Back to All Work
Auto Timelapser

Need for This Tool

Typical professional timelapse stitching occurs in a NLE (nonlinear editor), or with a tool like the Lightroom plugin lrtimelapse. These tools are great, but also can be difficult to operate, and lack set-and-forget automated timelapse creation functionality.

The following functions are core to a timelapse making app:

  • Ingest a wide variety of RAW files
  • Normalize exposure (see details below)
  • Apply exposure ramping, and keyframed edits
  • Export edit .xmls or a video in a modern format

To accomplish this today, you need Adobe Lightroom ($20/mo), lrtimelapse ($298), and the expertise to use them, or a tool like GlueMotion ($28), which can do some of this, but with limited control.

Building this tool, some additional first time features can be added on top of the current state of the art

  • Automatic HDR Bracket + Timelapse merge
  • Export standard .xml edits for each image (allow external RAW processing)
  • Export to Final Cut, DaVinci Resolve, or Premiere Pro timeline files
  • Export to ProRes with log profile for cinema workflow colouring
  • "Auto exposure points" functionality, allowing auto exposure with gradual ramping

I use this tool to create compelling timelapses, and think it can get 100+ sales at $10 on the App Store for Mac when released publicly.

Application Features

Cover image
A low-fi mockup of the exposure smoothing and Timelapse stitching UI

Exposure Normalization

Add adjustments to all frames to conform them to the same real-world exposure value (normalize for settings changes)

Exposure Smoothing / Ramp

Find the “exposure curve” of time vs frames, smooth that curve so that exposure changes over time appear gradual. Apply the exposure change to fit the frames to the smoothed exposure curve. Use a tunable smoothing factor

Auto HDR on Frame

When auto HDR is enabled, photos with a small (exposure time +1s) difference in capture time and different exposure are HDR blended. Photos with that time difference but no exposure difference can be mean or median blended as well. A frame can therefore be composed of either one photo, or multiple photos with combined exposure (HDR or median or mean)

To do this, we conceptualize frames as existing on a 2d continuum, of time and exposure. We use their relative clusterings in this plane to decide whether to make them separate frames (distant on time axis), blended in HDR (very close time, distant exposure), median/mean blended (very close time and exposure).

If we add a third axis for similarity to previous frame, we can also apply this algorithm to entire photo libraries, allowing auto HDR and panorama clustering for photo libraries, IE if photos are very close in time and bracketed in exposure, or have adequate % similarity to stitch as panorama, these procedures can be run automatically on a photo library.

Full Auto Mode

Dump in a batch of RAW files, and have it merge them correctly, regardless of poor exposure, presence of HDR bracketing, etc.

  1. Receive an image sequence
  2. Merge HDR bracketed sets of images
  3. Calculate auto exposure for each image
  4. Smooth out and remove outliers from the curve of exposures generated by the auto exposure calculation
  5. Apply auto exposure curve to the image sequence
  6. Export correctly exposed and smooth timelapse

Edit processing and saving

RawPY or another RAW library

Process RAWs in app, without support for the greatest colour science or HDR.

Export Adobe XMLs

Create Adobe compatible XML files for the edits for each frame, so they can be edited and exported from ACR or Lightroom, including in HDR.

Export NLE Timeline

Create a timeline file for DaVinci Resolve, Final Cut Pro, or Premiere Pro which contains the relevant image sequence and exposure edits.

Project Status

This app is currently being used by me to create Timelapse's. The UI is being polished, and it is planned to sell on the App Store for Mac in late 2024.