I recently shot this raw footage using my 50D, and I thought it might be helpful to walk folks through the post-production workflow process.
It’s definitely more labor intensive to shoot raw, but the difference is worth it I think. Here’s footage shot the same night in H.264:
The difference in dynamic range and low-light sensitivity is remarkable in my opinion. A lot of the frame just goes to solid black in the H.264 footage, while so much more detail is preserved in the raw footage. And I have to say I’m super impressed at the quality of the footage shot at night with no additional lighting at such a high ISO as 3200.
In addition, one benefit of shooting raw at this event was that I was stuck in the middle/back of the crowd, so the “crop factor” of shooting in crop mode, which effectively turned my 50mm lens into a 200mm, was of great use, giving me a much more telephoto image in order to see the organizers at the front of the crowd (for example).
Ok, so here are the details of the production. The first video above was shot on the 50D in raw rec mode. 50mm/f1.8 lens, ISO 3200, aperture set to 2.4 or 2.8 to preserve a bit of depth of field that would be lost at 1.8. The first shot was in regular mode at 1568×882. The rest was shot in “crop mode” to increase resolution as well as effectively zoom my 50mm lens from 80mm effective to 200mm effective. I wasn’t able to get continuous recording on the fly at 1920×1080 (I hadn’t had time to test the nightly build and prep before the protest started), so I dropped resolution to something like 1780 in order to record for over a minute or two at a time.
There are lots of options for post-production workflow, but this is mine given my current resources (a 2011 macbook pro and the Adobe CS6 suite):
- Convert the ML raw footage to DNGs using RawMagic Lite.
- Open the DNGs in After Effects using the Adobe Camera Raw plugin.
- Correct exposure, but not much else.
- Some people add a filter here to convert the raw footage into a “log” type look here, but I have yet to try that out, so I can’t comment.
- Set the project to 16bit (as ML raw records at 14-bit uncompressed, rather than 8-bit H.264 like we are used to).
- I use Magic Bullet Resizer to upres the footage to 1920×1080.
- Export each shot as ProRes4444 with trillions of colors. I think this is the highest quality way to export, but I’m not certain… would love to hear suggestions on this (or any other aspect of the workflow!) from folks who have more experience.
- Then I re-import the master ProRes files to After Effects to do noise reduction (AE won’t allow you to upres and use NR in the same sequence). I use the Neat Video plug in for my noise reduction, with the maximum setting of 5 for best results (but longer processing times).
- Set up my shots to render out again at ProRes4444 with trillions of colors and let it run over night.
- I now have all of my “master” footage set up as ProRes4444 1920×1080. I might also create ProRes proxy copies of all the shots to do my initial edit, and the relink to the ProRes4444 files when I’m done to create the final, edited master of the project.
This is the workflow I am currently using. Let me know if you have any suggestions for improvements!