How to join two clips with a transition made from what is in the shot

Last updated 2026-09-22 · 8 min read

What is the short version?

Two clips, and the join is the whole point. Foam in a cup and waves on a beach are the same texture at different scales, and a transition that uses that fact is invisible in the best way.

1. Tell Vizard Agent what the two shots contain.
2. Say the join should flow rather than wipe or dissolve.
3. Expect it to look at both ends of the join frame by frame.

What do you need before you start?

Both clips and a sentence about what connects them. Vizard Agent will find the visual rhyme itself, but naming it — foam to surf, steam to cloud, headlights to sunrise — saves a pass and usually produces a better join first time.

• Both clips. Even if they are only a few seconds each.
• What connects them. The shape, texture or movement.
• Which comes first. The direction matters.
• Whether the sound continues. It usually should.
• The finished length. How much of each clip survives.

What do you type into Vizard Agent?

Name the two things in the same sentence. "Join these nicely" invites a cross-dissolve; "from the foam in the cup to the waves on the beach" tells Vizard Agent that the foam is the transition and not merely the last thing on screen.

Prompt

Variants worth knowing:

• "From the foam to the waves." The rhyme, stated.
• "Flow rather than fade." Rules out a dissolve.
• "Keep the audio continuous." Joins fail on sound first.

What does Vizard Agent actually do?

Here is the Vizard Agent sequence on two very short clips joined at the moment a cup of coffee becomes the sea. The first version works and the second one is the one you want, which is normal for this kind of join.

1. Samples both clips, then pulls denser frames because each is only seconds long.
2. Looks at the foam's movement in the first clip specifically.
3. Looks at the waves' movement at the start of the second.
4. Checks the effects library for a transition suited to two foam textures.
5. Checks the audio so the join will not click.
6. Renders a first version with a soft diagonal transition.
7. Reviews the storyboard for direction and softness.
8. Extends the remaining wave footage and crossfades the audio.
9. Measures the final loudness and checks continuity end to end.
10. Goes back to the join: reviews the last of the foam and the first of the waves closely.
11. Picks the region of foam to use as the shape of the flow.
12. Holds the spill moment longer and blends the real foam texture as the thing that opens into the surf.
13. Confirms the foam flow is actually visible and the ending stays tidy.

Step eleven is the difference between the two versions. A diagonal wipe moves a hard edge across the frame in a direction the picture never suggested; using the foam's own shape means the boundary is something that was already moving that way.

Step eight is the practical detail that saves the join. Transitions eat frames from both clips, so a short second clip runs out — the wave footage is extended before the transition is built rather than after it fails.

Step five is worth doing early. A join that looks perfect and clicks audibly is a failed join, and the fix is a crossfade that has to be planned with the picture, not added afterwards.

What does the result look like?

Two clips that read as one movement: the foam rises, spills forward, and the spill becomes surf without an obvious moment where one clip ended. The sound carries across at a consistent level, and there is no hard edge travelling over the picture.

When does this not work well?

Some pairs have no rhyme to build on. If the two shots share no shape, texture or direction of movement, a constructed transition looks like an effect applied to hide a cut, and a plain cut is usually better.

• No visual connection. Nothing to carry the join.
• Static shots. A flow transition needs movement.
• Very short clips. The transition has no frames to work with.
• Hard cuts inside the source. The join competes with one already there.
• Strongly different colour grades. The moment of change shows anyway.

How do you fix a result that came back wrong?

Say what you actually saw at the join. Vizard Agent can change the shape of the transition, its timing or its length independently of each other, so telling it which of the three bothered you avoids a rebuild that alters all of them at once.

• "It still looks like a wipe." The shape is taken from the picture instead.
• "It happens too fast." The spill moment is held longer.
• "The audio clicks." The crossfade is planned across the join.
• "The second clip runs out." Its footage is extended before the transition.

How does Vizard Agent compare to doing it yourself?

By hand you drop a dissolve between them, look at it, try a zoom blur, look at that, and settle for whichever is least distracting. The match-cut version is entirely doable and it means masking a moving texture by hand, which is why almost nobody does it for a two-second join.

Common questions

What kinds of pairs work? Anything that rhymes — foam and surf, steam and cloud — and Vizard Agent will say if yours does not.

Can it find the rhyme itself? Often, but naming it gives Vizard Agent a much better first attempt.

Does it work with three clips? Yes. Vizard Agent builds one join at a time, each from its own pair.

Can it keep the audio from both? Yes. Vizard Agent crossfades them so the join does not click.

What if my clips are very short? Say so. Vizard Agent extends or holds footage rather than running out mid-transition.

Will it work on vertical video? Yes, though Vizard Agent usually rethinks the direction of the flow for that frame.

Can it match the colour across the join? Yes, and Vizard Agent usually should — a grade change at the join gives it away.

Is this the same as a morph? No. A morph reshapes one image into another; this uses a real texture as the boundary.

Can it do a hard cut instead? Yes, and for some pairs Vizard Agent will tell you a cut is better.

How long should the transition be? Long enough to read the movement; Vizard Agent keeps it short of becoming the subject.

Can I see both versions? Yes. Ask Vizard Agent for the simple one and the built one, side by side.

Does it slow the clips down? Only if you ask — holding a moment is different from slowing the whole clip.

What about a sound effect on the join? It can be added, though the continuous ambience usually does the work.

Will the join survive compression? Yes. Vizard Agent checks the finished file rather than the preview.