Cinematography

Cinema Raw – What’s Lost in Compressed and What’s Gained in Raw

Canon 5D Mark III Compressed vs. Raw

Magic Lantern, a software devised by hackers, created a code that gets placed on the memory card of the camera (and engages when you activate the firmware update of the camera—although it doesn’t actually alter or update Canon’s firmware). In essence it embeds a software interface that allows the user to tap into the potential of the camera—a potential that Canon refuses to give users access to within their software interface. Some of these changes include presenting functional overlays—from audio meters to histograms—that make the Canon DSLRs better, in many ways providing features you would find on a normal high-end video camera. The software makes the Canon DSLRs much more user-friendly, and Magic Lantern continues to improve the software. In May 2013, they announced the beginning stages of recording raw video uncompressed from the sensor data—creating 14-bit raw files that can be converted (or wrapped) into the CinemaDNG format. For those who already own Canon DSLRs and for those who don’t want to purchase a raw cinema camera, this software update—although requiring technical savvy to make it work—becomes another option for independent filmmakers working on a tight budget.

Alec Weinstein, a filmmaker based in San Diego, started out his career as a business consultant for over ten years, but when the Canon Rebel T2i (550D) was released in the first quarter of 2010, he bought one and made a short film with it (along with some friends). It got the attention of some television cinematographer types and within a few weeks of his short film’s release, he was offered commercial work within television. Needless to say, he quit his job and “dove head first into professional camera work,” he tells me in an interview. He uses Canon’s 5D Mark II and III doing network-level work at CBS, NBC, Fox, Bravo, A&E, VH1, and Style. Weinstein does shoot with higherend ENG video cameras, but he says he prefers the DSLRs for their “versatility and ability to shoot superior imagery.”

Using a Canon 70-200mm L series lens with a two times extender adapter (making the zoom lens a 140-400mm lens), Weinstein placed this gear on his Canon 5D Mark III, taking footage of palm trees and the ocean at an effective focal length of 400mm. He shot nice 5D Mark III “standard” footage in the normal 8-bit H.264 compression mode of the camera. He then compared this footage to similar footage shot in raw with Magic Lantern’s software, circumventing the limits of the camera’s 8-bit compression, allowing him to record 14-bit raw on a high-speed CF card.

Below are a series of frame grabs from the video shot by Weinstein described above, showing the difference between video shot in raw in contrast to the camera’s normal 8-bit H.264 compression. See Figures 2.13–2.17.

FIGURE 2.13 - Frame from 5D Mark III H.264 footage. The frame reveals the original 8-bit compression. It expresses a good DSLR cinema look. (© 2013 Alec Weinstein. Used with permission.)

FIGURE 2.14 - Frame from 5D Mark III raw footage. You can see sharper images, such as within the leaves and ocean in the background. The color is obviously different, appearing richer than the 8-bit compression. (© 2013 Alec Weinstein. Used with permission.)

FIGURE 2.15 - 5D Mark III H.264 color corrected for saturation and contrast. It reveals the warmth found in the 14-bit raw version and at first glance, looks pretty close— not a vast difference between the two. (© 2013 Alec Weinstein. Used with permission.)

FIGURE 2.16 - Now Weinstein takes the raw footage coming out of the Canon 5D Mark III and provides a contrast and saturation color correction. Already it’s looking stronger than the corrected compressed look. If you still can’t quite see the difference, he brings it home in Figure 2.17. (© 2013 Alec Weinstein. Used with permission.)

FIGURE 2.17 - A side-by-side comparison of Weinstein’s footage comparing the color graded raw version with the H.264 corrected compressed version. This is where we really see the difference. While the standard compressed footage—by itself—looks good, it lacks the sharpness and richness of the raw footage. (© 2013 Alec Weinstein. Used with permission.)

Side by Side: 8-bit Compression vs. 12-bit Raw

Marco Solorio of OneRiver Media demonstrates this in a series of tests comparing normal compression of 8-bit Canon 5D Mark III to 12-bit images from the Blackmagic Cinema Camera.3 See Figures 2.18 and 2.19.

FIGURE 2.18 - Solorio’s comparison of the BMCC (16-bit linear to 12-bit log uncompressed) to the 5D Mk III (8-bit compressed). He notes with arrows how the details of the grass and roof shingles are less defined in the 5D when compared to the BMCC. He also points out how the color of the sky and banding issues appear on the 5D, but truer colors appear with no banding for the BMCC. (© 2012 Marco Solorio/OneRiver Media. Used with permission.)

FIGURE 2.19 - In a night shot lit with practicals, Solorio notes how the colors and definitions of lit signs are washed out (Mel’s drive-in signs) in the compressed image. In addition, he points an arrow at a blown-out light in the diner for the 5D, but when compared to the BMCC, the highlights are held and the shape of the light is defined. Notice the practical lamp to the right of the “Mel’s drive-in” sign on the far right. On the 5D sign, the white light blows out. On the BMCC side, we see gradations of light detail. (© 2012 Marco Solorio/OneRiver Media. Used with permission.)

When you’re shooting with a compressed format, such as with the Canon 5D, Andrew Cochrane of Guilllermo Del Toro’s studio Mirada explains, you’re getting about five stops of usable latitude:

So your exposure needs to be right—you can’t miss it by a stop. In the H.264 compression color space—and it’s pretty heavily compressed—you don’t want to change your image very much in post. You want to use filters. You want to nail your stop and you want to nail your look, you want to set your white balance. You want to look at your LCD and have it be at 90 percent final image—if not 95 percent—because that’s how much wiggle room you have in post before the image just goes to hell.

Fundamentally, there are two camps of filmmaking, Cochrane explains:

There’s film and there’s digital. Film’s all about the shadows. Digital’s all about the highlights. The reason a lot of people have had trouble transitioning to digital is because the mentality is 180 degrees from film—you’re protecting your shadows, when you should be protecting your highlights in digital.

This can be astutely seen in the next section.

POSTPRODUCTION ADVANTAGES WITH RAW

Where there is no subtlety between compressed vs. raw footage is in the post -production world. The 12-bit CinemaDNG raw footage excels in post-production—when you need to change exposure, engage color grading, to really change the look and feel of your film. In the 8-bit compressed world of images, you need to get the look of the film as close as possible in-camera. If you shoot in a “flat” look format (such as Canon’s cinema log), then you are giving more leeway in post to change the look of the film—but it still doesn’t come close to the work you can do in post with 12-bit uncompressed files.

It’s important to note the difference between linear and logarithmic encoding of images. The linear process applies equal values to each bit of color data— linear. The log is a logarithmic curve, where the bits of color data can be spread out, allowing for a natural simulation of how film (and the eye) responds to light. While the film or cinema settings on cameras record a “flat” look, it’s creating a log space for getting cinematic-looking images, while the Rec 709 setting defines a broadcast standard look. One is designed to engage color grading, while the latter is for expressing a realistic look without using color grading.4

In the creation of a day for night scene, Solorio takes us through the process, showing how easy it is to manipulate the image in post without any degradation of the image.5 See Figures 2.20–2.22.

FIGURE 2.20 - The original day image, shot on the BMCC in 16-bit linear to 12-bit log raw. (© 2012 Marco Solorio/OneRiver Media. Used with permission.)

FIGURE 2.21 - The raw CinemaDNG file adjusted day for night using menu settings in Adobe Camera Raw.(© 2012 Marco Solorio/OneRiver Media. Used with permission.)

FIGURE 2.22 - The same image with compositing effects of stars, moon, and glow effect. (© 2012 Marco Solorio/OneRiver Media. Used with permission.)

This shows the power of shooting with 12-bit raw. In many situations, no such effects are needed and shooting 8-bit compressed files—such as in documentary or news shooting situations—will be the right choice, since having the space to record needed footage and edit it quickly is more important than lacking space and dealing with the time needed to convert raw files when shooting uncompressed. However, another case where shooting raw can reveal a decisive advantage occurs when you need to recover detail in the highlights and darks. Again, Solorio shows us this advantage later in the same video referenced above. See Figures 2.23–2.27.

Another form of postproduction that can be applied to CinemaDNG files is film grain. For some, the idea of shooting in raw reflects in some ways an approach to filmmaking that harkens back to the days of working on film. What is lost in digital filmmaking involves the loss of texture. Film grain is one way to help bring texture back to the filmmaking process.

FIGURE 2.23 Here’s Solorio’s original 8-bit decompressed H.264 image (likely converted to ProRes 422). Notice the blown-out highlights in the window as well as the dark corner of the room. (© 2012 Marco Solorio/OneRiver Media. Used with permission.)

FIGURE 2.24 - Solorio’s comparison of 8-bit H.264 compression, 8-bit decompressed, and 12-bit raw. He’s opened up the exposure four stops in order to reveal details in the shadowed corner. Everything looks similar— which shows again that the differences between the 8-bit image and the 12-bit image are subtle.(© 2012 Marco Solorio/OneRiver Media. Used with permission.)

FIGURE 2.25 - Here’s where we see the real difference. When adjusting for the blown-out window, Solorio’s 8-bit files turn gray—there’s no recovering of clipped highlights in the 8-bit world. However, the 12-bit raw image recovers the clouds and blue sky, providing a color corrected image that really shines. Notice how the details in the lower half of the curtain are also recovered in the raw image. (© 2012 Marco Solorio/OneRiver Media. Used with permission.)

FIGURE 2.26 - Solorio layers the corrected 12-bit raw file for the window and the dark corner (he exposed them separately and adjusted accordingly). (© 2012 Marco Solorio/OneRiver Media. Used with permission.)

FIGURE 2.27 - As for the argument that web videos don’t need to worry about utilizing raw, because it’s so compressed, Solorio begs to differ. Here he shows the compressed YouTube version of the 8-bit version contrasted with the YouTube compressed 12-bit source. (© 2012 Marco Solorio/OneRiver Media. Used with permission.)

Excerpt from Cinema Raw: Shooting and Color Grading with the Ikonoskop, Digital Bolex, and Blackmagic Cinema Cameras by Kurt Lancaster © 2014 Taylor and Francis Group. All Rights Reserved.

Related posts:

0 Comments
Tell us what you think!
*

Latest Tweets

Stay Informed

Click here to register with Focal Press to receive updates.


about MasteringFilm

MasteringFilm, powered by bestselling Routledge authors and industry experts, features tips, advice, articles, video tutorials, interviews, and other resources for aspiring and current filmmakers. No matter what your filmmaking interest is, including directing, screenwriting, postproduction, cinematography, producing, or the film business, MasteringFilm has you covered. You’ll learn from professionals at the forefront of filmmaking, allowing you to take your skills to the next level.