Creating Ultra HD (4K) video from RAW video shot with Magic Lantern

I recently created some Ultra HD video from RAW video I shot using Magic Lantern software on my Canon 5D Mark lll camera.  The video I created is 3641 x 2048 and has more detail than the RAW 1920 x 1080 video.  To complete this test project I upres’d the dng files from 1080 to 2048 in Photoshop RAW.  This blog post will explain how I did my work.

To shoot RAW video on my Canon 5D Mark lll, I use a super fast Transend CompactFlash 1000x 64GB memory card.  This write speed of this card allows my camera to record the high data rate needed to record RAW video.  I set my Magic Lantern settings to create 1920×1080 video, which is the largest size my camera and this card will allow me to record.  If I record my RAW video to an external recorder, or if I can find a faster card – Magic Lantern software may allow me to record larger than 1920×1080 RAW – but I have not tested whether or not this will work.

Each RAW shot is recorded on my card as a file that ends in .RAW.  I use software called raw2dng to convert my RAW file into a folder of digital negative files.  Each dng file is one frame of the video I shot and all of the dng files in the folder will be used to create a movie sequence. I drop each RAW file on the raw2dng software and it creates folders with dng files.

To open my dng files and to upconvert them to 3641×2048 – I followed a tutorial by RedDeerCityTV that I found on YouTube. The video explains how to set up Adobe Photoshop CC to use Scripts/Image Processor to select which folder you want to open, to open first image to apply settings, and to resize the images to 3641×2048.  RedDeerCityTV believes the dng file can be upres’d to a 4K file that is indistinguishable from native 4k because of the ability of dng files.

I altered this methodology a bit.  I dragged my dng files from one of my folders onto Adobe Photoshop CC icon.  The files will open into Adobe RAW – where I can color correct one of the images appropriately, and resize the image to  3641×2048.  I then select all, right click, and synchronize all of the images to be the same.  Then I Save Images to covert the dng files into TIF files.  I did not use 16bit color settings because I heard the quality is not much higher than the 8 bit files – and I save a bunch of file space.

I used Quicktime 7 to convert my folder full of TIF files into video by using Open Image Sequence.  The resulting video is 3641×2048 ProresHQ that I can import into FCP X and use to edit right away.  I found that the colors were not the same as how I set them in Adobe RAW but I was able to color correct them in FCP X.

I exported my master video from FCP X as a ProResHQ and uploaded it to YouTube.  I did upload an H264 version first, but it was blocky and compressed looking so I took the time to export and upload ProResHQ.  The video I created is below.  If you play the video at 1920 x 1080 and at 4K you will notice a big difference in the quality – especially if you play each version on a 4K monitor.  Please let me know what you think.  I did this project just to see what can be done with a 5D Mark lll and Magic Lantern and if you have suggestions please let me know.

Video Production with Green Screens

Several of my recent projects benefited from use of green screens. After shooting in front of a green screen, I can remove the green background in post and place any background behind my subject that I want.

One of my shoots was a promo video for SECRET OCEAN 3D, an IMAX film Produced by Francois and Jean-Jacques Montello and Directed by Jean-Michel Cousteau. I set up a green screen at the office of Jean-Michel Cousteau’s Ocean Futures Society so that we could record Jean-Michel and Sylvia Earle promoting the film.  After the shoot, the Montellos replaced the green background with scenes from the film.

Jean-Michel Cousteau and Sylvia Earle

Jean-Michel Cousteau and Sylvia Earle on the set promoting SECRET OCEAN 3D.

Jim Knowlton's Green Screen Studio

My portable green screen set includes lots of lighting and overhead mics for excellent sound.

I also shot another project with a green screen and Sylvia Earle at DOER Marine, Sylvia’s facility where she and her team make submarines.  I filmed Sylvia and her Deep Rover submarine with a green screen so that I could place Sylvia and her submarine in Bikini Bottom – the underwater cartoonland where Spongebob Squarepants lives.  Shooting with a green screen allowed me to add Sylvia in her submarine motoring underwater in a cartoon as she explained the beauty and fragility of sea life.  I will place a link to the short film once its released on iTunes.

Sylvia Earle's Deep Worker submersible filmed with a green screen.

Sylvia Earle’s Deep Worker submersible filmed with a green screen.

Jim Knowlton and Sylvia Earle

Director of photography Jim Knowlton and Sylvia Earle celebrate at the end of their shoot at DOER Marine.

Jean-Michel Cousteau and Sylvia Earle are amazing to work with.  Both of them are so positive and energetic and they inspire the people around them with their good energy and their message that we need to protect the ocean – our life support system.

Jim Knowlton records Jean-Michel Cousteau with a green screen.

Jim Knowlton recording Jean-Michel Cousteau with a green screen.

I also did some filming in my studio with my boys and a green screen – just for fun.  My boys are good sports and we had fun just goofing around.

Some tips for shooting with a green screen:
– Move your subject away from the green screen so that green light does not spill onto your subject.  If you don’t then you will need to spend extra time in post reducing the green spillover from your subject or parts of them will disappear along with your green screen.

– Light your subject correctly and separately from your green screen.  Once the green area drops off of your video you will want your subject to be well lit without shadows or hot spots.  Use lots of light on your green screen – the more the better to reduce shadows and to make the color of the green screen as uniform as possible.

– Stretch your green screen tight to reduce wrinkles and shadow that can complicate your post production work.

– Make sure your subject does not move outside of the green screen or off camera so that part of them is not cut off, such as a hand that moves off screen (you can see some examples of this in the video above with my boys).  It will make it harder to resize your subject and place them in a larger environment, such as a wide landscape.  For interviews, I make sure I do not crop off the top of their head, so that I can resize them later if I want without missing part of their head.


Pool Shoot with Live Video Feed to the Surface

I recently shot video and photos of a pool cleaner as it cleaned the bottom of a pool to create media a pool company could use to promote their products. It was a great opportunity to use the live HDMI output of my Canon 5D Mark lll (5DM3) camera. Canon recently updated their firmware to Version 1.2.3 and one of the upgrades is that the 5DM3 now allows video to be displayed on the rear LCD screen of the camera and also output from the HDMI port to a monitor. This new “mirrored” feature allows the shooter and a director to monitor the shots.


Cameraman Jim Knowlton uses live video feed to the surface on a Canon 5D Mark lll

The new firmware also sends YCbCr 4:2:2 uncompressed video out of the HDMI port to allow the signal to be recorded on an external recorder to get higher quality video than the 5DM3 records internally. For this shoot, I recorded in-camera using my fast CF cards and only sent the signal to the surface to get feedback from the director during the shoot.

To send out of my camera and through my Aquatica underwater housing, I used an HDMI bulkhead connector made by Dive and See. The bulkhead is installed into the housing using an already existing port and its mini HDMI connection is plugged into my camera. I also used a 100 foot long underwater HDMI cable made by Dive and See to connect my camera to an HDMI monitor on the surface.

Canon 5D Mark lll camera in Aquatica Housing with Dive and See HDMI cable to the surface

Canon 5D Mark lll camera in Aquatica Housing with Dive and See HDMI cable to the surface

I used a OTS Buddy Phone with Squelch to receive instructions and feedback from the director on the surface.  He used a SP-100D-2 Buddy Phone 2 Channel Surface Station to send his audio underwater.  The buddy phone allows the director and I to work together to get the shots the pool company needed for their commercial.  The director and the pool company were confident when the shoot was over because they saw each shot recorded as it happened and they knew that every shot on their shot list had been successfully completed.

If you need an underwater director of photography with live feed equipment and experience please contact me for more information.

Underwater director of photography Jim Knowlton

Underwater director of photography Jim Knowlton gives a thumbs up following a pool shoot with live feed to the surface

Using Magic Lantern software on a Canon 5D Mark lll

Jim Knowlton uses Magic Lantern RAW video

Two frames taken from video shot with 5DM3 H.264 video and Magic Lantern RAW video for comparison

I shoot with a Canon 5D Mark lll and I installed Magic Lantern software (ML) since ML adds a lot of functionality to my camera. ML software is a free software add-on that runs from the SD card slot and adds a host of new features to Canon EOS cameras that weren’t included from the factory by Canon. I have had no problems with this software though you install ML software at your own risk since it is not authorized by Canon.

I have been experimenting with shooting RAW video using ML software. ML RAW records video at 1080p 14bit color without any processing from the camera. My tests show that the RAW video is sharper than the H.264 video recorded by my 5DMlll (5DM3) and there is more dynamic range – meaning, I can get more details out of the shadow areas than with H.264. The 14 bit color information means that more color information is stored compared to the normal H.264 8 bits of color info.

I’ve put together a comparison of RAW vs H.264 in the following video. Make sure you watch it in 1080 for the best comparison:

Canon 5D Mark lll Magic Lantern RAW vs H.264

Notice the amount of detail in the RAW image versus the H.264 video – especially on the seat cushions and on the brick wall. You can see the details of the pattern on the seat cushion and the wall has more texture when recorded with RAW.

Notice the amount of detail in the shadow areas in the RAW image versus the H.264 video.

Notice the colors on the pillow next to my dog Lucas. They are much brighter and more saturated in the RAW image versus the H.264 video.

Technical Details:
The H.264 was converted to ProRes HQ and it was sharpened in FCP 7 using a level of 25. The RAW video was processed using raw2dng and Photoshop CC and no sharpening was applied. I regularly sharpen my 5DM3 video ever since I watched Phillip Blooms video with his recommendation to sharpen 5DM3 video.

If I did not sharpen the H.264 video the difference in sharpness between H.264 and RAW would have been very dramatic. But, since I believe the H.264 is best when sharpened, I did not provide an unsharpened sample for comparison. Canon uses filtering to get rid of moire problems, where distracting lines move around in video when filming some materials, such as when filming a patterned sport coat during an interview. The Canon filtering makes the image softer – which can be restored with sharpening. I have also heard that Canon thought that many HDTVs are over-sharpening video and that the image looks better on an HDTV without sharpening in post. I have done tests where I present on an HDTV without sharpening in post and I agree – 5DM3 video looks great on my HDTV without sharpening since my TV provides its own sharpening. Since my comparison video is being watched on YouTube I have added sharpening to compare the two formats as I normally use them.

I tried to color correct the scenes to be similar but making the color exact is not my purpose. I am mostly impressed with the increased sharpness and details in shadows.

I think the RAW results are fantastic. However, I would not use RAW for all my shoots since the RAW workflow is time consuming and because RAW video requires much more harddrive space to store. RAW video does not record sound – there could be a way around this that I have not found yet. So, I wouldn’t bother using RAW for an interview. Also, because I like the way people look using H.264 5DM3 video. The colors and details look creamy and film-like. I believe I will use RAW for timeless shots such as landscapes and for when I shoot underwater stock footage video clips. I have shot underwater RAW and I will post my results in a future blog entry.

I hope this post helps you try ML software. If you have suggestions for me or if you appreciate this post please feel free to post a comment below.

My gear:
Canon 5D Mark lll with Canon EF 24-105/4L
Transcend Compact Flash 1000x Card 64 GB

Some very helpful links if you want to try using ML software:

You can download Magic Lantern software here:

These websites provided excellent instructions on how to install ML software:

This website provided excellent advice on how to process RAW video:

especially this video: