Why RAW? The short answer is video that has a film look, natural light and color.
RAW video can provide a look and feel that consumer video can’t. It accomplishes this by saving the all the color detail it sees while recording. Consumer video, what I’ll call H.264 (the name of the most popular compression CODEC in use today) is designed to deliver a pleasing moving image—sharp, with rich colors—at a data rate that: 1) does NOT exceed most consumer electronic devices’ maximum bandwidth (usually under 4 megabytes per second); and 2) contains no more brightness, or darkness, than we can view on our display; and 3) when things are moving fast, throws out any information that we won’t notice, like the license plate number of a car in a fast-pan clip. In short, H.264, tries to get as much image quality in a low-data stream without our noticing. The more you get into video, the more you notice the small-data for image-quality trade-offs it makes.
Here is a frame taken from H.264 video shot with an EOS-M. Study these images closely. They are straight from the camera, on a light-table, with controlled exposure.
Here is a frame from Magic Lantern RAW shot with the same camera:
Notice how the “RAW” colors are evenly bright, unlike the H.265 where the more primary the color, the brighter it is (because compression adds contrast). You will notice more noise in the RAW image, but with it, more detail and dynamic range. You may say, but aren’t they both 8-bit images on my computer? Yes. If we pumped up the saturation and contrast on the RAW image it would look just like the other one. The difference is that RAW can move up the range of brightness while retaining color information while the 8bit cannot. In 8bit, so to speak, the colors “red-line” as soon as you pump up the brightness.
The total pixels in a frame of 1,920 pixels wide, and 1,080 pixels high, is 2,073,600, or about 2 million pixels. In one second, we watch 30 of those frames, so that is 2 million times 30, or roughly 60 million pixels per second. For a minute we’d need 60 million times 60 seconds, or 3,600,000,000 pixels per minute, or 3.6 billion pixels. Yes, when you’re watching your HD TV your eye is viewing 3.6 billion pixels every minute!
What makes up one of those pixels? A color of course. Colors can be described in their red, green and blue components. That is, every color can be separated into a red, green and blue value, often abbreviated RGB. When most digital cameras take an image each color is assigned a brightness value from 0 to 16,383 (14 bits). We need three numbers, red (0 to 16,383), green (0 to 16,383) and blue (0 to 16,383) to numerically describe ANY color. Some simple math tells us that we end up with a value that might reach 16,383 times 16,383 times 16,383 or 4.3 trillion As expected, a single 1080p RAW frame from a Canon camera is about 4 megabytes.
In the above images, the H.264 frame can be pulled out as a 114k JPEG. The RAW frame, a 256 JPEG, originated from a 2.4 megabytes RAW file, which means you can choose less contrast, or more detail and noise.
Even 8-bits (1 byte) per color “channel” is enough to create 24-bit (8+8+8), or 16 million colors. The human eye can see about 12 million colors at best (so we don’t need those 4.3 trillion “RAW” colors). That allows an H.264 to throw out over 96% of the original pixel data.
A consumer video camera can quickly figure out what we can see, and not see, in an image, so this isn’t difficult. It takes the “brightest” data and saves it–AND THROW OUT THE REST. However, the overall brightest image is not always the image we want! Sometimes we want a dim image with a lot of detail in the shadows.
Let’s go back to the optimum image we’d like to see, 3.6 billion pixels per minute times 24bits (3 bytes). That would be 10.8 gigabytes per minute. As you know, you’re not streaming 10 gigabytes of video to your TV every minute. Video compression does a marvel job of cutting that down to a manageable size
HD 720p @ H.264 high profile 2500 kbps (20 MB/minute)
HD 1080p @ H.264 high profile 5000 kbps (35 MB/minute)
It is the limitations of our computing devices that we can’t have what we really want–10 gigabytes of video data every minute. If for the sake of argument we had unlimited storage and speed we’d all save and view images without compression. That’s when they have the greatest fidelity.
Consumer video cameras record video using a “distribution” CODEC, not a photographic storage method. This means they’re making an immediate decision of what part of image to save, and what to throw away. The top image is what they end up with. The bottom image is what the sensor recorded BEFORE being put through H.264.
The benefit of RAW video to me, is that I can decide how to compress the image after it has been taken. I can make the decision of what the image should look like. I can get a photographic look.
The new 4K cameras coming out will offer more resolution (4 times more), but resolution will not give me more color depth. That isn’t to say 4K is phony; only that it doesn’t fix the color-depth problem inherent in consumer-level compressed video streams. 4K RAW is a different matter, of course.