gamma curves, how to read this chart? at DVinfo.net

Ron Wilber · August 25th, 2009, 01:53 AM

I grabbed this image from the fx1000 brochure and I'm having a hard time reading it. Which gamma gives the best dynamic range and how do you tell? what does the output and input percentage numbers mean?

Ben Hall · August 25th, 2009, 10:35 AM

They are just different contrast curves, aren't they?

There is no "better", just which you prefer for a given situation.

Perrone Ford · August 25th, 2009, 11:27 AM

The gamma curve defines the relationship of how light that is input to the camera is recorded.

Ignoring what is in the chart for a moment, a linear gamma curve would say that if light coming into the camera is zero, then a zero gets recorded for light level on the card or tape. If light level is equal to 50, then a 50 gets written to the card, etc. While this is fine in many cases, it means you run out of range quickly in a camera that only has 256 possible levels for light like our cameras do. Generally, scenes that have deep shadows you want those little fine steps of a linear gamma. But as the scene gets brighter and brighter, the small graduations just don't become so important.

So it's common to see that by the time the brightness is 50%, we are going to a 2:1 relationship. So you would need to raise the light level 2% to get a 1% change in what is recorded to the camera. This increases dynamic range, at the expense of being able to see fine changes in brightness. By the time the light is super-bright, like shooting into the sun, or white clouds on a sunny day, you don't really care about the tiny differences in light, you just want it to be "white" or near white. So way up there, you may be at a 5:1 ratio or higher.

Generally these brightnesses are measure in IRE. From 0 to 100 for video purposes. However, you can see values below 0 (superblack) or above 100 (superwhite). These are not legal values for TV broadcast, but certainly are usable for computer based use or film. So most cameras will record them up to or down to a certain level. How you choose to use them is up to you. Therefore on the chart you can see that the levels go beyond 100%, and how the gamma curve handles those super-high values.

From the chart, you can see that the "normal" gamma will offer the most shadow detail because it is giving nearly a 40% output for an input of 20%. There is going to be major detail in those shadows. But by 80%, it's a lot more linear. And then it overshoots WAY into the super-whites. The other two gammas simply won't let you record anything that is not legal for TV in the whites, but they step on the shadows a lot more, so detail is lost down there. You can see that type1 offers more shadow detail than Type 2. The pictures illustrate that clearly in the lower left and in the lower portion of the man's suit jacket. All detail is lost with Type2. That gamma might be useful if you are trying to record a VERY bright scene and you really wanted to make sure that nothing was over exposed. But in a scene that goes from very dark, to very bright, you're going to squash the image on both ends.

Ron Wilber · August 25th, 2009, 01:13 PM

wow, thanks for the explanation Perrone. I may have a few more questions after i digest this..

Ron Wilber · August 25th, 2009, 10:08 PM

okay, had time to digest...

So in theory, cinema gamma 2 will produce a film-like look by creating a clean seperation between light and dark elements in the scene.

In otherwards when lighting a scene:

1. shadow and black areas are crushed
2. backighting/rim lighting will turn out crisp and seperated
3. Key or fill light areas will have the widest dynamic range

Also, under further investigation, I found that the cinema gamma modes are less grainy than the regular mode. Is this because they aren't pumping up the output.. thus adding gain?

August 25th, 2009, 01:53 AM	#1
Ron Wilber Regular Crew Join Date: Nov 2007 Location: San Diego, CA Posts: 98	gamma curves, how to read this chart? I grabbed this image from the fx1000 brochure and I'm having a hard time reading it. Which gamma gives the best dynamic range and how do you tell? what does the output and input percentage numbers mean?

August 25th, 2009, 11:27 AM	#3
Perrone Ford Inner Circle Join Date: Nov 2006 Location: Tallahassee, FL Posts: 4,100	The gamma curve defines the relationship of how light that is input to the camera is recorded. Ignoring what is in the chart for a moment, a linear gamma curve would say that if light coming into the camera is zero, then a zero gets recorded for light level on the card or tape. If light level is equal to 50, then a 50 gets written to the card, etc. While this is fine in many cases, it means you run out of range quickly in a camera that only has 256 possible levels for light like our cameras do. Generally, scenes that have deep shadows you want those little fine steps of a linear gamma. But as the scene gets brighter and brighter, the small graduations just don't become so important. So it's common to see that by the time the brightness is 50%, we are going to a 2:1 relationship. So you would need to raise the light level 2% to get a 1% change in what is recorded to the camera. This increases dynamic range, at the expense of being able to see fine changes in brightness. By the time the light is super-bright, like shooting into the sun, or white clouds on a sunny day, you don't really care about the tiny differences in light, you just want it to be "white" or near white. So way up there, you may be at a 5:1 ratio or higher. Generally these brightnesses are measure in IRE. From 0 to 100 for video purposes. However, you can see values below 0 (superblack) or above 100 (superwhite). These are not legal values for TV broadcast, but certainly are usable for computer based use or film. So most cameras will record them up to or down to a certain level. How you choose to use them is up to you. Therefore on the chart you can see that the levels go beyond 100%, and how the gamma curve handles those super-high values. From the chart, you can see that the "normal" gamma will offer the most shadow detail because it is giving nearly a 40% output for an input of 20%. There is going to be major detail in those shadows. But by 80%, it's a lot more linear. And then it overshoots WAY into the super-whites. The other two gammas simply won't let you record anything that is not legal for TV in the whites, but they step on the shadows a lot more, so detail is lost down there. You can see that type1 offers more shadow detail than Type 2. The pictures illustrate that clearly in the lower left and in the lower portion of the man's suit jacket. All detail is lost with Type2. That gamma might be useful if you are trying to record a VERY bright scene and you really wanted to make sure that nothing was over exposed. But in a scene that goes from very dark, to very bright, you're going to squash the image on both ends. __________________ DVX100, PMW-EX1, Canon 550D, FigRig, Dell Octocore, Avid MC4/5, MB Looks, RedCineX, Matrox MX02 mini, GTech RAID, Edirol R-4, Senn. G2 Evo, Countryman, Moles and Lowels.

August 25th, 2009, 10:35 AM	#2
Ben Hall Regular Crew Join Date: Apr 2009 Location: United Kingdom Posts: 48	They are just different contrast curves, aren't they? There is no "better", just which you prefer for a given situation.

August 25th, 2009, 01:13 PM	#4
Ron Wilber Regular Crew Join Date: Nov 2007 Location: San Diego, CA Posts: 98	wow, thanks for the explanation Perrone. I may have a few more questions after i digest this..

August 25th, 2009, 10:08 PM	#5
Ron Wilber Regular Crew Join Date: Nov 2007 Location: San Diego, CA Posts: 98	okay, had time to digest... So in theory, cinema gamma 2 will produce a film-like look by creating a clean seperation between light and dark elements in the scene. In otherwards when lighting a scene: 1. shadow and black areas are crushed 2. backighting/rim lighting will turn out crisp and seperated 3. Key or fill light areas will have the widest dynamic range Also, under further investigation, I found that the cinema gamma modes are less grainy than the regular mode. Is this because they aren't pumping up the output.. thus adding gain?