wsmith96 said:
IglooEater said:
wsmith96 said:
For those of us not educated in video related features, what is C-Log and what does it do for you?
I hope someone else more competent than me answers this one, but here goes:
Ever tried editing JPEG's, especially pushing and pulling highlights and shadows? They break down incredibly fast. That's because they are compressed and don't have enough data in them to handle that kind of manipulation.
Same with compressed video. One solution is to shoot RAW video, as RED cameras do. That comes with astronomically large files. The other solution is to use more of the data in your compressed file in the areas you know you'll want to push and pull (i.e.: the shadows and highlights) and less of the data where you won't need to do such manipulations. So basically you get the ability to apply some edits without compromising file size. If I understand they call it a Log because of the mathematic formula they use to assign data to the tonal range there's C-Log, S-Log, (Sony) and others.
One downside is that the footage is rather greyed-out right out of the box and will
require grading to become useable. (same goes with raw footage incidentally)
Got it - Thanks!
That's kinda it
-- Most video video camera encode their data using a "linear" form that does a couple of things, it assigns many more code values to the tones above middle grey than those below middle grey (the quantity of encoding values, in theory, gradations, increases in the same way light does, doubling with each stop as exposure increases. So the darkest shadows have only a few possible gradations whereas the lightest "in gamut" highlights have hundreds or thousands of shades depending on bit depth(again, in theory). Typical linear encoding also ads a curve with the specific goal of creating 6-7 stops of consistent contrast in the middle of the tonal range, compressing those tones above and below that middle range. The purpose for this scheme it to make the footage look right on a typical Rec.709 display. (6-7 stops of range is also about what is possible to print on the high quality photo papers as well).
Log Gammas, in general, assign "relatively" equal amounts of code values to each stop in the available dynamic range of a captured image, and in the process, greatly expand the middle "straight line" portion (that 6-7 stops mentioned above) to 11, 12, 13 stops and more.. This produces an image that is extremely flat, without the "normal" contrast we are used to seeing in a standard linear gamma. The benefit to this type of gamma, especially in 10 or 12 bit versions, is that we have much more flexibility in the shadows and highlights (at the expense of some gradations in the middle of the curve). C-log was designed for 8-bit cameras, and so it fudges its log by slightly compressing and lifting the shadows to allow for a greater highlight range. It's very easy to grade, even without a LUT and actually looks pretty good untouched.
It's of course a little more complicated than all that, but here's a graph showing the code value distribution of various Canon Log gammas. Note C-Log2, which is a more traditional cineon style Log curve, and its long flat shape -- which displays the essence of what a log gamma is, versus the original C-Log, with its compressed, lifted shadows, and more sloped strait line portion, compared to the Rec.709 curve that the current 5dmark IV outputs.
