Guide
Footage wins a tie. A photograph that shows the thing wins outright
When a clip and a photo show the same thing equally well, use the clip, because movement holds the eye in a feed and a frozen frame reads as a slideshow. When the photo shows the thing clearly and the clip only nearly does, use the photo, and give it a slow move so it never sits still. Both halves matter, and getting the ordering wrong is easier than it sounds.
- Slots that came back as photographs before we looked
- 83 of 89
- Clips against stills the libraries offered
- 407 v 409
- Photographs that should ever sit frozen
- 0
Movement is worth something specific, and it is not cinematic quality
A moving shot gives the eye a reason to stay for another second. In a feed where the alternative is a thumb, that second is the whole game.
A still does not do that on its own. Held on screen without motion it announces that the video has run out of material, and viewers read it as cheap before they could tell you why.
So at equal evidence, footage wins. Same subject, same clarity, same fit for the frame: take the clip.
That is a preference, not a rule, and the difference matters. Preferences decide ties. They should never decide arguments.
A photograph that shows the thing beats a clip that nearly does
The second half of the rule is the one people drop. If the photo unmistakably shows what is being talked about and the clip is only in the right neighbourhood, the photo is the better cutaway by a distance.
Relevance is worth more than motion because relevance is what the cutaway is for. A moving shot of the wrong thing is still the wrong thing, and now it is the wrong thing with momentum.
This is measurable if you want it to be. We hold the motion preference to a weight smaller than one band of relevance, so a clip can never outrank a photo that is meaningfully better at showing the subject.
In practice that means the ordering is: does it show the thing, does it fit the frame, is it moving. In that order, every time.
Which one to use, in the four cases that actually come up
Only the first row is a tie, and only ties are decided by movement. The rest are decided by what is in the picture.
| What you have | What to use | |
|---|---|---|
| Both show it equally well | A tie | The clip |
| Photo shows it, clip is vague | Not a tie | The photo, with a move on it |
| Clip is horizontal, photo is vertical | Cropping costs two thirds | The photo |
| Neither really shows it | Not a choice | Neither. Stay on the speaker. |
This is the whole editor
Highlight a phrase and a clip lands on those exact words. No timeline, no keyframes, no layers.
How a preference for motion turns into all photographs by accident
This is worth telling because we shipped it. Across four recent production edits, 89 slots were filled with cutaway footage and 83 of them were photographs.
The libraries were not the reason. They returned 407 clips against 409 stills across the same batch, which is a dead heat, and the filters removed almost nothing.
The ranking was the reason. The search terms were one and two words long, and against a one word query the relevance score maxes out for every hit, clip and photo alike. With relevance tied, the ordering fell to the small tiebreaks.
One of those tiebreaks was a penalty for a clip running much longer than the slot it fills, and by construction only a video can ever pay it. So every clip landed a hair below an otherwise identical photograph, photographs filled the shortlist, and the stage that picks the final clip never saw a moving frame to choose from.
The fix was to reserve half the shortlist for moving footage so both kinds get heard, and to make sure running long is never a reason to prefer a still. Neither change forces a clip onto the screen: a weak one still loses to a photo that fits.
Whatever you choose, a still must never sit frozen
A photograph on screen gets a slow push or a slow drift, always. No exceptions, including on the photos you supply yourself.
The move does not need to be big. A few percent of scale over two seconds is enough to keep the frame alive, and anything more starts to read as an effect.
Every still that reaches a Cutroom cut gets one automatically, because a photograph with no motion on it is the clearest tell that a video was assembled rather than edited.
Direction is worth varying too. A dozen cutaways all pushing in the same way is its own kind of monotony, and alternating between pushing in and pulling out costs nothing.
Questions people ask
- Is stock footage always better than stock photography?
- No. It wins when both show the subject equally well. A photograph that clearly shows the thing being discussed beats a clip that is only loosely related, every time.
- Why do AI editors fill everything with photographs?
- Usually because the ranking has a tiebreak only videos can lose. Photo libraries are also larger and better tagged, so on a vague query there are simply more of them near the top.
- What is the right amount of movement on a still?
- Small. A few percent of scale across the time it is on screen. If a viewer can describe the move, it is too much.
- Should I use my own photos or stock?
- Yours, wherever they exist. Your own material is specific in a way no library is, and specificity is what makes a cutaway worth the seconds it costs.
Show the thing first, fit the frame second, move third. That is the ordering we score against, and the reason a photograph in one of our cuts is never sitting still.