Now that I think about your post here I think I'm seeing where my confusion is. If the logC profile is created from the full range the sensor detects doesn't that sensor data have the inherent inaccuracies at the low and top end exposures of the image? Am I wrong in thinking that LogC is somehow keeping the sensor from working in the top and bottom ends of exposure where it is less accurate? I'm beginning to think I misinterpreted that in reading up on LogC.
From what you just posted it seems that it is the handling of the sensor data that is being remapped into the logC format which compresses the data using a log formula rather than a linear formula. Does the logC remapping of the data do more than just compress the signal so it's not getting clipped on top or bottom. I realize the mathematical exponential nature of logs allowing for a greater range of numbers in a smaller space but is the LogC also somehow accounting for the nonlinearity of the sensors at the top and bottom too?
If I were to hypothetically shoot a chart with 1024 steps of grey perfectly exposed would that come out the other end of the LogC food chain with the 1024 steps? My gut says that when I end up with a washed out narrower range of level on the image that means there are fewer of the original 1024 steps available to be stretched out back to a full 1024 steps of grey. Perhaps my quandary is more about overall image range vs. detail. I'm not saying LogC is bad and reading that it does take longer to correct makes me feel better because I do have to tweak it more and that takes more time and I was thinking I must be missing something if it takes longer to correct. I was confusing better vs. easier and quicker. Unfortunately the expectations from the bean counters is it's better so it must be faster and easier.
---In Avid-L2@yahoogroups.com, <Greg@...> wrote :
On Oct 12, 2014, at 18:23, bigfish@... [Avid-L2] <Avid-L2@yahoogroups.com> wrote:
Well most of my dvcproHD footage peaks at 110% and might go a little below 0mV but it's not terribly clipped so I don't have a big problem with that. Now how is the exposure contained into a smaller space? Is the same amount of light hitting the sensor but the circuitry creates the log C? If it's the extremes of the ranges that are hard on the sensor I would think it's the light hitting the sensor that has to be tweaked before it hits the sensor, but how would that be done? Is there some kind of flashing equivalent for video sensors?
I'm no camera operator so what happens in the camera when you switch to logC? Does the viewfinder suddenly look all washed out etc..?
Therre must be something fundamental I'm missing here in that to me it seems like I'm using less of the range of the sensor expecting to pull more range out later. Is the only real benefit that the top and bottom don't get clipped? If the issue is that the top and bottom aren't as accurate, disregarding clipping, then wouldn't it make more sense to use the full range and alter the gamma etc... for the top and bottom?
Sometimes I think I spent too many years editing 3/4 to 1 inch and I just have a built in noise ignorer in my brain.
---In Avid-L2@yahoogroups.com, <Greg@...> wrote :Yes, you're amplifying noise. What makes log footage nice to work with
is YOU, the colorist, gets to choose what's thrown
away/deleted/compressed/clipped, on a shot-by-shot basis. I'll trade a
little noise for recoverable detail almost any day. When the shot's
low-contrast, and under-exposed in LOG to boot - well, that's when the
trade-off seems questionable - but about 85% of the time, the poorly
shot LOG is more gradeable than a poorly shot Rec709.
At least with c-log form a C300, in resolve, I use a touch (or more) of
NR on almost every shot. I suppose in media composer
I have not been happy with the Canon LUTs in Resolve - maybe I'm using
the wrong ones? - so I don't use them much. The black magic luts seem
better, I use it about half the time on a BM shot, and almost always if
the shot is 'raw' (dng.) The Arri LogC lut is excellent, when I do get
arri footage I will almost always put that lut somewhere in the nodes.
Mostly in Avid I've only used the Arri luts. Because of where Avid puts
the LUT - before any color correction - I find the ability to use luts
less than, well, terribly useful, because you can't adjust the shot
before it goes through the lut. Since the color correction is... let's
say 'concatenating,' you can't always recover shit that the LUT clips.
Overall, I'd much rather have a little noise than clipped highlights or
shadows. Yes, a perfectly composed and exposed shot, with
709-appropriate contrast levels, will probably be less noisy than the
same shot graded from a similarly properly exposed LOG shot on the same
camera using the same recording format, but in the real world, I'd much
rather correct log - it gives me a fighting chance.
gh
-------------------------------------
Greg Huson
Chief
Greg (at) SecretHQ.com
www.SecretHQ.com
DigitalServiceStation.com
> -------- Original Message --------
> Subject: [Avid-L2] Lots of LUTs?
> From: "John Moore bigfish@... [Avid-L2]"
> <Avid-L2@yahoogroups.com>
> Date: Sun, October 12, 2014 2:05 pm
> To: Avid L2 <Avid-L2@yahoogroups.com>
>
>
> I haven't jumped onto Ver. 7 and above yet. I'm waiting to finish a current series. My googled understanding of LogC and other log shooting is to center the exposure/level around the center of the overall energy of the light. I read the 14% grey, or whatever the proper units are, is the 50% energy point so LogC centers around that and keeps the exposure away from the bottom and top where the sensors are less accurate, not sure if that would mean less linear given things are exponential in LogC. Now when I get this into Avid it's washed out and I have to lower the blacks and raise the gain and noodle around to my liking. This is what the LUT would do ahead of time if I was on V7 and above. When I find myself adding 175 and more to the gain I start to wonder if using that much gain is adding noise. If it were a regular shot and that dark and I had to add gain it would most definitely add noise so is adding gain to a LogC shot any different?
> Electronically I don't see how it could be. Also the noodling around I assume is to approximate the "S" curve of the LUT which is a bit of an in exacting science. I would assume with the proper test charts it could be more analytical but I'm doing it by eye and waveform.
>
> Bottom line is if I have to lower the setup and add this much gain has the image really been improved by LogC? I'm sure it is given the adoption but I'm trying to understand the nuts and bolts better in my head.
>
>
> John Moore
> Barking Trout Productions
> Studio City, CA
> bigfish@...
Posted by: bigfish@pacbell.net
| Reply via web post | • | Reply to sender | • | Reply to group | • | Start a New Topic | • | Messages in this topic (6) |
No comments:
Post a Comment