UGC ad workflow examples: turn product photos into ads in one canvas

Three UGC ad workflows in Picsart Flow each fit on a single canvas, and each one takes a photograph of your product. What they hand back is where they split.

One turns that photo into a creator selfie, a styled hook and three vertical clips. One reads the label, writes the script from what it finds, and returns the ad as separate beats you can cut together. One fills out a whole placement set, several lifestyle videos plus static ad formats.

The difference between them is not the product or the aesthetic. It is how much of the ad each one finishes for you. These UGC ad examples run from a single hook still up to a full set of placements, so the question worth asking is which part of the job you want handled.

UGC hook text: one product photo, one creator selfie

One can photographed on a beach becomes a creator on a patio.

Create Engaging Lemon Sparkling Water UGC Content is built for a vertical 9:16 feed, and its video prompt asks for eight seconds. Its shape is a single straight line of six image nodes that splits three ways at the very end.

1. Product photo in. One ordinary shot: a lemon sparkling water can, photographed on a beach.
2. Background removal. The can gets cut out and the beach is discarded. Everything after this point is built around a floating cutout, never around the original scene.
3. Scene generation. A text node feeds the cutout into a new image: a “realistic UGC-style selfie photo of a male content creator holding the uploaded product close to the camera,” at “smartphone front-camera perspective.” The person, the patio and the pose all arrive here, from nothing.
4. Retouch, composition locked. The render goes straight back in. This prompt’s first instruction is to “preserve the exact original image and composition,” and only then to improve facial detail “while maintaining natural skin texture” and brighten the product “to make it stand out as the hero object.” Nothing is allowed to move. Only quality changes.
5. Hook text. A plain text node adds the line: “Obsessed with lemon sparkling water.”
6. Hook styling. A second, separate node dresses it: “a compact rounded highlight behind each line,” at “approximately 30% opacity,” “tightly wrapped to the text with minimal padding,” and explicitly “avoid large banners or oversized backgrounds.”

Then the split. That one finished still feeds three video nodes at once, and each branch pairs it with a different text node. One carries a spoken line about zero calories and staying hydrated. Another carries a different line, about crisp bubbles and citrus. The third carries no dialogue at all, just motion direction.

That motion prompt is the most instructive thing on the canvas, because it is mostly refusals. It permits “gentle blinking, soft breathing, slight head movement, and minimal handheld smartphone camera motion,” then closes the door: “no scene changes, dramatic motion, or exaggerated animation.” Composition, product placement, hook text and lighting all have to survive untouched.

The economics of the fan-out are the real lesson. One still is generated once and reused three times. Only the words beside it change, which is why testing a weak ugc hook here costs almost nothing.

What it hands you: a styled hook still, three eight-second vertical clips, and two spoken creator lines to choose between.

UGC script: benefits copy written before any footage

This one reads your label first, then writes the ad from what it finds.

Create Authentic UGC Ad for Skincare Serum is shaped the opposite way round: it opens wide, narrows to a single record, then fans out hard. Nothing renders until the copy exists.

1. Two readers in parallel. The uploaded product shot feeds two nodes simultaneously. One runs OCR across the label text. The other reads the bottle visually, for form, materials and finish.
2. The reads merge. Both outputs collapse into one structured product record: category and subtype, hero ingredients, the claims printed on the packaging, materials, and estimates of target audience and price tier. Each field carries its own confidence score, so the workflow is explicit about what it is guessing.
3. Benefits copy. That record feeds a node whose only job is turning extracted facts into selling language.
4. The script. Four labelled beats come out: a hook, two body sections and a call to action. The record and the benefits copy both feed it, so the script argues from the label rather than from the category.
5. A model reference, generated separately. One full-length image of the person, made from its own prompt. This is the identity anchor, and it is why seven clips can show the same woman rather than seven similar ones.
6. The fan-out. The script splits into a voiceover on an audio node, and into four beat clips named for their role: Video Hook, Video Body 1, Video Body 2, Video CTA. Three B-roll clips render alongside, one a dedicated product close-up.

The reading pass at stage one is what separates this from the lemon chain. Both end up with something spoken, but the lemon workflow writes a creator line, one testimonial sentence to camera. This ugc script restates what the packaging already says, in a deliberate order. One gives you a person saying they like something. The other gives you a reason to believe them.

The beats arriving separately matters just as much. What lands is not a finished video, it is a cut waiting to be assembled, which is exactly why B-roll ships with it.

Reach for this when the benefit is invisible in a photograph. A serum or a supplement needs the copy layer. A can of sparkling water mostly does not.

What it hands you: a structured product record, benefits copy, a four-beat script, a voiceover, a model reference, a still, and seven clips including three B-roll shots.

Product video ads: several placements from one brief

One product, several times of day, video and static together.

Create Engaging GlowSip Product Video Ads stops thinking about single assets. Its outputs are vertical, and it carries the most model firepower of the three: Sora 2 Pro and Veo 3.1 for video, with Flux 2 Max and Ideogram V3 alongside.

Its shape is the widest of the three and the shallowest. Rather than a long chain, a single input node fans straight out into a bank of parallel outputs that share no intermediate stages.

1. One input, taking product footage or a written description.
2. A row of video branches, each rendering the product in its own lifestyle context: a close, unhurried self-care shot; a gym scene with the product beside a kit bag and headphones; a woman working at a laptop.
3. A row of image branches running from the same input. One produces a clean headline ad, the product standing at a waterline that cuts across the frame with “hydration that actually fits your routine” set underneath. Another produces a split comparison creative.

Because the branches are independent, they run and re-run separately. A context that misses can be regenerated without touching the ones that worked, which is the practical advantage of a wide shallow graph over a long chain.

The statics are the part most video workflows skip. Campaigns need a video for the feed and a static for everywhere video will not run, normally two separate jobs with a real risk they drift apart in tone. One brief means they read as one campaign.

What it hands you: several lifestyle video contexts plus static ad formats, all from one product and one brief.

How to run a UGC ad workflow

1. Open one of the three templates above, or browse the library at Picsart Flow.
2. Sign in and clone it into the workflow editor so the original template stays intact.
3. Replace the sample product image with your own, shot clean and evenly lit.
4. Edit the text nodes you want to change before running anything.
5. Run the workflow and let every stage finish.
6. Export what you want to keep. Exporting usually requires signing in.

Start building

Pick the workflow that covers the layer you are missing, not the one with the nicest sample output. If the hook is your weak point, start with the lemon workflow and rewrite its one line. If the product needs explaining, start with the script.

Open Picsart Flow, clone a template, and swap in your own product photo.