Exploring the New Scene Cut Detection Features of DaVinci Resolve and Adobe Premiere Pro

Comparing the new Scene Edit Detection features and discussing their use cases in video post-production workflows

Filip Milovanovic
Post-production expert,

The competition between the development teams behind DaVinci Resolve and Adobe Premiere Pro appears to be hotter than ever, with each of them matching the other’s progress at a furious pace. And, having recently compared their respective speech-to-text functions, we decided to examine and compare both NLE’s new scene detection features.

How It Works with Resolve

The Scene Cut Detection feature in Resolve is used while importing footage. Open the Media page and locate the footage you want to use. Instead of importing the clip via drag and drop, you will need to right-click the clip and select Scene Cut Detection from the drop-down menu. This will open the new Cut Detection window. Use the Auto Scene Detect button in the lower-left corner of the window to start the process. After a quick analysis, the detected cuts will be listed in the Info tab, and a graph will be displayed.

The vertical lines on the graph represent the algorithm’s confidence in the detection of a scene cut. If the confidence is higher than the purple horizontal threshold line, a cut is detected, and a clip is added to the Info tab. Use the mouse to move the threshold line lower or higher to confirm more or fewer detected cuts.

The viewer displays the frame at the playhead position, as well as the final frame of the outgoing clip on the left side and the second frame of the current clip on the right side. Clicking the Add Cuts to Media Pool button in the lower right corner will import the detected scenes into Resolve for editing.

How It Works with Premiere Pro

To use this feature in Premiere Pro, insert the clip into the timeline, right-click the clip, and click Scene Edit Detection from the drop-down menu. A window appears, allowing the user to decide whether a cut or marker should be added to the clip at the detected edit point and whether subclips should be extracted after the detection. Premiere will automatically analyse the clip and mark the edits as selected in the settings.


The accuracy of the Scene Edit Detection functions was tested and compared using two different kinds of footage.

The first clip was a one-minute edit of a man entering a coffee shop and working on a laptop. The clip contained eighteen simple cuts without any transition effects. Resolve’s Scene Cut Detection with default settings recognised seventeen edits correctly, missing one and making one additional false detection. Premiere Pro, on the other hand, delivered a perfect score of eighteen cut detections.

Resolve’s algorithm interpreting a sudden pixel-value change in a part of the frame as a cut.

Unfortunately, Resolve’s algorithm delivered even worse results with our second test clip. The content of this clip was more abstract, featuring macro footage of different coloured paints being mixed. It featured six jump cuts, two dissolve cuts, and one additive dissolve cut. In this test, Resolve recognised only one cut correctly and made two false detections. Ironically, the correct cut had the lowest confidence score. The cut with the additive dissolve transition was detected, however, with the confidence level below the default value. Resolve’s algorithm seems to interpret sudden brightness changes as cuts.

Premiere’s feature proved its accuracy again by detecting all six jump cuts. The cut with the additive dissolve transition was not detected. As expected, both NLEs didn’t detect the cut with the dissolve transition.

Media Library Scene Edit Detection

The ELEMENTS Media Library is a powerful MAM platform that has been offering this feature for over six years. In our accuracy test with both clips, the Media Library’s scene detection feature delivered the same accuracy as Premiere Pro, detecting all eighteen cuts in the first test clip and all six jump cuts in the second test clip.

Unlike the NLEs, the Media Library automatically performs scene detection analysis for every clip. In preview mode, users can use the arrow buttons to jump between the previous or next detected edit.


The Scene Edit Detection features in both NLEs are designed to detect simple cuts without applied transitions, and both were unable to detect an edit with a dissolve transition effect. When an additive dissolve transition was used, Resolve registered a possible edit, but the confidence of the detection was very low and far below the default detection value.

Unlike Resolve, both Premiere and the Media Library’s set-and-forget style of Scene Edit Detection don’t allow the user to tweak the parameters for the detection, but both provided far better accuracy during our tests.

For this reason, we’ve concluded that Premiere Pro and Media Library offer a better user experience, delivering exceptional results without the need to learn how to use a new feature. Resolve’s rapid development, however, makes us confident that the accuracy of default detection will continue to improve with future updates.

For now, we would suggest using the Scene Edit Detection feature in Adobe Premiere Pro or our own Media Library.

Check out some of our other blogs about Resolve and Premiere Pro:

DaVinci Resolve – Performance and Troubleshooting Guide
Adobe Premiere Pro – Performance and Troubleshooting Guide
DaVinci Resolve Super Scale – Turn HD Footage Into 4K or Even 8K Footage
DaVinci Resolve 18 Reference Manual with Chapter Links
New Premiere Pro Beta Feature: Assemble Rough Cuts from Transcripts


Using the OpenAI Transcription Engine to Generate Subtitles


Automatic Transcription with DaVinci Resolve’s Speech-To-Text Function and Text-based Video Editing


Using an Alternative Microphone Input in DaVinci Resolve with a Blackmagic I/O Device



COBIT ist ein international anerkanntes Rahmenwerk für das Management und die Governance von Informationstechnologie. Es bietet ein umfassendes Regelwerk von Prinzipien, Praktiken und analytischen Instrumenten und Modellen zur Steuerung der unternehmensweiten IT.