Raw vs Compressed
This topic applies so much to both videographers and photographers alike that I’m hoping this will address some of the confusion. The principles are the same between the video and photo worlds, but the massive data rate of video means a slightly different approach.
Raw video is raw sensor data, not designed to make an intelligible image. It’s often chroma subsampled at the sensor level, meaning a “Bayer” array sits over the sensor restricting certain colors of light from registering at every pixel. The “de-bayering” and raw to readable conversion can happen within the camera or on your computer. The latter method provides more control because you have access to significantly more data (at significantly higher file sizes). Raw is no more an acronym than the raw meat in your freezer is.
Compression is a way of taking your data and optimizing it to retain everything you need to see visually and nothing more. This means if you want to alter hues or make dramatic exposure changes in post you’ll likely run into problems because the data has been so paired down. Compression can happen across a frame by averaging similar pixels into blocks and can happen between frames by only updating portions of the image that have significant differences from their surrounding frames. This is why you hear people complain about mushy “macro blocking” or editing “long-GOP codecs”. These are the pitfalls of compression.
Many image elitists (particularly those who don’t shoot much) naysay compression but it’s something to be aware of more than avoided.
Redcode Raw from RED in an amazing, wavelet-based video compression format, that, while lossy, is still one of the best innovations I’ve seen in the video world. RED developed a way to work with high resolution video four times the size of the then-current standard.
Common video codecs (a word conveying the process of _co_mpression and _dec_ompression) include:
Each manufacturer has their own proprietary raw format: CR2, NEF, ARW, etc.
Video: MP4, XAVC-S, XFAVC,
DNG, Canon Raw Light, R3D
Log is the new raw
Log is NOT raw, but, especially once people started shooting raw, the large data rates made them wonder if it was worth it. Surely there was a way to smash a lot of visual information into a smaller format? Turns out this problem had been solved long ago when we faced the same problem of getting all the meaty information available on a film frame into the limiting technologies of video. Cineon’s log standard is still very much in use today.
Log is more than a “flat” picture profile. You can add contrast to a flat image, sharpness to a soft image and saturation to a muted image more easily than the reverse. If a contrasty picture profile in your camera is clipping shadows and highlights then that data is lost. At a certain point in the “DSLR revolution” this was extremely popular. Most cameras today have a provision for shooting with at least a softer highlight roll-off so whiles don’t instantly clip.
Log means getting MORE right in camera
Shooting raw gives you more ability than you should have to save yourself in post. Modern sensors allow you to resurrect a drastically overexposed shot with very little noise penalty. You can resurrect a large amount of highlight detail (assuming all color channels aren’t clipped), lift shadows, and even set white balance in post on a raw shot. I cannot stress enough that shooting log means the opposite. Shooting log means you’re still compressing data and because of that you need to get your settings as close as possible to optimize the given scene for the format. This is especially true when shooting in 8 bit. See more information on working with the Sony cameras [here].
Proxies see my post on “Post Workflow for Videographers” [here].