I just came back from a video shoot and I thought I’d use the time to share my video workflow.
My usual video workflow uses a Canon 5D Mark III with the Magic Lantern firmware. This gives me the possibility to record raw footage and since we all agree recording raw is a great thing, this is the best way to record video for me. Having said, that Magic Lantern is cool – and open source – the inconvenience starts exactly here: long story short what I need to do is convert the MLV into a sequence of DNGs, convert these into an intermediate format, edit that, color grade it and then apply the final touches to it.
Now lets dig into more detail on that workflow.
First you need to prepare yourself for a huge amount of data compared to just shooting single pictures. I use 64GB UDMA 7 CF-Cards, which provide the ~100MB/s write speed I need to record RAW footage. This leaves me with roughly 12 min. of record time per card.
Once a card is full I copy the files to an external USB 3.0 hard drive using a USB 3.0 CF-card reader. This takes about 12 min per card, which makes it clear, that you need at least two card to record without the unwanted “I need to copy the files” intermission.
At the end of a session – or maybe during it, if I could afford a digital assistant – I use a (Bash) script to convert all the MLV files into a sequence of DNGs and a WAV files.
Now these files are pretty useless, as of today I haven’t found a way to use them directly in any NLA, except for DaVinci Resolve. So what I need to do is import them into DaVinci Resolve and then export single clips as ProRes 422 or ProRes 4444.
Having done that, I import those files into Adobe Premiere Pro and making all my edits. This also requires me to match video and audio manually more times than I would which to admit. Though Premiere Pro offers the possibility to work pretty seamlessly with programs like After Effects I don’t do that at this point. At this point I literally only make my edits, meaning arranging the shots and cutting them.
After that I export the sequence as an XML file, which I then import back into DaVinci Resolve. Here I make all of my color grading and yes I shoot a color card to get the most accurate results. When I’m done with that I export the sequence as an XML/ProRes once again import it back into Premiere Pro.
Pros and Cons of this workflow
- I use RAW
- I use at least three different programs to get to my end result
- There is no room for dynamics in this workflow. If I notice something, when I do my lower thirds, worst case is I need to go back to my edits and rerun all the steps in between to correct this mistake
- This workflow uses a lot of intermediate files, which isn’t optimal imo.
Finally it comes down to this: The Adobe Creative Cloud software is great for editing and special effects, but isn’t that great for dealing with a sequence of DNGs, let alone MLVs. DaVinci Resolve is great for dealing with a sequence of DNGs and doing color grading, but isn’t that great for editing.
If I were to make a feature request for the Creative Cloud software it would be to be able to import MLV files directly into Premiere Pro and then send them over to Resolve once, directly before delivery. I mean seriously, from all I can gather Magic Lantern is widely used for filming on Canon cameras. It is an open format, from all I can gather. How hard can it be to merge that into the Creative Cloud?
How about the video from this shoot? You can watch this masterpiece right below 😉