EX3 Picture profiles for wildlife

Garrett Low · January 12th, 2011, 10:56 AM

Alister, I think it is important to note again, as you've pointed out, that cinegamma curves are meant to be graded in post. If you don't, you'll be scratching your head wondering how this wonderful camera can produce such flat, dull pictures when you view your footage. I know that before I understood what was going one, I was scratching my head and getting frustrated.

-Garrett

Jyrki Hokkanen · January 13th, 2011, 07:04 AM

Now I see why there has been controversy about the -3db gain setting. After another go at it, I joined the club and found a dependence between highlight IRE levels and gain, when using cinematic gammas.

When shooting with gamma cine1 embedded in an otherwise bare bones picture profile (everything turned off), gain +3db made luma to peak at 107 IRE, gain 0db at 105 IRE, and -3db at 100 IRE. Peak levels were read from Premiere's YC Waveform scope by eye, with an error of about 1 IRE.

However, any gain dependent highlight level drop was still not detectable to me, when using standard gammas (knee off) or when picture profile was turned off.

I suppose that, when using standard gammas with no (or steep) knees, or cine gammas with sufficient level boost, the low gain effects on highlights would take place mostly above 109 IRE and escape attention.

These effects are of course qualitatively seen directly in the EX viewfinder. Just open up the aperture and see if the histogram squeezes tightly against the bounding box on the right hand side. If a gap remains, the luma peaks at lesser values than 109 IRE. And, the zebras tell when values above 100 IRE (and 102 IRE) go missing (zebra 1 at 107).

Jyrki Hokkanen · January 17th, 2011, 04:52 AM

To finish up my previous post, here's how to use the low noise -3db gain setting without clipping the highlights.

Dialing in negative Gamma Compensation Level values should make the picture darker. All right, works as expected for darker shades but at the same time the light tones grow brighter. The dynamic range is boosted up.

And this is not the only thing that happens. The highlight cut off level gets lifted up, if it was below the 109 IRE mark.

So, when using cinegammas and -3db gain in daylight, lowering the gamma level by a barely visible amount of -5 is enough to push the cut off to the 109 IRE limit.

Just to make life more interesting, the highlight cut off limit is also related to white balance value. In tungsten light (say 2800) the cut off takes place lower and correspondingly the gain level has to be lowered more (say -25) to get rid of the highlight cut effect.

Be aware that negative Gain Compensation Level values boosts cine2 whites up to illegal superwhites!

This gamma level trick applies also to standard gammas, but lower-than-109 IRE cut off would be best handled by adjusting the knee settings.

All of this holds good for simple picture profiles where most parameters are set "off" or zero. The effect of other parameters not mentioned above is probably small but should be checked in the field from the viewfinder, using histogram, brightness level indication, and zebras.

January 12th, 2011, 10:56 AM	#16
Garrett Low Trustee Join Date: Jul 2007 Location: Novato, CA Posts: 1,774	Alister, I think it is important to note again, as you've pointed out, that cinegamma curves are meant to be graded in post. If you don't, you'll be scratching your head wondering how this wonderful camera can produce such flat, dull pictures when you view your footage. I know that before I understood what was going one, I was scratching my head and getting frustrated. -Garrett __________________ Garrett Low www.GLowMediaProductions.com

January 13th, 2011, 07:04 AM	#17
Jyrki Hokkanen Regular Crew Join Date: Oct 2003 Location: Helsinki, Finland Posts: 35	Yes, no, maybe, sometimes Now I see why there has been controversy about the -3db gain setting. After another go at it, I joined the club and found a dependence between highlight IRE levels and gain, when using cinematic gammas. When shooting with gamma cine1 embedded in an otherwise bare bones picture profile (everything turned off), gain +3db made luma to peak at 107 IRE, gain 0db at 105 IRE, and -3db at 100 IRE. Peak levels were read from Premiere's YC Waveform scope by eye, with an error of about 1 IRE. However, any gain dependent highlight level drop was still not detectable to me, when using standard gammas (knee off) or when picture profile was turned off. I suppose that, when using standard gammas with no (or steep) knees, or cine gammas with sufficient level boost, the low gain effects on highlights would take place mostly above 109 IRE and escape attention. These effects are of course qualitatively seen directly in the EX viewfinder. Just open up the aperture and see if the histogram squeezes tightly against the bounding box on the right hand side. If a gap remains, the luma peaks at lesser values than 109 IRE. And, the zebras tell when values above 100 IRE (and 102 IRE) go missing (zebra 1 at 107).

January 17th, 2011, 04:52 AM	#18
Jyrki Hokkanen Regular Crew Join Date: Oct 2003 Location: Helsinki, Finland Posts: 35	Tails of the gamma level To finish up my previous post, here's how to use the low noise -3db gain setting without clipping the highlights. Dialing in negative Gamma Compensation Level values should make the picture darker. All right, works as expected for darker shades but at the same time the light tones grow brighter. The dynamic range is boosted up. And this is not the only thing that happens. The highlight cut off level gets lifted up, if it was below the 109 IRE mark. So, when using cinegammas and -3db gain in daylight, lowering the gamma level by a barely visible amount of -5 is enough to push the cut off to the 109 IRE limit. Just to make life more interesting, the highlight cut off limit is also related to white balance value. In tungsten light (say 2800) the cut off takes place lower and correspondingly the gain level has to be lowered more (say -25) to get rid of the highlight cut effect. Be aware that negative Gain Compensation Level values boosts cine2 whites up to illegal superwhites! This gamma level trick applies also to standard gammas, but lower-than-109 IRE cut off would be best handled by adjusting the knee settings. All of this holds good for simple picture profiles where most parameters are set "off" or zero. The effect of other parameters not mentioned above is probably small but should be checked in the field from the viewfinder, using histogram, brightness level indication, and zebras.