Kdenlive Feature Proposal: Real-Time Timeline Rating Tool for Rough Cut Acceleration

Summary

Introduce a real-time rating tool that allows editors to assign subjective interest values to timeline segments during playback. Ratings are stored as timeline-linked metadata and can be used for conditional navigation and batch editing.

This feature is intended to significantly accelerate rough cut workflows, especially for long-form footage.

Estimated time saving:20–30% reduction in rough cut timefor long recordings.


Core Concept

While playing back footage, the editor assigns a rating to the current moment.

The rating:

  • Is stored only when the value changes

  • Creates timeline segments based on rating transitions

  • Is linked to precise timecodes


Input Methods

Primary input via:

  • Keyboard shortcuts (e.g., 0–3 or 1–5)

  • Mouse buttons (configurable)

  • Mouse scroll wheel (increment/decrement rating)

Optional:

  • UI slider for visual interaction

Keyboard and scroll input should be usable without moving focus away from playback.


Rating Modes (Selectable)

  1. 1–5 star mode

  2. 3-level mode (Keep / Maybe / Remove)

  3. 0–3 quick numeric mode (optimized for speed)

Users can select preferred granularity in settings.


Timeline Visualization

A thin colored overlay at the top of the timeline:

  • Green = highest rating

  • Red = lowest rating

  • Gradient in between

This provides an immediate visual heatmap of subjective interest.


Navigation Functions

  • Jump to next segment where rating < n

  • Jump to next segment where rating ≥ n

  • Configurable pre-roll offset before jump target


Batch Operations

  • Remove all segments with rating < n

  • Create sequence from segments with rating ≥ n

  • Playback only segments ≥ n (automatic skipping of others)

Batch operations should allow configurable trim margins before and after cuts.


Team Workflow (Optional Advanced Feature)

  • Up to three independent rating tracks

  • Optional average calculation

  • Ability to toggle visibility per reviewer


Technical Considerations

  • Ratings stored only when value changes

  • Represented internally as time-segment metadata

  • Lightweight storage footprint

  • No frame-level continuous data required


Rationale

Current rough cut workflows require either:

  • Manual markers

  • Subclip creation

  • Repeated timeline scanning

This feature enables:

  • Fast subjective filtering

  • Structured first-pass review

  • Efficient second-pass refinement

It does not replace editing tools — it accelerates the selection phase.

- - - - - - - - - - -

Some additional ideas:

I suggest adjustable trigger time for appraisal sections. For example, if value stays less than 0,3 seconds, that section is not yet registered. The next value that stays for more than specified amount of time, is registered instead to the section. This minimizes unintentional interest values, also minimizing storage foot print.

Maximum feasible appraisal granularity is six steps. I suggest:

0= remove

1= maybe needed for continuity, not very interesting

2= needed for continuity, somewhat interesting

3= interesting, plot is clearly developing

4= high points of action

5= decisive moments

Last two may also be combined. Then: 4= most interesting moments

There should be a number indicating current interest value in the review window.

In my opinion at least using of up and down arrows is ergonomically good enough way to use this many steps. Also using mouse up and down movement may be, if constant clicking is not needed. Maybe mouse button down + movement up or down is best option? Using mouse scroll wheel, maximum feasible granularity is five steps (0-4), because then moving between highest and lowest value is quick enough.

It may be a good idea to implement a progress bar of appraisal.

When using feature “Create sequence from segments with rating ≥ n”, resulting video should have fade outs of sound at the end of segments, if high sound levels are detected at the end of those clips.

Later it may be feasible to implement AI integration. Some things that may be used in this regard are sound levels, voices; e.g. recognition of shouts, and movement value estimations.

Hi!

You gone into a lot of detail about your proposed Solution - but you haven’t really told us much, if anything, about the problem that you’re trying to solve.

Which is a problem, because your solution seems to be all about “rating” the “timeline” to “save time” - when surely the way to reduce wasted effort is to have rated and tagged and marked up your clips before you wasted the time dropping them all into the timeline in the first place if you know you only want to use some subset of them.

So maybe rewind a bit and tell us something about what kind of things you’re editing, and what your current workflow to do that is, and what problems that workflow is creating for you that you’d like to fix. Because there may be things we can do better here - but I am getting the feeling that you’ve found yourself in a hole that you dug yourself, and have imagined some very elaborate way to have a Magic Wizard lift you out of it, when possibly the most important thing you need to do first is Stop Digging.

It’s an interesting idea. It reminds me to the Pancake-Timeline workflow, where you have two timelines above of each other. On the top timeline you have all raw footage where you select the best clips and drag it down to the main timeline.

Assuming you have above concept in mind, I have the following questions:

- Are your ratings clip centric or are your ratings span over several clips ?

- Such a star rating system already exists in the project bin and is clip centric.

It would be interesting to think about your rating concept and once your rating is done to move or copy all best rated clips by one click from the top-timeline to the main-timeline without any gaps.