Visual DNA
Brand visual DNA: the look brief for an AI film and its photos
For the brand or creative lead who wants an AI-made film to look like the brand's best work, and wants that look written down once, for the film and the campaign photos.
The short answer
A brand's visual DNA is the set of visual habits its best images share: the palette in hex codes, the light, the lens and framing, texture and grain, styling and casting. A film adds the habits of motion: how the camera moves, how long shots hold, the frame rate, the grade and the sound. Written as lines someone can check, the DNA becomes the brief that every generated shot and every campaign photo is measured against.
Definitions
Visual DNA, brand identity, guidelines and the brand world
The four terms get used as if they were one thing. They answer different questions, and an AI production needs all four answered before the first shot:
| Document | What it answers | Example line |
|---|---|---|
| Brand identity | Who the brand is and how it signs its name | Logo, name, voice, values |
| Brand guidelines | The rules for using the marks | Clear space around the logo, the type scale, color use |
| Visual DNA | How the brand’s images look, measured | “Warm key light from camera left, blacks never fully black” |
| Brand world | Where this campaign happens, and who is in it | “A salt flat at dawn, one runner, a red jacket” |
Visual DNA sits between the rules and the campaign. It carries from one campaign to the next, while each campaign builds its own world on top of it. How to write that world is in brand world building for AI film and photos.
The shared fields
What a film and its photos share
The fields that describe a still image carry into film unchanged, because every frame of a film is a still:
- Palette, as five to seven hex codes, each with a job: background, shadow, skin, accent, the product's own color.
- Light: direction, hard or soft, warm or cool, and how dark the shadows are allowed to go.
- Lens and framing: the focal lengths the brand uses for wides and faces, and where the subject sits in the frame.
- Texture: grain, sharpness, halation around highlights, how clean or worn surfaces look.
- Styling and casting: wardrobe, props, hair, the age range and look of the people.
For stills, the method is set out in Overs' guide to a brand's visual DNA, which lists ten fields and a template. The rest of this page covers what a film needs on top.
The film fields
What a film adds to the DNA
A photo DNA says nothing about time, and a film is made of it. A generated shot can get the palette right and still feel wrong because it moves, holds or sounds like another brand.
| Field | Example line | How to check it |
|---|---|---|
| Camera movement | Handheld when following a person; one locked-off wide per scene | Label every shot of your best film as static, push, pan, handheld or aerial, and count |
| Shot length | Two to three seconds in action, five or more in quiet moments | Divide the film’s running time by its number of shots |
| Frame rate and blur | 24 frames a second, natural motion blur, no stutter on fast pans | Check the export settings and watch fast moves at full size |
| Transitions | Hard cuts only; one match cut per film at most | Scrub the edit and list every transition |
| Grade over the sequence | Lifted blacks, warm highlights, skin never orange | Put frames from shot 1 and shot 30 side by side |
| Sound | Breath and footsteps first; music enters at the midpoint | Listen with the picture off |
| Type on screen | One typeface, lower third, on for at least two seconds | Pause on every title card |
Shot length matters more in AI film than in live action. Generated shots are short: Veo 3.1 makes clips of 4, 6 or 8 seconds, so the rhythm of the cut is a choice the studio makes on purpose. A brand that cuts slowly should not get a film that cuts fast because the tool made short clips.
Method
How to pull the visual DNA from your own films
- Pick the evidence. Your three best films and ten to twenty of your best stills, the ones the brand team would hold up as “us”. Leave out work you are trying to move away from.
- Pull frames. One frame a second is enough. With the free tool ffmpeg:
ffmpeg -i film.mp4 -vf fps=1 frames/%04d.png - Measure the palette. Use an eyedropper on the recurring colors and write each one as a hex code with its job. A color that shows up in only one film belongs to that campaign; leave it out.
- Count the shots. The free tool PySceneDetect lists every cut:
scenedetect -i film.mp4 detect-content list-scenes. Running time divided by the number of shots gives the average shot length. - Label the movement. Tag each shot static, push, pan, handheld or aerial, and note which moves the brand never uses.
- Write ten to fifteen lines. Each line has to be checkable by someone who was not in the room, like “one light source, from camera left”. Words like “premium” fail that test.
- Test it. Make three shots and three stills from the lines alone, then check every line. Any line the output breaks is either unclear or wrong. Rewrite it before production starts.
Why it matters in AI production
What happens to a film with no DNA written down
A generator draws every shot from scratch. Nothing carries from shot 12 to shot 13 unless it is in the prompt or the reference images. So a film made without a written DNA drifts in small steps: the blacks lift a little, the palette warms, a lens gets wider, and by the end the film looks like the model's idea of a brand film rather than yours.
Every frame Ruminate X delivers is made with generative AI. There are no film crews, sets or location shoots. At Ruminate X, the look is fixed in the brand-world stage, before script and boards, and each generated shot is checked against its board and generated again until it holds. The stage-by-stage process is on how we make an AI brand film. The grade then does the last part: one color decision applied to every shot, so frames that were generated days apart match on screen.
The photo side
Where Overs comes in
A campaign rarely ends at the film. The key visual, the social posts and the store photos need the same DNA, and that is the job Overs was built for. Overs is the sister product from the same founder: an AI tool that makes product and campaign photos that match a brand, without a photo shoot. It was built on the process 100creatives used to make hundreds of photos a week for hundreds of brands.
Give Overs a brand's website, the photos it already has and, if there is one, its guidelines PDF, and it writes the visual DNA for you: colors with hex codes, light, lens, grain and styling. You fix anything it gets wrong once, and every campaign after that reuses it. Its Brand Watch reads the brand's recent Instagram posts so the DNA keeps up with how the brand looks now: when you ask on the free plan (at most once a month), weekly on Pro, daily on Team. Brand rules are applied last, so they have the final say over every photo. You see the shot plan and an estimated cost before anything renders.
Overs makes stills and no video. For each photo it can write a motion prompt for a video model such as Seedance or Veo, to run with that photo as the first frame of a shot. The free plan makes 40 photos a month, and the AI costs a few cents a photo on your own OpenRouter key, with no markup from Overs. Start on Overs' free plan at www.overs.studio and see your brand's DNA written out before you spend anything on a render.
Should Ruminate X build the look for your film?
Hire Ruminate X if
- A brand with strong photography and no film language yet, that wants its first AI brand film to look like its best stills.
- A marketing lead whose past films each looked different, who wants one look carried through the next film and its cutdowns.
- An agency producer who needs the DNA written, tested and held across a whole AI commercial.
Hire someone else if
- You need a new logo, name or typeface: hire a branding agency first, because visual DNA builds on an identity that already exists.
- You only need campaign photos in your existing look: a photographer, or Overs for AI-made photos, costs less than a film studio.
- Your film is interviews with real staff or customers: a documentary crew fits better, and their look is set on the day.
Questions buyers ask
What is visual DNA?
A brand's visual DNA is the set of visual habits that repeat across its best images: the palette written as hex codes, the direction and hardness of the light, the lenses and framing, texture and grain, styling and casting. For a brand film it also covers motion: how the camera moves, how long shots hold, the frame rate, the grade and the sound. It is written as lines someone can check, such as "light from camera left, soft, slightly warm".
What is the difference between brand DNA and brand identity?
Brand identity is who the brand says it is and the marks that say it: the name, logo, type, voice and values. Visual DNA is narrower and measurable: how the brand's photos and films look, from the palette to the light to the pace of the cut. Brand guidelines usually cover the identity well and say little about how images are lit, framed or moved, which is the gap visual DNA fills.
How do you find a brand's visual DNA?
Take the images and films the brand is proudest of, pull frames from the films, and write down what they share: the recurring colors (measured with an eyedropper as hex codes), where the light comes from, the lenses, how the camera moves and how long each shot lasts. Keep only lines that most of the images obey, then test them by making a few shots and checking each line.
Does an AI brand film need different guidelines than photos?
It needs more of them. The photo fields (palette, light, lens, texture, styling, casting) carry over as they are. A film adds camera movement, shot length and cutting rhythm, frame rate, transitions, the grade across a sequence, sound and the timing of type on screen. Without those lines, each generated shot drifts toward the look the model defaults to.
Can AI keep brand colors exact in a film?
Not on its own. Image and video models do not measure color, so a hex code in a prompt is a hint rather than a setting. Studios show the color on the reference images, check frames with an eyedropper against the brand's hex codes, and correct the rest in the grade, where one color decision is applied to every shot of the film.
When is the look fixed in an AI film production?
Before anything is generated. In the Ruminate X pipeline the palette, light, texture, camera language and casting are fixed in the brand-world stage, ahead of script and boards, so the last shot matches the first. Each generated shot is then checked against its board and generated again until it holds, and the grade ties the shots together.
Show us your best work
Send the films and photos your brand is proudest of, and what the next film has to do.