Magic Lantern, Raw Video and the Importance of Workflow

In my last post I mentioned that I have a Canon 5D Mark III running Magic Lantern firmware. I made that choice and installed the firmware before I had even shot with the camera. I read an extensive amount about the so-called hack, and, even though it is risky because it could void the warranty on the camera (even though Canon makes no such claim), I chose to do so because the benefits are quite desirable.

If you don’t know about Magic Lantern, I’ll give you a very brief rundown: There is a group of programming and photo enthusiasts who graciously create this software (it’s continuously added to and adjusted for bug fixes) that allows some Canon DSLRs to capture 1080i HD video in 14 bit. Normal video capture occurs in 8 bit, meaning that there is a ton more data captured in each frame. Essentially, the camera’s sensor takes a 1920 X 1080 (2 megapixel) capture for each frame. That’s the main feature that I need Magic Lantern for. However, there are many other features that the firmware brings to the table, most of which I likely won’t ever use. Some of the other features include bulb ramping for time lapse (which I would like to use, but need to learn first), fine and larger focus adjustments, and sound recording control. The features added to the camera by running Magic Lantern’s firmware are really useful and Canon and the other major camera manufacturers could learn a bit by analyzing this firmware.

My Workflow:

I’m definitely still finding my way through this process, but I’m doing what I can with the software I have. When recording raw video, Magic Lantern’s firmware creates a file with the extension .mlv (magic lantern video, I’m guessing). These files are ENORMOUS. But, if you think about it, it makes sense. Each frame is 2 megapixels and 14 bits. Yesterday I shot about 7 minutes of footage for a friend’s real estate listing and it came out to 40 GB of data. Needless to say, you’ll need lots of memory, and it needs to be able o record quickly (x1000 or faster).

This file must be converted into a series of .dng. After a little searching, I was able to find software on Magic Lantern’s website that runs on Mac. It’s is called MlRawViewer1.3.3 and was developed by the fine folks at Magic Lantern, and as such, is free. It installs like a normal application on Mac. Only trick is that Mac’s security will not allow the file to open. You have to open settings>security and privacy>click allow at the bottom of the window.

Go back to the folder with .mlv files and double click one of those files. It opens in MlRawViewer by default, and there you can adjust for the contrast curve you like best. I’ve found that Log 8 worked well for this last shoot. But, I don’t believe that it matters what contrast curve I choose for the workflow I use. MlRawViewer, like Magic Lantern firmware, gives you lots of “pro” editing options. However, these options are only useful if you choose to export your .mlv to .mov format. That is not a part of my workflow. In fact, I haven’t even tried it for fun. Anyway, the 7th option on the left side of the screen has an arrow pointing down to the right at a 45 degree angle. Click on that to choose between exporting to .mov or .dng. Also useful for exporting to .mov are the boxes on the bottom right. One allows you to change the exposure of the clip, and the other the color (blue-yellow, green-magenta axes). There’s a white point sampler as well in case you’re super organized and on top of your shit and make a frame of your color checker.

Magic Lantern Raw Viewer 2

Back to my workflow, after viewing the clip, if it looks usable, I click the “+” button just below the arrow and DNG button. The application begins to process each frame as it was recorded in camera, makes a separate folder that contains the accompanying .wav audio file, and populates the folder with each frame that was captured as .dng. The application shows a progress bar, and upon completion your file has been converted to a (large) sequence of .dng still images.

Magic Lantern Raw Viewer 1

My only gripe with MlRawViewer is that I have to close it each time I want to work on a new .mlv file. When I have my finder windows open and the application open, I want to be able to double-click on the next .mlv file and have it open. Otherwise, this thing is gold and it’s free.

The next step is where things slow down quite a bit. I’m not running the fastest system (21.5″ iMac Late 2012, 2.7 GHz Intel i5 with 8 GB RAM, 512 MB graphic card), so this part takes some time to complete. And I certainly plan on upgrading to avoid the wasted hours (yes, hours of wasted time) lurking Instagram while waiting for this to complete.

I digress. I use Lightroom to edit color and contrast. I have used Lightroom almost exclusively for the last 7 years. It is incredibly powerful and I’m so used to it that I would have to sacrifice a lot of time to become as familiar with another editing/cataloging application as I am with Lightroom. In Political Scientist and Economist terms, this is known as “path dependence” (or path dependency). Lightroom does have its downsides, but for the most part it is the most effective application I currently have for what I’m trying to accomplish.

Using Lightroom CC, I create a new catalog for the clips. Note: I haven’t decided yet whether I should create a new catalog for each clip. I think that is the way to go because the more clips there are, the more .dng files, and anyone who knows Lightroom knows that a catalog of 10,000 image files, no matter what the resolution, moves as slow as a snail. Plus, if you’re shooting at 60 frames per second, you’re looking at double the number of image files than if you are shooting at a normal or “cinematic” frame rate. Alas, for yesterday’s shoot, I did not create catalogs for each clip, and here is why: I need to easily access each different clip to compare color and contrast. Opening each clip’s catalog is extremely time consuming, as Lightroom has to boot each time you open a new catalog. On the other hand, a lean LR catalog does open fairly quickly – much faster than a 10K image one – and you are able to sync file adjustments and export those images faster as well. It’s really a job-by-job decision at this point. This time, I chose to have all the images in one catalog, and I’m doing a lot of waiting.

File organization is key here. The clips generally come out of the camera in the order shot. Like I said, once converted to .dng, there is a new folder accompanying the original .mlv file containing all of the .dng images. LR’s import dialogue will only allow for import of the .dng files, and does not recognize the .mlv files. For this job, I imported all of the files (not the individual folders. Please learn from my mistakes and don’t do this. What a pain in the ass.). Each .dng went into a folder named the shoot date. You can see the problem here; the clips were not separated! So, I went through all 7,337 images and found the first and last frame from each clip, made selections accordingly, and put each complete clip into its own subfolder. This allows me to toggle between clips to compare contrast and color pretty easily. Using this strategy, I’m able to select multiple folders and apply the same color corrections to all of the interior clips, for example.

Another thing I have noticed on a couple different occasions using this workflow is that a few single frames will be distributed throughout the entire folder in error. For example, a single frame from one clip will end up in the middle of another clip. The way to work around this is to sort all of the images BEFORE putting them into individual folders. Click View>Sort>File Name. Totally takes care of the problem. Don’t forget this step!

Once the .dng files are edited for contrast, color, grading, etc…, I open the export dialogue and choose my file settings. Depending on your preferences this can also take up substantial amount of disk space. I prefer 16 bit files, so I export to .tiffs. I have exported to jpegs in the past, but almost always notice significant banding and other undesirable features that I don’t want in my finished product. Now, I understand anything I, or one of my clients, upload to the internet is going to be in 8 bit. I get that, but while I’m working on the entire piece, I want to have as much data to work with as possible. Maybe I’m a little paranoid, but coming from a photo background, and seeing the tragedies editing jpegs, this is a must for me.

I export to the source folder of the .dng files. I make a subfolder named “Processed Video Frames” and name each .tiff file but the clip it belongs to and start numbering at 1000. For example, “Clip 19-1000”. This ensures that a longer clip will not “turn over the odometer” for lack of a better term, and screw up the order in which the files appear in the subfolder.

File Organization

Once all of the files have been exported as .tif files, we are ready to move into the video editor. I am using Premier Pro CC. I’m mostly self taught (which is scary, yes), although I did learn how to make a time lapse, and this is workflow is pretty similar to that process. Upon opening Premier Pro, I choose to create a new project, or open project if I’m continuing work on another video. Follow the instructions in the dialogue and get into the editor.

Once in the editor, I click File>New Sequence. For the current project I’m choosing DNxHD>1080p 23.976 and randomly going with DNX 175X 1080p 23.976 because I really don’t know any better. Maybe someone can leave me a comment and help me out. Then I’m clicking File>Import…, finding my .tif files and clicking the first file, then clicking the check box next to “Image Sequence” and finally click Import. Now my still image sequence has been imported into my project, turned into a video clip, and I can begin arranging them within the sequence.

Like I said, this is still work in progress. I’m going to continue adding posts on this topic.

Update: I wrote this post last night, as I was processing some of the raw video files from the shoot. I’m really kicking myself for not capturing this footage at 60FPS, or even 30, for that matter. Even though I’m using DJI’s Ronin-M, some of my movements were a little too quick, and could have used a bit of slowing.

I hope this post was helpful to anyone looking for a different workflow using basic photo editing programs. Please leave comments/suggestions/questions regarding this workflow. I’m open to new and faster workflows that don’t use all of my CPU!

Thanks for reading,


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s