Let's talk about Dynamic Range

Ivan Mosny · November 7th, 2007, 08:17 AM

Quote:

Originally Posted by Kris Bird

If you alter Setup or Master Ped then the whole curve shifts up and down, so the gamma correction does happen before that in the image chain... I'm not saying this is a major factor, just that black and white aren't absolutes.

Black an white levels are defined in analog videosignal with absolute Voltage. Your tv respect this as mina and max. levels in Picture reproduction. In digal - are they defined with minimal or maximal level in its bit resolution. So can be a black level 0,0,0 an white level 255,255,255. It is not necessary to have points at black or white level in Picture. Any gamma curve always beginn in Black level on both sides input and output and ends in max. white level. Even curves with master pedestal - the are only constant for a smal range of input.
CCD sensors have own maximal dynamic range - u can`t change this with gamma corrector. And after CCDs is ist always a input/ouptut matrix in picture signal chain.

Kris Bird · November 7th, 2007, 09:01 AM

Analogue min and max IRE 'are' at points on a scale- say from 0IRE to 100IRE for luma. These are voltages as you said. But the cameras can record super blacks and/or super whites- i.e. values above and below the analogue broadcast legal values. If your TV is displaying 0 to 100 IRE when the cam records around -4 to 114 IRE (which seem to be approx the A1s min and max for composite), then the TV is displaying 0.034 to 0.881... so if you gamma correct in camera (i.e. 0-1), there will be values which scale in and out of your TV's visible range. I presume that this what you're seeing when adjusting gamma in camera seems to change where the 100% zebras are on a recorded luminance ramp?

Peter Ralph · November 7th, 2007, 09:30 AM

in the days when everyone was shooting for broadcast there was a strong imperative to have whites at 100 and blacks at 0 - it was the only way to get a full volt of signal and maximize the sn ratio.

Kris - what you are asking for is a "maximize dynamic range" setting. The default setting is about the best you will find. Dynamic range needs to be evaluated for each scene - what you will normally want is to maximize the tonal values that fall in the midrange - in many circumstances it's fine to blow out the whites and crush the blacks - and that is where the cine settings help.

Kris Bird · November 7th, 2007, 12:51 PM

Hi Peter- I'm definitely seeing more DR with knee low and blacks stretched- cine (i.e. more linear) gamma expands the steps at the top of the step chart, so while it might not be increasing DR, I'm getting more usable/recoverable detail for grading.

You said- "in many circumstances it's fine to blow out the whites and crush the blacks - and that is where the cine settings help"- the cine gammas don't do this in my tests, they expand the highlight detail and compress the lows (without necessary affecting DR at all of course). They don't clip the blacks.

"Dynamic range needs to be evaluated for each scene - what you will normally want is to maximize the tonal values that fall in the midrange"- Agreed, but in my mind this has everything to do with the gamma, recording with a result as close as possible to your end result in order to use the bits where you need em.

Kris Bird · November 7th, 2007, 09:18 PM

Okay, I'm 'absolutely' seeing more DR with cine2, then video gamma. Sometimes you just have to trust results (and charts..). Without a doubt, the cine gamma is unclipping values in an over-exposed test scene, without clipping lower values. Obviously the lows get compressed down a bit (i.e. get less bits).

For anyone totally confused, check these links-

http://www.usa.canon.com/app/html/HD...justment.shtml

http://www.motionfx.gr/Files/HD%20Va...Film%20Rec.pdf

(The second is regarding the varicam, but it's relevant.)

To be clear- I'm not talking about getting something for free. It's clear that the camera as set up by default is optimised for legal broadcast delivery on a standard monitor/screen, for a nice bright image with minimum contouring. The CCDs *do* capture more DR than you see with typical settings, but capturing this within the 8bit file format necessitates compromises- extra bits used to expand highlights (i.e. knee low, or cine gamma) is basically taking bits from further down the scale, minutely increasing risk of contouring, etc., etc... Cine2 gives you more DR from the chips, but makes the image innapropriate for display on a monitor without tweaking-- as the dynamic range is too wide for a typical monitor/viewing environment it'll look too dark until corrected in post.

I've started experimenting with cine2 on shoots and, where it's appropriate for the scene/style, I'm much prefering it in the grade.

Brandon Freeman · November 8th, 2007, 11:26 AM

As I've been further investigating this issue, I've decided to purchase a Tiffen Ultra Contrast 3 filter from Bhphotovideo.com. According to Tiffen's website, this filter will reduce contrast and bring up dark areas without flare, as shown in the image attached.

Here is the link to the filter:

http://www.tiffen.com/displayproduct...&itemnum=72UC3

Kris Bird · November 12th, 2007, 10:44 AM

Hi Brandon! Yeah they're definitely interesting, haven't tried them yet. Charles' comment about low-levels shifting up and down as you pan around (due to the ambient lift of the ultra-con being dependent on how much light there is available in total) is a tad worrying, since I shoot a lot of hand-held cine type stuff, but I'm definitely planning on checking them out. Obviously it doesn't 'capture' the extended dynamic range, in the way that cine gamma does, but 90% of the time (when shooting video) you're mostly just worried about highlight clipping in outdoor scenes, etc., so it's definitely interesting. Our feature shoot (brevis, nikon glass, largely hand-held via shoulder mounted rod setup) will feature about 20% outdoor daylight material, so let me know how you get on.

Peter Ralph · November 14th, 2007, 09:58 AM

Kris my sentence read:

"...what you will normally want is to maximize the tonal values that fall in the midrange - .... - and that is where the cine settings help."

The "- in many circumstances it's fine to blow out the whites and crush the blacks -" was parenthesized with the intent of underlining the fact that it is often OK to blow out the highlights or crush the blacks precisely because this may yield more usable tones in the midrange where you need them.

November 8th, 2007, 11:26 AM	#21
Brandon Freeman Major Player Join Date: Feb 2007 Location: Boise, ID Posts: 226	As I've been further investigating this issue, I've decided to purchase a Tiffen Ultra Contrast 3 filter from Bhphotovideo.com. According to Tiffen's website, this filter will reduce contrast and bring up dark areas without flare, as shown in the image attached. Here is the link to the filter: http://www.tiffen.com/displayproduct...&itemnum=72UC3 Attached Thumbnails __________________ Canon EOS 7D, Final Cut Studio 2

November 7th, 2007, 09:01 AM	#17
Kris Bird Regular Crew Join Date: Sep 2005 Location: UK Posts: 173	Analogue min and max IRE 'are' at points on a scale- say from 0IRE to 100IRE for luma. These are voltages as you said. But the cameras can record super blacks and/or super whites- i.e. values above and below the analogue broadcast legal values. If your TV is displaying 0 to 100 IRE when the cam records around -4 to 114 IRE (which seem to be approx the A1s min and max for composite), then the TV is displaying 0.034 to 0.881... so if you gamma correct in camera (i.e. 0-1), there will be values which scale in and out of your TV's visible range. I presume that this what you're seeing when adjusting gamma in camera seems to change where the 100% zebras are on a recorded luminance ramp?

November 7th, 2007, 09:30 AM	#18
Peter Ralph Major Player Join Date: Feb 2007 Location: Denver/Vail Colorado Posts: 254	in the days when everyone was shooting for broadcast there was a strong imperative to have whites at 100 and blacks at 0 - it was the only way to get a full volt of signal and maximize the sn ratio. Kris - what you are asking for is a "maximize dynamic range" setting. The default setting is about the best you will find. Dynamic range needs to be evaluated for each scene - what you will normally want is to maximize the tonal values that fall in the midrange - in many circumstances it's fine to blow out the whites and crush the blacks - and that is where the cine settings help.

November 7th, 2007, 12:51 PM	#19
Kris Bird Regular Crew Join Date: Sep 2005 Location: UK Posts: 173	Hi Peter- I'm definitely seeing more DR with knee low and blacks stretched- cine (i.e. more linear) gamma expands the steps at the top of the step chart, so while it might not be increasing DR, I'm getting more usable/recoverable detail for grading. You said- "in many circumstances it's fine to blow out the whites and crush the blacks - and that is where the cine settings help"- the cine gammas don't do this in my tests, they expand the highlight detail and compress the lows (without necessary affecting DR at all of course). They don't clip the blacks. "Dynamic range needs to be evaluated for each scene - what you will normally want is to maximize the tonal values that fall in the midrange"- Agreed, but in my mind this has everything to do with the gamma, recording with a result as close as possible to your end result in order to use the bits where you need em.

November 7th, 2007, 09:18 PM	#20
Kris Bird Regular Crew Join Date: Sep 2005 Location: UK Posts: 173	Okay, I'm 'absolutely' seeing more DR with cine2, then video gamma. Sometimes you just have to trust results (and charts..). Without a doubt, the cine gamma is unclipping values in an over-exposed test scene, without clipping lower values. Obviously the lows get compressed down a bit (i.e. get less bits). For anyone totally confused, check these links- http://www.usa.canon.com/app/html/HD...justment.shtml http://www.motionfx.gr/Files/HD%20Va...Film%20Rec.pdf (The second is regarding the varicam, but it's relevant.) To be clear- I'm not talking about getting something for free. It's clear that the camera as set up by default is optimised for legal broadcast delivery on a standard monitor/screen, for a nice bright image with minimum contouring. The CCDs do capture more DR than you see with typical settings, but capturing this within the 8bit file format necessitates compromises- extra bits used to expand highlights (i.e. knee low, or cine gamma) is basically taking bits from further down the scale, minutely increasing risk of contouring, etc., etc... Cine2 gives you more DR from the chips, but makes the image innapropriate for display on a monitor without tweaking-- as the dynamic range is too wide for a typical monitor/viewing environment it'll look too dark until corrected in post. I've started experimenting with cine2 on shoots and, where it's appropriate for the scene/style, I'm much prefering it in the grade.

November 12th, 2007, 10:44 AM	#22
Kris Bird Regular Crew Join Date: Sep 2005 Location: UK Posts: 173	Hi Brandon! Yeah they're definitely interesting, haven't tried them yet. Charles' comment about low-levels shifting up and down as you pan around (due to the ambient lift of the ultra-con being dependent on how much light there is available in total) is a tad worrying, since I shoot a lot of hand-held cine type stuff, but I'm definitely planning on checking them out. Obviously it doesn't 'capture' the extended dynamic range, in the way that cine gamma does, but 90% of the time (when shooting video) you're mostly just worried about highlight clipping in outdoor scenes, etc., so it's definitely interesting. Our feature shoot (brevis, nikon glass, largely hand-held via shoulder mounted rod setup) will feature about 20% outdoor daylight material, so let me know how you get on.

November 14th, 2007, 09:58 AM	#23
Peter Ralph Major Player Join Date: Feb 2007 Location: Denver/Vail Colorado Posts: 254	Kris my sentence read: "...what you will normally want is to maximize the tonal values that fall in the midrange - .... - and that is where the cine settings help." The "- in many circumstances it's fine to blow out the whites and crush the blacks -" was parenthesized with the intent of underlining the fact that it is often OK to blow out the highlights or crush the blacks precisely because this may yield more usable tones in the midrange where you need them.