Shotcut tutorial · companion page
Split twelve frames off each side of a hard cut, put Color Grading on those two short segments only, and keyframe the picture out to white and back. The cut then happens inside a frame that is entirely white, so there is nothing left of it to see. The method is MusicalBox’s; this is Ben’s build of it.
Reading in another language? Use the translate picker in the header, or your browser’s own translate command — the Shotcut filter, control and value names are marked to stay in English so they still match what is on your screen.
There is no plugin, no downloaded overlay and no light-leak image. The whole thing is one stock filter on two short pieces of your own footage, keyframed to blow the picture out to white right where the edit lands.
There is no transition object anywhere in this build. The cut stays a hard cut — it is simply hidden inside a frame that has gone completely white, which is why the join reads as deliberate rather than as an edit.
Color Grading ships with Shotcut; nothing is downloaded and no image is imported. Only Shadows (Lift) and Highlights (Gain) are keyframed, and Midtones (Gamma) is left alone from start to finish.
The burst lives entirely inside two short segments split off the clips either side of the cut. Everything outside those twenty-four frames is untouched footage.
The method belongs to MusicalBox. Ben says so in the title, in the description, and in text on screen at 0:06 before he builds anything.
[ 12 frames on each side ]
on-screen text · 0:26
This video carries no chapter markers and nobody speaks in it. The sections below were derived from the screencaps themselves, sampled one frame every ten seconds, so each one is anchored to a frame that actually exists rather than to a list Ben published. Click any screencap or timestamp to play from that second, here in the card. Coverage is thin at that sampling rate — where a step falls between frames it is called out.
Before a single clip is touched the video puts a credit on the screen: Courtesy of MusicalBox. The technique is not Ben’s own invention — it comes from the MusicalBox channel, and the title, the description and the picture itself all say so.
The on-screen credit at 0:06, over the same night beach clip used for the whole build. The title card a few seconds earlier reads SHOTCUT 3-Minute Tip / LIGHT BURST TRANSITION.
The entire project is two files butted together on a single video track. There is no second track, no overlay, and nothing dragged onto the join — the cut stays a plain cut all the way to the end.
The Playlist holds video01.mp4 at 00:00:08:19 and video02.mp4 at 00:00:09:16. Project light-burst.mlt, total 00:00:18:11, one track V1. The cut falls at 00:00:08:19.
This is the one number the video states outright, and it states it in text rather than in speech. Park the playhead twelve frames before the cut and split; park it twelve frames after and split again. Everything that follows happens only inside the two short segments that leaves behind.
The caption 12 frames on each side is held from 0:26 to 0:40. The project’s video mode is never opened on screen, so how much running time twelve frames buys cannot be read off the video.
After both splits the track reads video01.mp4, a short video01.mp4, a short video02.mp4, then video02.mp4. The two long outer clips are never touched again.
The short incoming segment is selected, outlined in red. Both short segments sit flush against the cut with no gap and no overlap between them.
Select a short segment, open Filters, press the plus, and type col into the search box. Three results come back and the one you want is the first, Color Grading. That is the whole filter list for this effect.
Searching col returns Color Grading, Invert Colors and Old Film: Technocolor. No other filter is added to either segment at any point in the video.
Color Grading offers three wheels — Shadows (Lift), Midtones (Gamma) and Highlights (Gain) — each with a reset button and a stopwatch. Clicking a stopwatch drops a keyframe at the playhead and opens a track for that parameter in the Keyframes panel.
The Keyframes panel lists two parameter rows and only two: Shadows (Lift) and Highlights (Gain). Midtones (Gamma) is never given a stopwatch and never leaves 0.0 %.
Dragging the puck around the Highlights (Gain) wheel moves red, green and blue by different amounts, and the preview turns cyan instead of white. What the video does instead is leave the puck alone and type the same figure into the R, G and B boxes underneath it.
Caught mid-correction: G and B are sitting on 98.8 % while R is being retyped, and the preview above is visibly teal. That tint is the symptom of an uneven wheel, not a step in the recipe.
Highlights (Gain) on its own lifts the bright parts and leaves the dark ones dark. Pushing Shadows (Lift) up on all three channels as well drags the floor of the image up with it, and the frame goes flat white rather than merely washed out.
Shadows (Lift) caught mid-adjustment at 37.3 % on R, G and B together, stopwatch lit. The keyframe rows below carry a marker at the playhead on both keyframed parameters.
The incoming segment gets its own Color Grading with the ramp running the other way: white at the edge touching the cut, back to normal by the far edge. The two segments together read as one flash that peaks exactly on the join.
The filter panel header now reads video02.mp4. Both wheels are back on 0.0 % in this frame — this is the far end of the incoming segment, where the picture has to look untouched.
With both segments keyframed the cut has gone. The picture ramps to white over twelve frames, the edit happens inside the white, and it ramps back down over twelve more. There is nothing left of the join to see.
Final state on the incoming segment: Midtones (Gamma) at 0.0 %, Highlights (Gain) at 100.0 % on all three channels. The timeline shows the four clips with the two short ones straddling the cut.
Every number below was read off the pixels of the video. Nothing here is a recommended-sounding guess — where a value was never shown, the page says it was never shown.
Drop both files onto V1 so they butt up against each other. No second video track is used anywhere in this build.
The project’s video mode is never opened on screen, so the resolution and frame rate behind these timecodes are not something the video states.
Move the playhead twelve frames back from the join and split the clip. Move it twelve frames past the join and split again. You now have four clips: two long ones you will not touch, and two short ones that carry the whole effect.
This is the only quantity the video states explicitly, and it states it as on-screen text: 12 frames on each side.
Select the short piece of video01.mp4, open Filters, click the plus, and search col. Choose Color Grading. It is the only filter this effect uses, on either segment.
Put the playhead at the far edge of the segment — the edge away from the cut — and click the stopwatch beside Highlights (Gain) with all three channels still on 0.0 %. Then move to the edge touching the cut and set R, G and B to 100.0 %.
Type the figures into the R, G and B fields rather than dragging the puck. Dragging moves the three channels unequally and tints the blowout; the video shows exactly that going wrong at 1:50, with G and B stuck on 98.8 % and the preview turning teal.
Repeat on the Shadows (Lift) wheel: 0.0 % at the far edge, high at the edge touching the cut. This is what turns a bright, washed-out frame into a genuinely white one, because it lifts the black point as well as the highlights.
The peak here is a judgement call rather than a fixed figure. The wheel is seen passing through 18.3 %, 29.0 %, 30.5 % and 37.3 % while it is being tuned, and settles on 39.6 % across four separately sampled frames. Anything that reaches white will do.
Select the short piece of video02.mp4 and build the same two ramps in reverse: full white at the edge touching the cut, back to 0.0 % at the far edge. Both segments end up with two keyframed parameters each.
Midtones (Gamma) is never given a stopwatch and stays on 0.0 % throughout — the Keyframes panel only ever lists two rows.
Scrub over the cut. The picture should ramp cleanly to white, cut inside the white, and ramp back down. If you can still see the edit, the two ramps are not peaking on the same frame.
The title promises three minutes. The build on screen runs from roughly 0:18 to 3:31 — a little over three minutes of actual work, so the claim holds up.
Read directly off the frames. Where a value was tuned on screen rather than settled, the tuning path is shown separately below.
| Setting | Value | Where it lives |
|---|---|---|
| Shotcut version | V20.09.27 | Video description — see the note below |
| Project file | light-burst.mlt | Title bar |
| Project video mode | not shown on screen | — |
| Source clip 1 | video01.mp4 | Playlist |
| Source clip 1 duration | 00:00:08:19 | Playlist |
| Source clip 2 | video02.mp4 | Playlist |
| Source clip 2 duration | 00:00:09:16 | Playlist |
| Timeline total | 00:00:18:11 | Player |
| Track stack | V1 only — one video track | Timeline |
| Clips after splitting | 4 — long, short, short, long | Timeline |
| Burst length | 12 frames each side of the cut | On-screen text at 0:26 |
| Filter | Color Grading | The two short segments only |
| Filter search term | col | Filters panel |
| Keyframed parameters | Shadows (Lift), Highlights (Gain) | Keyframes panel |
| Midtones (Gamma) | 0.0 % throughout, never keyframed | Color Grading |
| Highlights (Gain), far edge | R 0.0 % / G 0.0 % / B 0.0 % | Color Grading |
| Highlights (Gain), at the cut | R 100.0 % / G 100.0 % / B 100.0 % | Color Grading |
| Shadows (Lift), far edge | R 0.0 % / G 0.0 % / B 0.0 % | Color Grading |
| Shadows (Lift), at the cut | R 39.6 % / G 39.6 % / B 39.6 % | Color Grading |
The Help → About Shotcut dialog is never opened in this video, so the build cannot be read off the screen the way it can on the other tutorials in this series. The version above is the one Ben states in the video description, V20.09.27, which is consistent with an upload on 28 October 2020. The interface in the screencaps matches that era of the application.
The Shadows (Lift) figures are a path, not a target. The wheel is seen passing through 18.3 %, 29.0 %, 30.5 % and 37.3 % while it is being tuned; 39.6 % is where it settles across four separately sampled frames. Highlights (Gain), by contrast, is typed to an exact 100.0 %.
This tutorial has no narration. The audio is a continuous music bed for the full 3:45, and the only words in the video are four pieces of text burned into the picture. They are reproduced below in full — that is genuinely all of them. The technique itself is shown, not explained: the mouse moves, the numbers change, and you follow along.
Three checks agree that there is no speech here. A machine transcription of the audio returns six segments across three and three-quarter minutes, and five of the six are the identical sentence — the phrase a speech model emits over music with no voice in it. A level analysis of the audio track shows a mean of −9.4 dB with no gap of any kind until 3:31, which is a continuous music bed rather than the stop-start shape of somebody talking. And the frames themselves carry no burned-in subtitles, only the four captions above. So this page carries no transcript: a near-empty one would only look as though something had gone missing.
