files you can post as they are.
Example direction Cut a vertical and a square version. Leave room for captions.
Short answers. If your question isn't here, write to me directly. Contact page
UGC-style characters, short reels and edits, campaign films, social posts and product visuals. The work page has examples of each.
I build a reference set from your own material and check every output against it. I don't train my own model. Before delivery I go through details like product shape, labels, colours and the character's face one by one.
The work on this page comes in these formats: 9:16 vertical, 4:5 feed post, 16:9 widescreen. In the brief, I work out with you which size each channel needs.
Everything on this page is work I made with AI, and I don't hide it. How to label it on your own channels is something you and I agree on in the brief.
No. Some jobs are mostly AI, some mix it with real photos and footage, and some don't use it at all. I work out with you which one fits your job while going through the brief.
Write to me through the contact page with what you need, where it will run and a few references you like. I'll reply with the questions I need answered.