Thursday, July 10, 2014

Further experiments with Magic Timelapses


As you will be aware from some of my previous posts, Magic Lantern offers a silent (shutterless) DNG capture in LV or Video mode. The current ‘downside’ is that this capture is about a quarter of the size of a full frame still capture and there is no embedded EXIF data. Although it is early days, there is a hint that the ML gurus are close to creating a full frame silent capture; and I believe they are also trying to address the missing EXIF.
 
Until the silent DNG EXIF is sorted out, undertaking a Holy Grail sequence using this route, ie sunrise or sunset, is problematic. Which means falling back on to ‘normal’ shutter-based .CR2 captures.

So how do you do an in-camera ML-based Holy grail Timelapse? Well here is the basic workflow:

  • Compose and focus;
  • Take a base ETTR exposure; 
  • Put the ETTR to ‘Always On’ and set the minimum shutter – I set 1 second for the sequence below, at which point the ISO gets bumped up a stop at a time (when required, although last night this wasn’t necessary); 
  • Set your timelapse parameters, ie delay, number of frames and interval; 
  • Start the capture – the sequence last night was 300 frames at a 10 second interval, ie 50 minutes in the real world or about 12 seconds at 24fps; 
  • Ingest into Lightroom and initialize in LRTimelapse; 
  • Use the Holy Grail wizard to undertake some corrections, set some addition key frames and auto reimport the Metadata .XMP files back into LR; 
  • Using the key frames tweak the look of the timelapse and, because it is an ETTR Holy Grail capture, ie the exposure is rather ‘jumpy’, export the sequence to HQ JPEGs (you could try deflickering at this stage, but it could get ‘messy’);
  • Import the JPEGs into LRTimelapse and deflicker, twice if required; 
  • Bring the deflickered JPEGs into LR and export a second time to JPEGs for video rendering; 
  • Either create the video at this stage or import the JPEGs into Panolapse and undertake any dolly/zoom actions you wish and deflicker again for safety, then render the video.

I know it sounds complicated, but I can assure you the LR/LRT combination works seamlessly and fast.

Here are the two videos from last night’s testing. The first is the static capture, the second a Panolapse version.

Bottom line: Timelapses for me are the ‘acceptable’ way for still photographers to experiment with motion and show the World in a way that is not normally visible.

No comments:

Post a Comment