Gain or pain in post?

Daniel Browning · May 11th, 2008, 01:54 AM

Quote:

Originally Posted by Vito DeFilippo

before compression

I think this is an important difference. Brightening in post will accentuate the compression artifacts in addition to the noise.

Seun Osewa · May 11th, 2008, 02:05 PM

Yeah. I'm also with the do-it-in-camera guys.

Giovanni Speranza · May 11th, 2008, 06:29 PM

The in camera gain ADDS noise. The post gain rises the video level. So you will see a more luminous noise, not "more" noise.

Patrick Moreau · May 12th, 2008, 11:45 AM

We do a ton of color work on all of our images and I am extremely picky about exposure and controlling the highlights and shadow detail so I can share with you what I know from that perspective.

The quick answer to your question is to use both techniques - gain and post. As you use more gain, you can do less post though. If you can properly exposure your image with gain and no need to brighten, I use gain first as it generally looks quite good as long as it isn't pushed. If you need to push the footage, keep the gain fairly low, such as 3 or 6db. I've had some odd circumstances that led to vastly underexposed footage at 0 gain and I have pushed it a good 3-4 stops and it looks usable, probably better than shooting at 12gain and brightening in post (which is a terrible combination).

Overall, it is almost like an alternating sequence. If you need a little bit of light, add gain. If you need a little more than that, use no gain and fix in post. If you need more than that, add twice the gain. If you need even more than that, use minimal gain and post. The trick is learning what each of those points correspond to with your camera and lighting situations.

Patrick

Martin Pauly · May 12th, 2008, 12:38 PM

In camera is - at least theoretically - better, and the experience that many have posted here seems to confirm this in practical use.

Here is why: the ADC (analog-to-digital converter) inside the camera takes a voltage (representing brightness) from the CCD for each pixel, and converts it into a number. The number is then stored on tape (for simplicity's sake, we'll disregard encoding/compression here) and read by a computer when the tape is captured.

To keep it simple, let's assume that these numbers range from 1 to 100, 1 meaning the darkest and 100 the brightest possible value. If you are shooting in a really, really dark place, only a small portion of this range is going to be used - for example, from 1 to 10. Note that these are discrete numbers, i.e. there is no 3.5 between 3 and 4 (due to the digital nature of the DAC's output). So in this example, what's recorded is an image with only ten discrete brightness values, one of which is assigned to each pixel. Therefore, a gradual transition between intensities is no longer possible.

The example may be exaggerated, but you can probably see where this is going: apply an analog gain between the CCD and the DAC in the camera, and you can record the brightness data to tape using finer increments. As long as the gain is "clean", i.e. doesn't introduce significant noise in addition to what's already there at the CCD's output, this will result in more picture information being on the tape, and it will increase your options in post processing. Worst case, you'll have to lower the brightness in post if it looks to bright, so you'll undo (some of) the gain that occured in the camera. But if we tried to do it the other way around - no gain in camera, increase brightness in post - the increments between the few discrete brightness levels would be amplified and become painfully visible.

- Martin

May 12th, 2008, 11:45 AM	#19
Patrick Moreau Still Motion Join Date: Oct 2004 Location: Toronto, Canada Posts: 1,186	We do a ton of color work on all of our images and I am extremely picky about exposure and controlling the highlights and shadow detail so I can share with you what I know from that perspective. The quick answer to your question is to use both techniques - gain and post. As you use more gain, you can do less post though. If you can properly exposure your image with gain and no need to brighten, I use gain first as it generally looks quite good as long as it isn't pushed. If you need to push the footage, keep the gain fairly low, such as 3 or 6db. I've had some odd circumstances that led to vastly underexposed footage at 0 gain and I have pushed it a good 3-4 stops and it looks usable, probably better than shooting at 12gain and brightening in post (which is a terrible combination). Overall, it is almost like an alternating sequence. If you need a little bit of light, add gain. If you need a little more than that, use no gain and fix in post. If you need more than that, add twice the gain. If you need even more than that, use minimal gain and post. The trick is learning what each of those points correspond to with your camera and lighting situations. Patrick __________________ site // www.stillmotion.ca blog // www.stillmotionblog.com edu // www.stillmotionexperience.com

May 12th, 2008, 12:38 PM	#20
Martin Pauly Major Player Join Date: Apr 2006 Location: Cedar Rapids, IA Posts: 563	In camera is - at least theoretically - better, and the experience that many have posted here seems to confirm this in practical use. Here is why: the ADC (analog-to-digital converter) inside the camera takes a voltage (representing brightness) from the CCD for each pixel, and converts it into a number. The number is then stored on tape (for simplicity's sake, we'll disregard encoding/compression here) and read by a computer when the tape is captured. To keep it simple, let's assume that these numbers range from 1 to 100, 1 meaning the darkest and 100 the brightest possible value. If you are shooting in a really, really dark place, only a small portion of this range is going to be used - for example, from 1 to 10. Note that these are discrete numbers, i.e. there is no 3.5 between 3 and 4 (due to the digital nature of the DAC's output). So in this example, what's recorded is an image with only ten discrete brightness values, one of which is assigned to each pixel. Therefore, a gradual transition between intensities is no longer possible. The example may be exaggerated, but you can probably see where this is going: apply an analog gain between the CCD and the DAC in the camera, and you can record the brightness data to tape using finer increments. As long as the gain is "clean", i.e. doesn't introduce significant noise in addition to what's already there at the CCD's output, this will result in more picture information being on the tape, and it will increase your options in post processing. Worst case, you'll have to lower the brightness in post if it looks to bright, so you'll undo (some of) the gain that occured in the camera. But if we tried to do it the other way around - no gain in camera, increase brightness in post - the increments between the few discrete brightness levels would be amplified and become painfully visible. - Martin __________________ Martin Pauly

May 11th, 2008, 02:05 PM	#17
Seun Osewa Major Player Join Date: Oct 2006 Location: West Africa Posts: 255	Yeah. I'm also with the do-it-in-camera guys.

May 11th, 2008, 06:29 PM	#18
Giovanni Speranza Regular Crew Join Date: Aug 2006 Location: Lugano, Switzerland Posts: 149	The in camera gain ADDS noise. The post gain rises the video level. So you will see a more luminous noise, not "more" noise.