I’ve been playing around with C-log quite a bit over the past few days, so far only in AVCHD, and I’m a little bummed out with the results. Even a minor amount of grading seems to wreak noisy havoc on the image, which kinda defeats the purpose of using C-log to begin with. I’m keeping my ISO around 850, occasionally knocking it up a bit. I find that I’m getting a better image when shooting WideDR and making lighter grades versus shooting C-log and making any grades at all. I’m swapping out my Ninja 2 for a Blade, so haven’t been able to play around with grading ProRes footage from the C100 yet. I guess my question is – is it even worth shooting in C-log if you’re using the AVCHD codec? I know that the answers will probably vary based on your quality needs and what platform you’re shooting for, but my general goal is to get the best image possible. On a side note, how well does WideDR grade with ProRes? Having shot with a Canon XH-A1 for the past five years, I’ve gotten really used to perfecting my image in-camera, so I guess I’m still working in that mindset, which is another reason I’ve enjoyed using WideDR.
C100: is it worth using C-log with AVCHD?
- Thread starter jthabel
- Start date
remsyatassi
Well-known member
I don't have an issue with Clog in AVCHD. You need to expose to the right and drop shadows in post. If you raise mids it will look very noisy and thin. Better to lose a little highlight detail than underexpose the mids.
ProRes helps a little, but not very much in my experience.
ProRes helps a little, but not very much in my experience.
Eric Coughlin
Veteran
I find Wide DR to be the best profile in my experience in terms of image quality. C-log is too noisy for me once graded to a normal image.
I find if you over expose the C100 to try to lower the noise the image looks washed out when you bring it back to normal, so it's best to expose properly for natural looking colors.
I find if you over expose the C100 to try to lower the noise the image looks washed out when you bring it back to normal, so it's best to expose properly for natural looking colors.
ReelWorksMedia
Well-known member
I only shoot C-Log. Pic profiles are of no interest to me. I have been very happy with the results and how the AVCHD holds up. That said, I usually capture to an external prores recorder. The results I am getting gave my RED cam a run for its money. In a lot of environments, the C100 is better.
AndreeOnline
Veteran
I would likely almost always shoot Canon Log if I still had my C100. The nature of the Log curve gives the image characteristics that are above normal video, in my humble opinion. It might of course depend on what you are shooting and what look you are after.
While you shouldn't under expose it, I also would not recommend over exposing Canon Log. The brighter parts of the picture profile flattens out and you'll lose some tonality if you put the meat of your picture there. That said, don't be afraid to let highlights and lights blow out, if that is the result of getting a face at the right exposure level.
It is a misconception that just because you are shooting log you should try not let anything blow out in the image. If your scene has 15 stops of DR and your camera has 12... well... some filtering has to happen or things will blow out (also fine).
While you shouldn't under expose it, I also would not recommend over exposing Canon Log. The brighter parts of the picture profile flattens out and you'll lose some tonality if you put the meat of your picture there. That said, don't be afraid to let highlights and lights blow out, if that is the result of getting a face at the right exposure level.
It is a misconception that just because you are shooting log you should try not let anything blow out in the image. If your scene has 15 stops of DR and your camera has 12... well... some filtering has to happen or things will blow out (also fine).
This, it works absolutely fine, great in fact. Would rather a bit of neat video than take my camera off C-log
So I know this gets asked a million times, but I still don't have a handle on it. How do you c-log enthusiasts set exposure? Especially during run and gun? Do you enable the view assist on the LCD and just eyeball it? Since the zebras don't go below 70%, it makes it hard to identify the IRE value of skin or a grey card.
How do you know when you're exposing skin correctly?
How do you know when you're exposing skin correctly?
Good question, drummondb. I'm a little lost on this too. I tend to expose most shots up to "just under zebra," unless there are excessive hot spots outside the center of focus, in which case I make the necessary judgment calls. But with C-log, I'm unclear about how to expose a shot, considering I'm going to very likely be upping exposure somewhat in post.
remsyatassi
Well-known member
Exposing to the right does not mean overexposing. But I would rather have things run a little hot to get the right exposure for my subject. Underexposing looks much more washed out in my experience.
Eric Coughlin
Veteran
I'm not terribly familiar with the terminology, but my guess is that the term and use of "exposing to the right" generally applies to cameras that don't have highly compressed codecs. Let's just pretend for a moment that the C100's codec was so terrible that you couldn't adjust the image in post at all, or perhaps a more realistic example, your handing the footage directly to a client who is not going to do any post adjustment; in that case, exposing to the right would be overexposing. And to what I'm saying, if you expose higher than what you want your final result to be (after grading), then the post adjustment will take its toll on the image (due to the poor codec) to the extent that it is essentially overexposing, in my experience.
My experience is that overexposing looks washed out and clean and underexposing looks less washed but still washed out and is noisy (but since it is less washed out and retains detail in the shadows it generally looks better than over exposing), and exposing properly so as little as possible post adjustment is needed on the poor codec is generally the best way to go.
Finnegan
Well-known member
Canon's white papers show that middle grey (18%) comes in around 32.79 IRE on the waveform when shooting Canon log on a C300/C100, while 90% white comes in at 62.74 IRE. With that knowledge, you should be able to more accurately expose your subjects using the waveform monitor. I've found 40-50 IRE to be a pretty solid range for exposing subjects with bright, white complexions without muddling up the mids and blacks.
Here's an example of a web series I shot on C300 where I do just that:
Here's an example of a web series I shot on C300 where I do just that:
Last edited:
This is where I, as a C100 mk1 and 2 user, remain confused. On the C100 Mk1, there is no way to tell the percentage of IRE on the waveform. I can MAYBE tell 50% by finding the middle line, but I sure as hell can't tell what 32.97% is. Perhaps the Mk2 has some percentage indicator on the spot meter? But I don't think so. Hell, I don't even think the mk1 has ANY numbers on the waveform. I have a hard time telling where zero is, is that 100? 110? A frame above the 110? AAAaaaa!
One more thing Finnegan. Please take this in the spirit in which it is intended: Your IGN video above is an entertaining, well thought out, well put together bit of corporate video. But I would not hold that up as an example of proper exposure of skin tones. Maybe this is a case of c-log not being necessary? Or in fact a case of it doing more harm than good?
Last edited:
AndreeOnline
Veteran
"Expose to the right" I would say is primarily a raw footage technique. The idea is to maximize the signal strength relative to the noise floor. With raw footage this is fine since you can pull the image data afterwards and it does not have any baked in curves to consider.
If you expose skin in the higher regions of a log curve you run the risk of getting flat skin with few discrete brightness values.
I guess I did some testing at first, but I see no problem at all with exposing log simply by looking at the image on your camera's monitor. This assumes that your monitor uses a fixed brightness and you have done some testing to know what the footage will look like once ingested. Waveforms are ideal, and I like to use them, but I don't find them necessary for most things.
I think the real trick is to just start shooting a bit of Canon Log and then play with it. You'll quickly learn how to handle it.
If you expose skin in the higher regions of a log curve you run the risk of getting flat skin with few discrete brightness values.
I guess I did some testing at first, but I see no problem at all with exposing log simply by looking at the image on your camera's monitor. This assumes that your monitor uses a fixed brightness and you have done some testing to know what the footage will look like once ingested. Waveforms are ideal, and I like to use them, but I don't find them necessary for most things.
I think the real trick is to just start shooting a bit of Canon Log and then play with it. You'll quickly learn how to handle it.
AndreeOnline
Veteran
I don't know Sony that well, but if I remember correctly they shoot at ISO3200 to maximize DR. I'm guessing there is some internal pull-pushing going on to get to "14 stops". Something similar to what the 1Dc does, but 3 stops more (Canon 1Dc uses 2 stops ISO400 as base ISO and I think Sony is 5 stops ISO3200?).
More importantly the A7s shoots 10bit. 1024 code values allow for a lot more flexibility compared to 256 for 8bit when it comes to stretching the image in post.
My responses in this thread relate to C100 AVCHD.
Pretty sure it's regular 50Mbps 8-bit 4:2:0.
AndreeOnline
Veteran
Yes... you actually wrote h.264 above. I guess it's just the external uncompressed signal that's 422 10bit. I was thinking about the external recording—it's what I've read most about—but you were clearly talking about internal.
I would be very hesitant in shooting Slog into 8bit. It's very flat from what I've seen. Seems designed for 10bit+. Anyway, I'd leave it up to actual Sony users to make recommendations for that.
remsyatassi
Well-known member
When I said ETTR I meant in relation to Canon's white paper. I expose Caucasians around 60-65 IRE and it looks very clean and natural. I'm not advocating pushing skin tones up to 80 or something, but if you read Canon's white paper you'd be looking at more like 45-50 which is a recipe for noise. I think we've beat a dead horse here so that's all I've got to say on it. YMMV
No, it's 8-bit 4k external. Technically if you downsample that it's something like 10-bit HD but I'm not really sure what difference that makes. The codec on the a7s is extremely robust though. The real issue with Slog is ISO 3200. Codec can handle it, but it's just noisy. That camera requires a hotter signal too because of the noise floor. Source: I own and shoot on one regularly.
No, it's 8-bit 4k external. Technically if you downsample that it's something like 10-bit HD but I'm not really sure what difference that makes. The codec on the a7s is extremely robust though. The real issue with Slog is ISO 3200. Codec can handle it, but it's just noisy. That camera requires a hotter signal too because of the noise floor. Source: I own and shoot on one regularly.
AndreeOnline
Veteran
ETTR most of the time means putting the histogram as far to the right as possible before the image clips, regardless of content. But like you say: no need to beat it further.
Ah.... I'm probably filing FS7 and A7s stuff internally in one bucket and mix it up.
Ah.... I'm probably filing FS7 and A7s stuff internally in one bucket and mix it up.