Skip to content →

Shard Cinemagraph

We’ve had a spell of lovely weather here in London, and I wanted to take advantage of it by making my first cinemagraph!

Here’s an outline of my workflow, including how I went about fixing some important issues. Don’t hesitate to leave a comment below with any questions if something isn’t clear, and I’ll try to answer them as best as I can!

  • I mounted my camera on a tripod, pointed it at the shard, and waited. 
  • Video was taken rather than a burst of photos, as the camera didn’t have a burst mode. But I’m glad I did it this way, because bursts of photos would have had a few problems. Even if their resolution was higher, I would have lost the information in between exposures and had less freedom with the speed that I ran the gif at. More importantly, the result would have been jittery as the clouds would have jumped from picture to picture: with 24fps video, I was able to merge adjacent photos together to get a much smoother video.
  • I had to record constantly, so I didn’t miss the perfect shot when it came along. I ended up with over an hour of footage I didn’t need, but when the right cloud came along in the reflection I knew that it would be the one and stopped recording.
  • I stabilised the video as there was some slight drift over the shot, even though I had used a tripod. I did this in Blender, and you can find some great tutorials for how exactly at /r/ImageStabilisation.
  • The video was quite noisy, especially as I had to zoom in a considerable amount for the shot I wanted. I couldn’t composite it how I liked without making the noise even more obvious. Luckily, the way I sped up the video, which was by merging each 8 adjacent frames, eliminated this very well as the noise was averaged over the pictures. This was done in Blender’s sequence editor.
  • I applied some post production, again in Blender. RGB curves were all I needed to get the desired look.
  • The next part, where I made it loop seamlessly, was the most time consuming part of post production. I’m sure there’s a more efficient way, and if anyone knows it I’d appreciate a comment telling me how! What I had to do is load the desired frames into GIMP, and manually merge the last few with the first few, with varying opacity so that the end smoothly merged with the beginning. This needed a lot of trial and error before I overlapped the right number of frames to make it look smooth.

Published in Tutorials