How Do You Make a Before-and-After Video With AI From Two Photos?
Turn two photos into a before-and-after clip: framing and light rules, transitions, the models that actually end on the after photo, and a six-photo timeline.
A before-and-after video is two photos and one transition: the first image opens the clip, the last image closes it, and the model fills the seconds between. In VdoBloom it uses Seedance 2 or Seedance 2.5, and a 5-second 720p clip costs 82 or 126 credits. The rule that decides whether it works is simple: only some models actually land on the after photo, so pick a model that does.
Updated October 2026.
The clip is not proof of anything. It is a transition between two photographs you supplied, and it should be labelled that way. What it can do is make a change legible in two seconds, which is why the format travels in renovation, styling, cleaning and product work.
What a before-and-after clip is: two photos and one transition
The tool is narrower than a general image-to-video model and that is the point. You upload a start photo and an end photo, choose a look, and generate the motion between them. There is no prompt to write the scene, because the scene is in the two photographs.
The output is a short vertical-first clip. The tool offers 9:16, 1:1, 16:9, 3:4 and 4:3, with 9:16 the natural choice for Reels, Shorts and TikTok. The presets are skincare, fitness, hair, home, product results and custom, and each one adds a style instruction for the transition. Custom is for prompts such as a specific room change or a product swap.
If you need more than one change, the pair mode is not the tool. Use the chain mode in Keyframe Morph, covered further down.
Photos that work: same framing, same light
The transition is only as good as the pair. Four rules cover most of it.
- Same camera position. Put the phone or camera in the same place for both shots. A tripod mark on the floor is enough. If the angle moves, the model has to invent a camera move and the clip becomes a different shot rather than a transition.
- Same framing. Crop both photos to the same aspect ratio before upload. A wide before and a tight after forces an automatic reframe that usually crops the subject.
- Similar light direction. Morning on one side and evening on the other makes the colour change look like a different day. That is fine for a story, wrong for a comparison.
- One clear change. A renovated room, a finished haircut, a cleaned surface, a rebuilt product. If three things changed, the viewer cannot tell what the clip is about.
For a fitness or body-comparison idea, the pair has to be two photos of the same person taken by that person, and the caption should say what the clip is: a transition between two photos, not clinical evidence and not a health claim. Do not publish it as proof of a programme you did not run.
Choose the transition and a model that actually lands on the after photo
Not every model honours a supplied last frame. In the Seedance family, Seedance 2 and Seedance 2.5 end on the after photo; Seedance 2 Fast and Seedance 2 Mini ignore the last frame, so the clip never arrives and the result looks like a slow pan of the first image. The tool only offers the two models that work, which is the simplest way to avoid the mistake.
| Model | Honours the end photo | 5-second 720p price |
|---|---|---|
| Seedance 2 | Yes | 82 credits |
| Seedance 2.5 | Yes | 126 credits |
| Seedance 2 Fast | No | 50 credits |
| Seedance 2 Mini | No | Cheapest tier |
The Fast and Mini rows are there to explain the price you might see elsewhere in the app, not as options here. A 50-credit Fast clip is cheaper because it is a different job. For this tool, the choice is Seedance 2 at 82 credits or Seedance 2.5 at 126, and the difference is mostly how much motion and how long a transition the model will hold. The exact price is shown before you render, and a failed render is refunded.
Five uses, each with the pair it needs
- Renovation. The same room, the same corner, before and after the work. Shoot the second photo from the tripod mark. The home preset adds a cinematic reveal. For more local-service ideas, see AI video for home services.
- Makeover or hair. The same face, the same background, the same light, before and after the cut or colour. Do not mix a bathroom selfie with a salon photo.
- Cleaning. A stained surface and the same surface clean, framed the same way. This is the easiest pair to shoot because the camera can stay fixed between the two exposures.
- Fitness progress. Two photos of the same person in the same pose and light, with a factual caption. It shows a transition between two photos, not a measured result.
- Product refurbish. A worn item and the restored item on the same surface. The product preset keeps the object centred; the appeal is the visible change on one object, not a before-and-after of performance.
Each of these works because one variable changed and the camera did not. If you are rebuilding a product from several stages, the chain mode below is the better format.
Chain up to six photos for a progress timeline
Keyframe Morph takes up to six photos in chain mode and animates a transition between each pair. Pair mode is the two-photo case; chain mode is for a timeline: bare room, stripped room, plastered wall, painted wall, furnished room, finished room.
Two constraints shape the plan. Each transition has its own duration, so the total length is the number of transitions multiplied by the seconds you choose. And every photo needs to share the same framing and light, or the chain reads as a slideshow of six different places. If you need a longer finished piece, chain mode is also the right way to extend the idea; how to make an AI video longer covers the general options.
The same technique is behind the “meet your younger self” format, where the two frames are the same pose at different ages: see the younger-self guide for the framing that makes it land.
Common mistakes and fixes
- The camera moved. Re-shoot the after photo from the same mark. No prompt recovers a different angle.
- The two photos have different crops. Crop both to the same ratio before upload. The tool inherits the frame from the images.
- The model ignores the end frame. Check the model. Fast and Mini do not land on it; Seedance 2 and 2.5 do.
- The transition is too slow for the length. A 4-second clip with three changes reads as a blur. Cut to one change.
- The claim is overstated. Label the clip as a transition between two photos, especially for fitness, cleaning and product results.
Frequently asked questions
How many photos do I need?
Two for a before-and-after clip: a start photo and an end photo. Keyframe Morph extends the same idea to a chain of up to six photos.
Which model should I choose?
Seedance 2 or Seedance 2.5, because both end on the supplied after photo. Seedance 2 Fast and Mini ignore the last frame and should not be used for this job.
How much does one clip cost?
A 5-second 720p clip is 82 credits on Seedance 2 and 126 on Seedance 2.5. The exact price is shown before you render.
Can I make a longer before-and-after video?
Use chain mode with up to six photos, or render a longer duration on a model that honours the end frame. Each transition in a chain has its own duration.
Is a before-and-after clip proof of a result?
No. It is a transition between two photographs you supplied. Label it that way, and never present a fitness or product comparison as evidence it is not.
Open the Before/After tool, upload both photos at the same crop and start with the cheaper model. New accounts start with 10 free credits and no card, and credits in one-time packs never expire, so a first test clip does not need a plan.
Ready to try it?
Create your first AI video in minutes — no credit card required.
Start Creating Free →