Guides / Ads

From one product photo to a video ad and a poster set

You have a product photo on a white background. That is enough to make the ad and the posters, in that order, without a second photo.

  • 8 min read
  • Presets: Social and poster pack, Hands-on product ad
Styled stillThe adPosters

What you make, in order. Tap one to jump to its step.

To turn one product photo into a video ad and a poster set, first make a styled still with the photo attached as a reference, then animate that still for the ad and re-run the same reference at three sizes with your headline for the posters. Three tools, in that order: the image console, the video console, the image console again. This page shows the whole run on a pair of fictional boxing gloves, and every picture and clip here is the output of the step next to it.

The product is invented so we could show it: crimson leather gloves with gold laces and no brand. Your starting photo is the one you already have.

Before you start

  • One product photo on a plain background, sharp, the whole product in frame. Ours is square; any shape works as a reference.
  • One line that describes the product the way a buyer would: material, colours, the details that must not change. It goes into every prompt.
  • One headline of two to six words for the posters, and nothing else printed. The fewer words, the fewer chances to misspell.
  • The sizes you need. We made 9:16 for Stories, 4:5 for a feed and 16:9 for a banner.
A clean product photo of a pair of crimson red leather boxing gloves with gold laces and gold wrist straps on a white background
The starting photo. Generated with GPT Image 2.5 Flare so we could show the method on a product that does not exist. Yours is a real photo.

Restyle the photo into the ad's look

Open the image console, press Add reference image and add the photo. Describe the product exactly as it is in the photo, then the place, the light and the lens. Say what must stay: shape, colour, laces, straps. This still becomes the first frame of the ad, so make it 16:9 (or 9:16 for a vertical ad).

The image console on a desktop: the gloves photo uploaded and shown as Image 1 under References, one image, 16:9 selected, native quality
The image console on a desktop. The photo rides as Image 1; the aspect ratio is 16:9 because this still becomes the ad's first frame.
The same form on a phone: the uploaded photo as Image 1 and the aspect ratio chips
The same form on a phone.
The styled still (GPT Image 2.5 Flare, 16:9, one reference)
Premium sports advert still, 16:9. The same pair of crimson red leather boxing gloves with gold laces and gold wrist straps from the reference photo, keeping their exact shape, colour, laces and straps, now hanging by their tied laces from the top rope of a boxing ring in a dark empty arena. Thick haze, a hard red gel from the left and a cold blue gel from the right, a white rim light tracing the leather, dust drifting in the beams, deep black background, strong contrast, 50mm lens, shallow depth of field, photoreal. No text, no letters, no logos, no badges, no watermark.

The photo

The product photo: crimson boxing gloves on white

The styled still

The same gloves hanging from a ring rope under red and blue light with haze
One run. The laces, the gold straps and the red held; the arena, the gels and the haze are new.

First run

The first styled still: the gloves hanging from the top rope, red light left, blue right

Same prompt, run again

The same prompt run again on the live console: the gloves hang closer together with the straps at the top
The same photo and prompt, twice. The gloves, laces and straps held both times; the hang and the ring changed. Expect the second run to differ, and pick.

Animate the still into the ad

Open the video console, choose Animate an image, press Add start frame and pick the styled still from your Library with Pick an asset (or upload it). Choose Kling v3 (Kling 3.0 in the menu) with the Choose model button, set 6 seconds, and write the prompt as motion: what sways, what drifts, where the camera goes, what the sound is. The price sits on the Generate button. This is the same path the drink ad guide runs shot by shot.

The ad (Kling v3, 6 s, from the styled still)
Premium sports advert, one continuous shot. The two crimson boxing gloves hang from the top rope and sway very slowly, the gold laces catching the light; dust drifts through the red and blue beams, the haze rolls gently, the camera pushes in slightly toward the gloves and stops while the gloves stay fully in frame. Nobody enters the frame. The gloves keep exactly the same shape, colour, laces and gold straps. Photoreal, strong contrast. No text, no letters, no logos, no badges, no watermark. Sound: a low arena rumble, a distant crowd murmur and one soft ring bell, no music, no voice.
The gloves swaying gently on the rope while haze drifts through red and blue light and the camera eases in
The ad. Kling v3, 6 seconds, one take, with the sound it rendered. The gloves sway on the laces and stay fully in frame; the laces and gold straps hold. Add your text in Studio if the placement needs a line.

Make the poster set at three sizes

Back in the image console, keep the photo as the reference and run the styled-still prompt three times with the headline added and the aspect ratio changed each time. Put the headline where the size wants it: top on 9:16, bottom on 4:5, left on 16:9. Say the headline is the only text in the picture.

The Stories poster (GPT Image 2.5 Flare, 9:16, one reference)
Premium sports poster, 9:16. The crimson red leather boxing gloves with gold laces and gold wrist straps from the reference photo, kept exactly as they are (shape, colour, laces, straps), hanging by their tied laces from the top rope of a boxing ring in a dark empty arena, thick haze, a hard red gel from the left, a cold blue gel from the right, a white rim light on the leather, dust in the beams. A bold headline in clean white condensed sans-serif capitals across the top quarter of the poster, in two lines: TWELVE ROUNDS. / ONE PAIR. The headline is the only text in the picture: no other words, no logo, no brand mark, no watermark.
A 9:16 poster: the gloves hanging from the rope, the headline Twelve rounds. One pair. across the top
9:16, for Stories. The headline across the top.
A 4:5 poster: the gloves hanging from the rope, the headline across the bottom
4:5, for a feed. The headline across the bottom.
A 16:9 banner: the headline on the left, the gloves hanging on the right
16:9, for a banner. The headline on the left, the gloves on the right.

Tip:If the words come back wrong, use the poster pack. All three posters here spelled the headline correctly, and we checked each at full size. If yours do not, the Social and poster pack preset generates the picture and sets the words as real text on top, so they cannot be misspelled.

What goes wrong

MistakeWhy it hurtsFix
Animating the product photo itselfA white-background photo gives the model a white world to animate; the ad looks like a catalogueRestyle first, animate the styled still
Describing the scene but not the productThe model redraws the product to fit the sceneDescribe the product from the photo first, then say what must stay
A long headlineEvery extra word is a chance to misspell or crowd the productTwo to six words, and only the headline
Reframing the finished ad for a new sizeReframe works on stills, not clipsMake the still and the ad at the size you need
Trusting the words at thumbnail sizeA wrong letter hides at small sizeOpen each poster at full size before you post

Which tool for which job

JobUseNotes
The look, from the photoImage console, GPT Image 2.5 Flare, one referenceUp to four references; the photo is enough
The adVideo console, Animate an image, Kling v33 to 15 seconds; keeps a hanging product's sway believable
Posters with words drawn inImage console, same reference, headline in the promptCheck the spelling at full size
Posters with words set as textSocial and poster packOne headline, four sizes, real text on top
A hands-on demo clip insteadHands-on product adPhone-footage look, no presenter
Bigger files for printImage console, 2K or 4K qualityPriced per picture; the console shows it

Where you still regenerate

  • Small details drift. Our laces stayed laced and the straps stayed gold, but count stitches and buckles on yours before you post.
  • The camera can overdo it. Ask for a push-in that stops and say the product stays fully in frame; keep a second take in the budget.
  • Text is a gamble the model usually wins, not always. Three of three posters were right here; the preset with real text is the safe path.
  • One size per render. A 9:16 ad is a second animation, not a crop.

What it costs

This set is two stills, one clip and three posters at the standard picture size. Prices are on the Generate button; a failed render is refunded.

  • GPT Image 2.5 FlareEvery plan

    Per image

    Native 8 credits · 2K 21 credits · 4K 58 credits

  • Kling v3Pro and up

    Per 5s clip (audio on by default)

    720p 35 credits · 1080p 47 credits · 4K 117 credits

Read live from the catalog when this page loads. Every price is shown before you generate. See every rate.

Checklist before you post

  • The product matches the photo: shape, colour, every part you would be asked about.
  • The headline is spelled right on every size, checked at full size.
  • Nothing else is printed in the picture.
  • The ad keeps the product in frame for its whole length.
  • Each size was made at that size, not cropped from another.

Try it yourself

Download the gloves photo, open the image console, add it with Add reference image, paste the styled-still prompt and generate at 16:9. Then animate the result in the video console with the ad prompt. We ran the still recipe on the live console and show both runs above; the ad clip came from the same request sent through the API.

The inputs

  • The product photo (the reference)JPEG photo · 2048x2048 · 727 KBDownload
  • The styled still (the start frame)JPEG photo · 2560x1440 · 462 KBDownload
Both are generated, so you can use them to practise.

Questions and answers

Is one photo really enough?
For a product with one good side, yes. The model shows the sides the photo shows; if the ad needs the back or the sole, add that photo as a second reference.
Does it work with a phone photo?
Yes. Plain background, sharp, the whole product in frame. The guide on [product photos from a phone photo](/guides/product-photos-from-a-phone-photo) covers the photo itself.
Will the model spell my headline correctly?
Usually, and it did on all three posters here, but not always. Check at full size, or use the Social and poster pack preset, which sets the words as real text.
Can I get a 9:16 version of the ad?
Make the styled still at 9:16 and animate that. Reframe changes the size of a still, not a clip.
Can I put my logo on the gloves?
Keep it off the picture and add it as a layer afterwards. A logo drawn by the model changes from picture to picture.
Can a person hold the product?
Yes, describe the hands and the pose in the still prompt. We kept people out of this example on purpose, so we have not measured it here.

Start from the photo you already have

Open the image console

Product images

One phone photo to a full shop set

Turn one clear phone photo into a studio hero, a seasonal version, a product page banner and a short clip, with the preset for each step and what the set costs.

· 6 min read

Ads

A drink ad in fifteen seconds, from stills you generate

Make one styled product still, animate three shots from it, cut them to a sound bed in Studio and type the end line. The appetite cues that worked on a fictional tonic, and the two that did not.

· 10 min read