I want to preserve as best I can the quality of digital photos taken in RAW format and then rendered as animation sequences in Blender. My workflow is as follows:

  1. Process RAW (.cr2) images with Darktable
  2. Export as OpenEXR
  3. Track and composite animation sequence in Blender
  4. Save animation as OpenEXR image sequence
  5. Open multiple image sequences in VSE and render final video file

The OpenEXR format is very large (over 200MB per image). Do I need to use it in both steps 2 and 4? Also, what settings should I use for the final video render?

  • $\begingroup$ What's your delivery requirement? $\endgroup$ – user1853 Jun 4 '14 at 20:11
  • 1
    $\begingroup$ Related: blender.stackexchange.com/a/3472/599 $\endgroup$ – gandalf3 Jun 4 '14 at 21:00
  • $\begingroup$ @cegaton I don't have any specific requirements. I may want to upload to the web as HD video. I may also want a higher quality (possibly 4K) for archival purposes. $\endgroup$ – Reed G. Law Jun 4 '14 at 23:39
  • 1
    $\begingroup$ Seems to me that having number 2 with OpenEXR is overkill. The cr2 format is 14 bit. Trancoding to a 16 bit container should be enough to keep all the original information. You might want to try using 16bit PNGs or Tiffs. PNGs will certainly be smaller. Also, for tracking, consider making jpg proxies, as they might yield better performance in blender. If you use OpenEXRs on step 4 make suer you use the multilayer kind, so you can save your render passes without having to re render in the future. $\endgroup$ – user1853 Jun 5 '14 at 2:02
  • $\begingroup$ You mention jpg proxie which brings up a related question of mine. How can I change the clip format without losing my tracks? $\endgroup$ – Reed G. Law Jun 5 '14 at 2:29

It is more common to save the output of blender rendering as a set of PNGs. You can turn up the compression in the Render > Output section, but I can not predict if it will save space compared to the OpenEXR.

Making a VSE output to compressed video (as you do) is very common, and almost required if there is any sound involved. The settings you use for the final video render will vary depending on

  • how much bandwidth you are willing to allocate for streaming
  • resolution of video (3.75Mbit/s MPEG-2 is CableLabs VoD standard for SD. You will want 4-9Mbit/s for H.264 1080p video)
  • what formats are supported by the client software (motorola DCT-2000 -vs- iPhone)
  • does the container support the codecs you want (MPEG-2 program stream doesn't support AAC and blender does not seem to support transport streams)

Some more reading is at http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Output/Video

  • 1
    $\begingroup$ I'm pretty sure that a PNG will almost always be smaller than an EXR. Maybe if you using 16 bit PNG with half-precision EXR you might get occasionally larger PNG, but I don't know. $\endgroup$ – gandalf3 Jun 4 '14 at 21:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.