Two shots.
One visual connection.

1. Choose the connection the cut should make.

A match cut connects separate shots through a visual or movement relationship. The useful question is not simply whether two images contain a circle, a face or the same colour. It is whether the viewer can recognise the relationship at the moment the edit happens, and whether that relationship adds something to the story.

The research paper Match Cutting: Finding Cuts with Smooth Visual Transitions describes matching through framing, composition or action and studies ways to find candidate shot pairs. It supports treating the relationship between shots as the subject of review, rather than judging each clip alone. It does not demonstrate the fictional production below or guarantee that an AI generator will deliver matching footage. Source: Chen and colleagues, 2022.

This is a graphic match, not a request for the light to transform into a wheel. That distinction changes the scope. A transformation needs a believable intermediate action; a cut needs two useful boundary frames and enough surrounding footage for an editor to choose the moment. Do not ask for both unless the concept genuinely needs both.

Write the intended meaning in the brief before preparing images. If the connection cannot be explained without describing a software effect, reconsider it. A technically neat transition can still distract from the product or interrupt a line of thought.

2. Define two anchor frames on the same canvas.

Start with the last intended frame of shot A and the first intended frame of shot B. Put them on canvases with the same aspect ratio. Mark the centre and apparent size of the matching feature, then note its angle and the contrast around it. These are the properties the viewer has a chance to recognise across the cut.

For the Orven example, propose that both circular forms sit at 55% of the frame width and 48% of its height, measured from the top left. Give each an apparent diameter of roughly 30% of the frame width. The exact values are less important than recording a shared target. They can change during layout review.

A wheel seen at a steep angle reads as an ellipse. A light viewed head-on may read as a circle. Calling both subjects circular does not solve that mismatch. Either change one viewpoint or decide that the looser association is intentional. Record the decision before making moving versions.

Use a simple overlay of the two proposed frames in the edit or design application to inspect position and size. This is a layout check, not a claim that the entire transition works. It cannot tell you whether the motion feels continuous or whether the cut arrives at the right moment.

3. Prepare the images for the relationship you need.

Once the anchor plan is clear, create or select images that support it. Keep a separate reference for the object’s appearance and a layout reference for its placement. A beautiful source image with the feature in the wrong part of the frame may require more compromise than a simpler image designed around the cut.

Runway’s Gen-4 Image References documentation describes using reference images and sketches to guide composition and object placement. It also recommends iterating individual elements for complex changes. Those are documented controls and vendor recommendations, not a promise of exact pixel alignment or a verified result for this example. Source: Runway’s reference-image guidance.

For the fictional wheel shot, establish the wheel design before treating its rim as an abstract circle. If the subject were a real product, the correct design would remain a requirement even when an alternative wheel happened to match the light more neatly. The transition should serve the product, not quietly redesign it.

Prepare enough room around each feature to make modest framing adjustments. Check what would be lost if the image were enlarged or shifted: a recognisable product detail, the edge of the light, or space needed for another crop. Do not assume an editor can repair any mismatch by zooming in.

Use neutral file labels such as “A anchor candidate 02” and “B anchor candidate 03.” Keep approved anchors separate from experiments so a later generation does not accidentally start from an obsolete composition. The reference-image guide explains how to make source material useful without overwhelming the team.

4. Match the movement around the cut.

Two still frames can align while their moving versions feel unrelated. Inspect how the feature approaches the cut in shot A and leaves it in shot B. Is it growing in the frame, moving sideways or rotating? Does the second shot continue that tendency, stop it deliberately or reverse it without explanation?

In the first Orven pass, keep the movement restrained. The proposed camera approach makes the circular feature grow slightly in both shots. Avoid adding a fast wheel spin, a moving vehicle and a large camera orbit at the same time. Each new action introduces another relationship to control.

Brief the motion in terms of what must be visible around the edit point. Ask for usable material before and after the intended cut rather than a clip whose only matching frame sits at its very edge. Those extra frames give the editor room to find a better rhythm. The amount needed depends on the edit and the available generation or production method; there is no universal handle length.

Review the pair without a transition effect first. If the direct cut does not communicate the intended connection, a dissolve may hide the problem without solving it. A dissolve can still be the right creative choice, but it should be chosen for the idea rather than used automatically to cover a framing error.

The camera movement guide covers how to choose a move. Here the additional task is to judge two moves as a pair. If one shot has attractive motion that cannot connect to the other, decide whether to revise that shot or abandon the match.

5. Diagnose the cut one variable at a time.

Place the two clips together and watch at normal speed. Ask someone who has not been staring at the anchor frames what they noticed. Then inspect the boundary more closely. The aim is to separate a compositional problem from a timing problem so the next revision has a clear purpose.

For a hypothetical failed Orven pass, imagine that the wheel is noticeably smaller than the light. The first revision would be a framing adjustment, provided it preserves the wheel detail and required crop. Changing the soundtrack, colour grade and camera speed simultaneously would make it harder to learn whether scale caused the problem.

After each change, watch the full surrounding sequence again. A pair that looks convincing in isolation can still interrupt the film’s rhythm. Keep a record of the chosen clip versions and edit points, especially if the images are being regenerated while the edit develops. The guide to editing separate AI shots covers the broader assembly process.

6. Approve the pair in every required version.

A wide match cut does not automatically survive a vertical crop. Recheck the two anchor frames on each delivery canvas. A subject position that is consistent in the wide master may become inconsistent when the crops follow different points of interest. Treat the crop as another composition decision, not merely an export setting.

Sound can reinforce the connection, but first establish what the picture is doing. A shared sound across the cut, a deliberate silence or a musical accent each creates a different emphasis. Review the transition with the intended voice and soundtrack so the visual event does not compete with the main message. This morning’s music briefing article provides a way to describe that priority.

There is no obligation to keep a match cut simply because it took effort to prepare. If it needs excessive reframing, obscures a product or makes the film harder to follow, a straightforward cut may communicate the idea more clearly. Agree that fallback before production so the team can make an editorial decision without defending sunk work.

The value of the exercise is the shared specification. Instead of asking for two impressive clips and hoping they connect, the team defines the relationship first, prepares the boundary frames and reviews the actual pair. That turns an attractive transition idea into a production task that can be discussed, revised and approved.

Sources checked 21 October 2026. Orven is an imaginary label. Coordinates, motion notes and failure diagnosis are illustrative planning examples; no generated test or client outcome is claimed. Runway’s documentation establishes available guidance, while the research paper establishes the cited definition and candidate-pair approach. Explore images are independent references, not before-and-after results.