This IS the Avid L2: Re: [Avid-L2] Canon log profile, wtf?

Thursday, September 12, 2013

Re: [Avid-L2] Canon log profile, wtf?

The way I think of this (initially coming from an audio background) is the same way I think of a double-ended noise reduction system. Analog magnetic tape was a great recording medium back in the days when it was being developed, but in those same times, one of the main gripes that people had about it was the noise floor. Tape hiss, and how you always heard it, even when you tried maximizing the signal level before distortion, and even more evident in multi-track recording, where the hiss is added per track. It builds up, and isn't desirable. So the geniuses at Dolby came up with this way to game the audio spectrum going in to the part of the tape that could 'hear' more of the signal (more headroom). The noise reduction system they developed was a double-ended one. Whatever signal you wanted to record to tape went through the input electronics of the noise reduction system first before hitting tape. This input path would do a few crazy things. It would boost the level to a point that would normally have overdriven the tape, but it would also 'squeeze' that signal's frequencies, steering them away from low end and toward high end. This is where the tape's greater headroom lies. This skewed signal was recorded to tape. If you listened to it as is, it sounded awful. That's where the second part of the double-ended system came in to play. Upon playback, you passed the 'encoded' tracks through the noise reduction decoding section. This effectively stretched the signal frequencies back out over the spectrum, and then by lowering the level to the normal recording levels, you also lowered the tape hiss. Voila, noise reduction.

Log recording is doing something different, but very similar in ways. It is also a double-ended system. The sensor sucks at low level and high level detail. Its midrange is the best part of what it can preserve. So the incoming signal is squeezed into the midrange. Keep in mind that there's still usable stuff on the upper and lower parts of the recorded signal too. If you look at log footage on its own, looks terrible, same as Dolby encoded audio sounds. Then you decode it, stretching it back out. All of the low and high level detail that could have been clipped on linear capture (because of the inherent not great sensor technology) is preserved with proper log recording, and the second part of the double-ended system is where you recover it. This is a way to do things now. It will be completely unnecessary once sensor technology gets good enough at low enough costs. Some super high end chips (Sony F65) are really good and therefore log recording isn't necessary. Log is a way to get great looking footage with low/high level details with inferior sensor chip tech. As good as Alexa's sensor is, we always demand Log-C footage from our DPs, because the results speak for themselves. We can get better looking stuff with the log process from most cameras. Even the Canons can sometimes benefit, especially in high or low light shooting.

I hope this is helpful. We are lucky now that there are somewhat better solutions out there (for extreme amounts of money) but the more people invest in digital capture technology, the lower these great sensors will cost. Eventually it will get passed down the product lines, but for now, shoot log and process log and "it gets better".

On Thu, Sep 12, 2013 at 7:52 PM, Edit B <bouke@editb.nl> wrote:

Mark,

Please, re-read. I'm talking about the former.

And even if i miss the point, please educate me, why does ANY remapping / preserving could help me getting more out of a squished image than a properly exposed / balanced image....

The 'otherwise no one' quote is meaningless to me. Zillions of people are doing stuff that don't make sense at all.

I'm not one of them, i am:

Bouke

VideoToolShed
van Oldenbarneveltstraat 33
6512 AS NIJMEGEN, the Netherlands
+31 24 3553311

----- Original Message -----

From: Mark Spano

To: Avid-L2@yahoogroups.com

Sent: Friday, September 13, 2013 1:12 AM

Subject: Re: [Avid-L2] Canon log profile, wtf?

I wonder if you are joking. If not, I'm surprised. Are you talking about "why should anyone shoot in Log with a C300?" or "why should anyone shoot in Log?" - if it's the former, you may have a point. If it's the latter, I can vouch for many camera systems that the resulting log media has benefitted from gaming the light into the sensor's money zone. The lower and upper ranges of the light get squished going in, the sensor picks it up, then you stretch it back out with a LUT or a correction so you can preserve highlight and shadow detail. It works, otherwise no one would use it.

On Thu, Sep 12, 2013 at 7:04 PM, Edit B <bouke@editb.nl> wrote:

Ok, so i could (re) import while compensating for the original ' makes no sense to me' settings.

But that doesn't answer my question...

(I repeat, is there ANY sense in shooting this way?)

Bouke

VideoToolShed
van Oldenbarneveltstraat 33
6512 AS NIJMEGEN, the Netherlands
+31 24 3553311

----- Original Message -----

From: Steve Hullfish

To: Avid-L2@yahoogroups.com

Sent: Friday, September 13, 2013 12:17 AM

Subject: Re: [Avid-L2] Canon log profile, wtf?

I've got one in mine: Canon C-log to REC 709.

On Sep 12, 2013, at 5:07 PM, owen@thenowcorporation.com wrote:

mc 7, i thought i saw a lut for canon in the source settings window, no?

Owen's iphone

On Sep 12, 2013, at 4:46 PM, "Edit B" <bouke@editb.nl> wrote:

Got some material (well, a feature doc worth of) Canon C300 material, shot
in LogC (?)
Now, can anyone explain to me why this way of shooting is actually better in
any other case than ' I haven't got the faintest idea where this is going'
scenario?

I mean, it is still 8 bit, nothing raw like, blacks are lifted (but without
detail, so why even bother?), no usable LUT's (or i am truely misinformed,
please let it be so),
so i find myself just yanking setup / gamma / curves to remap a few pixel
colors that could have been in 'about' the right spot anyways on a normal
video cam approach.

Again, granted, this might be good if the shooter has no clue at all, but
c'mon, how often is that the case where you bring a cam like this? Why not
shoot regular and define a look in the cam?
I still can do a lot on properly white balanced / exposed material, and in
most cases the in-cam defined look will be close, so the end result after CC
will be better than after grading these images.

Or, what am i missing?

(Note to self, do NOT, i repeat, do NOT start a discussion about putting a
light (or perhaps three...) on the scene will improve things above shooting
in a custom 'make it pretty in CC' mode...)

Bouke

VideoToolShed
van Oldenbarneveltstraat 33
6512 AS NIJMEGEN, the Netherlands
+31 24 3553311