How to make a before-and-after transformation video

Last updated 2026-08-16 · 5 min read

What is the short version?

A transformation video works on contrast, and contrast is a deliberate choice rather than a side effect. The before has to look ordinary. If both halves get the same polish, the change reads as smaller than it was, which is the opposite of the point.

1. Go to Vizard Agent and upload every clip from the shoot.
2. Say clearly which clips are the before and which are the after.
3. Say how the two halves should differ in treatment.

What do you need before you start?

Clips from both halves and, crucially, a note about which is which. Vizard Agent inspects all the footage and can see that the person changed, but "before" and "after" is a narrative fact rather than a visual one — and getting it backwards is the one failure that ruins the whole video.

• All the clips, in whatever order. Vizard Agent probes them, samples frames and sorts out what it has.
• Which clips are before and which are after. Say it explicitly. This is not something to leave to inference.
• How the halves should differ. "No polish on the before, grade and music on the after" is the classic treatment.
• The shape and length. Vertical and under forty seconds suits a feed.
• Whether the original audio matters. Salon and studio footage is often noisy, and Vizard Agent will measure the levels either way.

What do you type into Vizard Agent?

Label the halves in the prompt. Everything else about a transformation video is a taste decision Vizard Agent can make well; the labelling is the one input it genuinely cannot derive from the pictures, and it is one clause.

Prompt

Variants worth knowing:

• Two separate cuts. "Give me the before as one clip and the after as another" if you post them apart.
• Split screen. "Show them side by side at the end" instead of sequentially.
• Any transformation. The same shape works for renovations, fitness, restoration and makeovers.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real transformation shoot of eighteen clips. Most of the run is inventory and measurement rather than editing, and the audio measurement partway through is the step nobody expects — it is also the one that decides whether the finished montage sounds professional.

1. Looks at the reference images the user supplied alongside.
2. Probes every clip for resolution and duration.
3. Extracts a representative frame from each.
4. Downloads all eighteen clips locally to work on them.
5. Builds a contact sheet of the wide clips and looks at it, then does the same for the vertical ones separately.
6. Measures the volume of each of the wide clips, because footage from a shoot varies wildly and a jump in level is more jarring than a jump in picture.
7. Checks what the music and sound effect tools can do.
8. Generates music and sound effects in parallel while extracting dense frames from the key sections.

Step 6 is the one worth stealing. In a montage cut from many clips, inconsistent audio levels are the thing that makes it feel amateur, and it is invisible until you watch with sound.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 35.2 seconds, AAC audio. Vertical and under forty seconds, with music Vizard Agent generated rather than the raw room noise from the salon, which is almost always the right call on footage shot in a working space.

Ask for the two halves as separate clips in the same conversation and Vizard Agent splits from the same assembly rather than re-editing.

This category was not separately measured, so treat the timing as a range rather than a promise: comparable work runs a median of 28 to 38 minutes end to end. Across all projects the median cost by tier is Flash 47, Pro 55, Max 242, Ultra 263 credits.

When does this not work well?

The technical work here is straightforward and rarely the problem. The failure is almost always narrative: either Vizard Agent has the story backwards because nobody told it which half was which, or the two halves got the same treatment and the contrast the video exists to show quietly disappeared.

• It cannot reliably tell before from after. In the run behind this page the user asked directly whether Vizard Agent could distinguish them. Label the clips and the problem disappears.
• A polished before kills the video. Resist the urge to make the whole thing look good. Say plainly that the before stays plain.
• Inconsistent framing between halves. If the before was shot on a phone from waist height and the after on a camera from the front, the change reads as a change of equipment.
• Consent for the person on screen. A transformation video is about someone's body or home. Whether you may publish it is not a question the edit answers.
• Very noisy source audio. Vizard Agent measures and balances it, and beyond a point music over the top is the honest fix.

How do you fix a result that came back wrong?

Say which half or which clip. Vizard Agent keeps the probe results, the contact sheets, the level measurements and the assembly, so re-ordering or re-labelling is a change to the edit rather than another pass over eighteen files.

• "You have the halves the wrong way round." A relabel, not a re-edit.
• "The before looks too good." Vizard Agent strips the treatment from that half.
• "The music starts too early." A timing change on the audio only.

How does Vizard Agent compare to doing it yourself?

By hand this is import eighteen clips, watch them all, sort them into two piles, pick the good seconds from each, level the audio, grade one half and not the other, then score it. The sorting and the levelling are the parts that take the evening and that nobody notices when they are done right.

Common questions

How does it know which clips are before? It does not, reliably. Tell it, and the rest of the job takes care of itself.

Can it do split screen instead? Yes. Say so, and Vizard Agent builds the comparison rather than the sequence.

Does this work for renovations or fitness? Yes. The shape is the same: two halves, two treatments, one contrast.

What if some clips are wide and some are vertical? That is normal on a shoot, and Vizard Agent groups them and inspects each group separately before deciding how to fit them into one shape.

Can I get both halves as separate clips? Yes, and asking in the same conversation is cheaper than a second job, because Vizard Agent splits from the assembly it already made.