M. Black vs. Stretch vs. Gamma

[Yaron Y. Dahan]

Hi,

I am somewhat of a newbie to video, and I am usind the Sony DSR300, and have some confusion

so, from a visual perspective, I noticed that all three of these essentailly affect the contrast, but what is the difference from changing on or the other? both from a technical point and a visual point? any help is appreciated.

[Alessandro Machi]

Hi,

I am somewhat of a newbie to video, and I am usind the Sony DSR300, and have some confusion

so, from a visual perspective, I noticed that all three of these essentailly affect the contrast, but what is the difference from changing on or the other? both from a technical point and a visual point? any help is appreciated.

 

Your question is one of the most important ones there is for a newbie, also for everybody, lol, and everybody doesn't always agree on the answer.

 

From an observational point of view (rather than technical), I am a fan of stretching the black levels. As I understand it, stretching the black levels is a way of decreasing the contrast difference from the lower to mid range gamma signals as compared to the brightest parts of the scene. It's a form of electronic contrast reduction since the darker parts of the picture are made more visible while the brightest parts of the scene stay at the leve they are at. I think this type of stretching works better with an analog video signal than a digital signal if the digital signal is being compressed, but as long as one wasn't going into gain mode on the camera the picture should look fine with even with a digital camera. Stretching is basically an essential element for E.N.G. style of work or in situations in which lighting is not controllable to the degree that is needed, yet it can also make conventional style of lighting less of a task as well.

 

Ironically, many lower cost digital cameras actually INCREASE the overall contrast of the image during acquistion, maybe that would be called "white stretch" since the brightest parts of the scene will read close to 120 IRE and the darkest parts will be at zero black. This is way too much contrast for standard def monitors to handle. Perhaps the HD monitors can handle this range better but when all of the video compression poop is added in even that becomes doubtful.

 

I think it's safe to use black stretch in most situations because the additional video information it helps make more discernible/visible can always be adjusted out during the editing process, but if the information is not there to begin with, it cannot be added in later.

 

Example, an Actor with black hair, no hair light to separate from the background, and a dark background, the black stretch can in many instances show that there is a slight separation in video levels between the background and the black hair, whereas in the "normal mode" the two levels may just merge together as one practically unified black level that has no separation.

[Simon Wyndham]

Stretching the blacks can often be a good thing. However remember to make the distinction between black level and black gamma(stretch).

 

The black stretch will not affect the absolute black level. I like deep blacks (ie when something is black, I want it to look black, not grey.) Often I find that the standard black level setting on cameras is too high for my liking. I nearly always end up adjusting it in post. As long as you don't go below the zero line on a waveform with the master black you will not crush detail. So I wonder if getting a good master black setting, and then using black stretch to bring back some detail might be the best way to go?

[Alessandro Machi]

Stretching the blacks can often be a good thing. However remember to make the distinction between black level and black gamma(stretch).

 

The black stretch will not affect the absolute black level. I like deep blacks (ie when something is black, I want it to look black, not grey.) Often I find that the standard black level setting on cameras is too high for my liking. I nearly always end up adjusting it in post. As long as you don't go below the zero line on a waveform with the master black you will not crush detail. So I wonder if getting a good master black setting, and then using black stretch to bring back some detail might be the best way to go?

 

You're getting into very complex issues that have people on both sides of the issue thinking they are right and those with a differing opinion are wrong. I think you can actually lose detail even if the darkest part of the signal is barely above the zero black (aka IRE of 1,2,3) compared to leaving set-up at 7.5 and making 7.5 equal zero black.

 

Where the confusion may come in relates to the overall contrast in the scene, how much resolution the camera is capable of, how good the lens is, AND, what type of screen one is viewing the signal on. When a picture tube television is used, and the three pluge bars are set up correctly on the screen, anything under 4 IRE is considered black, yet if one "brightens" the screen, that darkest pluge level is now visible!

 

Keep in mind I think there is a difference between graphics images and real life images. Graphics images seem to handle zero black fine, but real life images can easily lose detail in the darks if one uses zero IRE (for NTSC) as a standard and some of the picture information is right around the 1-5 IRE range.

[Yaron Y. Dahan]

As I understand it, stretching the black levels is a way of decreasing the contrast difference from the lower to mid range gamma signals as compared to the brightest parts of the scene. It's a form of electronic contrast reduction since the darker parts of the picture are made more visible while the brightest parts of the scene stay at the leve they are at.

 

Thanks Alessandro.. So stretch, would be the equivalent of axpanding the low t- mid range tones?? Essentially the same thing as changing the bottom half of the gamma curve? Do you know how it does this? or to be more precise at the expense of what?

 

 

Stretching the blacks can often be a good thing. However remember to make the distinction between black level and black gamma(stretch).

 

The black stretch will not affect the absolute black level. I like deep blacks (ie when something is black, I want it to look black, not grey.) Often I find that the standard black level setting on cameras is too high for my liking. I nearly always end up adjusting it in post. As long as you don't go below the zero line on a waveform with the master black you will not crush detail. So I wonder if getting a good master black setting, and then using black stretch to bring back some detail might be the best way to go?

 

thats a very interesting question, and I will have to do some tests and get back to you on that. Actually the project I am working on calls for low key deep images, and I want to emphasise the entire lower range of colors and hues, and then have some bright spots that will go out.

 

Unfortunately, I have no access to a wavceform monitor for the moment, so from a purely scientific point of view, I wont be able to know exactly.

 

 

You're getting into very complex issues that have people on both sides of the issue thinking they are right and those with a differing opinion are wrong. I think you can actually lose detail even if the darkest part of the signal is barely above the zero black (aka IRE of 1,2,3) compared to leaving set-up at 7.5 and making 7.5 equal zero black.

 

Keep in mind I think there is a difference between graphics images and real life images. Graphics images seem to handle zero black fine, but real life images can easily lose detail in the darks if one uses zero IRE (for NTSC) as a standard and some of the picture information is right around the 1-5 IRE range.

 

Hmm. Interesting stuff. So what about the difference beetween M. Black and Gamma? If M Black changes the whole line of black how is it different from Gamma? OR does M Black just lower the gamma line, and the Gamma changes the curve???

 

But I still have one question that is still amystery to me, and I posted it actually in the next post, but since its linked I'll allow myself to ask it again. All of these settings changes - M Black Gamma, etc.... do these changes affect the sensitivity of the chip or are they "in camera post production" changes, and they're just doing "as if"?

[Alessandro Machi]

Thanks Alessandro.. So stretch, would be the equivalent of axpanding the low t- mid range tones?? Essentially the same thing as changing the bottom half of the gamma curve? Do you know how it does this? or to be more precise at the expense of what?

 

I don't know technically how this is done. Others have stated on other threads about this topic that additional video noise can be induced. While I think this is possible, simply crushing that black should make the noise go away if it appears, so it still appears to me to be a risk free venture. Sort of like overexposing negative by 2/3's of a stop for a thicker negative.

 

 

Actually the project I am working on calls for low key deep images, and I want to emphasise the entire lower range of colors and hues, and then have some bright spots that will go out.

 

Sony made a component color corrector about 10-15 years ago that does a really good job of color enhancement in the low, mid and high gamma ranges. Since it's a betacam sp uncompressed signal that is being fed this component color corrector, it may still be better than simply using mini-dv and a basic firewire codec for what you are describing.

 

Unfortunately, I have no access to a waveform monitor for the moment, so from a purely scientific point of view, I wont be able to know exactly.

 

People are actually throwing away older waveforms. They can be had for under a hundred bucks on ebay.

You should look into getting one. The only other option is to use SMPTE color bars and learn how to perfectly adjust the monitor, but that can go haywire if you can't verify the waveform and vectorscope settings for the SMPTE color bar image that you are using.

 

Hmm. Interesting stuff. So what about the difference beetween M. Black and Gamma? If M Black changes the whole line of black how is it different from Gamma? OR does M Black just lower the gamma line, and the Gamma changes the curve???

 

this stuff does get confusing. I've been shooting with 7.5 set-up so that is what I am most familar with. On top of that, when I do get zero set-up master black in my studio (usually these are lower budget or prosumer cameras), I can inevitably pull more detail out of the blacks, without making them go milky, by simply lifting the set-up back to 7.5.

 

But I still have one question that is still a mystery to me, and I posted it actually in the next post, but since its linked I'll allow myself to ask it again. All of these settings changes - M Black Gamma, etc.... do these changes affect the sensitivity of the chip or are they "in camera post production" changes, and they're just doing "as if"?

 

That's a great question. I don't know.

[Simon Wyndham]

You shoot with 7.5 setup with a digital camera?

[Alessandro Machi]

You shoot with 7.5 setup with a digital camera?

 

I shoot with a Betacam SP camera. However the DVX 100 incorporates a 7.5 set-up option.

 

PAL is a different issue when it comes to set-up.

[Yaron Y. Dahan]

That's a great question. I don't know.

 

Okay, So I finally caved in and called Sony. Talked to some guy named Dominick. He told me that all of the changes in Gamma and M Black, etc. were made after the Image was recorded by the chip. I asked him what the point was then, if this stuff if usually done in post production, and he said that there really wasn't any. So why would you want to change stuff in camera at the time of the shoot? So you can see the "feel" of the shoot on the monitor? Dominick told me that it transform the image without any loss, but is this really possilbe? If I boost the contrast to the max, and then in post, I realize it was too much, isn't the information lost, becuase I discarded the mid tones??? I don't really know enough about how video works.

[Simon Wyndham]

he said that there really wasn't any.

 

He's wrong. For example where the adjustment of knee point and knee slope is concerned that will cram in more highlight information. You couldn't leave the camera on a default setting and hope to bring that back in post. Thats why the likes of the F900R have hypergamma, and the Varicam has Film Rec gamma.

 

Then there are issues of noise. The camera has coring and level depend adjustments which can minimise the noise in certain areas of the picture (such as the darker areas). This is more difficult to achieve in post. The fact is that you will get a cleaner picture by having these adjustments made in-camera.

[David Mullen ASC]

Okay, So I finally caved in and called Sony. Talked to some guy named Dominick. He told me that all of the changes in Gamma and M Black, etc. were made after the Image was recorded by the chip. I asked him what the point was then, if this stuff if usually done in post production, and he said that there really wasn't any. So why would you want to change stuff in camera at the time of the shoot? So you can see the "feel" of the shoot on the monitor? Dominick told me that it transform the image without any loss, but is this really possilbe? If I boost the contrast to the max, and then in post, I realize it was too much, isn't the information lost, becuase I discarded the mid tones??? I don't really know enough about how video works.

 

"Recorded by the chip" is not the correct phrase -- the chip can't record, it can only respond to light, converting photons into voltage. Yes, after that, signal processing is applied before the image is recorded to tape. But certain things do affect the exposure information recorded, and if it's not recorded, it's not there in post for you to play with. Knee compression is the most obvious thing, hence why all Sony pro cameras have a "DCC" switch on them ("auto knee") to hold more overexposure detail. Black stretch / black gamma can do the same thing at the other end.

 

The amount of information that comes off of the CCD is not the same amount that gets recorded. Less gets recorded that gets originally captured, hence why image control set-ups do affect the amount of information.

 

However, there is no free lunch with video -- usually attempts to dramatically increase the total range recorded by playing with knee and gamma, etc. can cause other artifacts like noise, sometimes color casts in whites, some other odd-looking effects. So there is a limit and usually one will get a cleaner signal if one lights and exposes, filters, etc. the scene in front of the camera to control the exposure range to more manageable levels.

 

You also have to factor in the recording format -- you will have less flexibility in post to manipulate an 8-bit 3:1:1 signal with a lot of compression (like HDCAM) versus a 10-bit 4:4:4 signal with minimal compression (like HDCAM-SR). So you have to be more careful about what you "leave for post" to do to the image when using the more compressed tape formats. Some things in post are more artifacty than others.

 

In terms of contrast, more contrast tends to mean less exposure information whereas low contrast tends to mean more exposure information. And generally it is easier to throw away information in post than try to recover it if it was never recorded. So if you want a high contrast image, I would probably record a normal-contrast image in-camera and add the contrast in post to the level you want. Within limits, of course -- if you light a scene too flatly, it will be hard to make it look shadowy and high-contrast in post. On the other hand, if you're talking about a DVCAM camera, I'd try and get halfway there in-camera and with lighting rather than save it all for post.

[Alessandro Machi]

Okay, So I finally caved in and called Sony. Talked to some guy named Dominick. He told me that all of the changes in Gamma and M Black, etc. were made after the Image was recorded by the chip. I asked him what the point was then, if this stuff if usually done in post production, and he said that there really wasn't any. So why would you want to change stuff in camera at the time of the shoot? So you can see the "feel" of the shoot on the monitor? Dominick told me that it transform the image without any loss, but is this really possilbe? If I boost the contrast to the max, and then in post, I realize it was too much, isn't the information lost, becuase I discarded the mid tones??? I don't really know enough about how video works.

 

If a scene is lit to spec, than the sony tech's informatin is probably valid. Keep in mind it would be kind of silly to invest in a high end Sony camera and then point it at poorly lit scenes, so the Sony tech may have "filtered" you question with the idea that you are shooting with a decent budget.

 

However, in the world of E.N.G. or low budget, quick acquisition, his statement would appear to be erroneous but perhaps he really wasn't focusing on that aspect when he was discussing details with you.

 

I think what messes up high end tech guys more than anything else is they base their technical knowledge and observations on properly lit, high production value content. These tech guys probably understand the high end production world moreso than the guerilla side of cinematography. When I once explained how off the contrast ratio was on my Hitachi-Z-one C as compared to other similar style cameras, the camera service tech suggested I use more light to fill in the contrast even though other similar type of video cameras could handle contrast situations that the Z-one C could not because it was improperly set up.

[Yaron Y. Dahan]

"Recorded by the chip" is not the correct phrase -- the chip can't record, it can only respond to light, converting photons into voltage. Yes, after that, signal processing is applied before the image is recorded to tape. But certain things do affect the exposure information recorded, and if it's not recorded, it's not there in post for you to play with. Knee compression is the most obvious thing, hence why all Sony pro cameras have a "DCC" switch on them ("auto knee") to hold more overexposure detail. Black stretch / black gamma can do the same thing at the other end.

 

...

On the other hand, if you're talking about a DVCAM camera, I'd try and get halfway there in-camera and with lighting rather than save it all for post.

 

Hmm.. yes, I should have said, "senses" the information

That's quite interesting. I didn't realize that the camera doesn't record all that the sensor "sees"

 

He's wrong. For example where the adjustment of knee point and knee slope is concerned that will cram in more highlight information. You couldn't leave the camera on a default setting and hope to bring that back in post. Thats why the likes of the F900R have hypergamma, and the Varicam has Film Rec gamma.

 

Then there are issues of noise. The camera has coring and level depend adjustments which can minimise the noise in certain areas of the picture (such as the darker areas). This is more difficult to achieve in post. The fact is that you will get a cleaner picture by having these adjustments made in-camera.

 

Also interesting. Instead of continuing on abstract level, why don't I describe my specific upcoming shoot, and maybe get some advice. I am shooting with a pro DV cam (Either DSR 300 or 390). The scence will be a low key scene, with one decent size highlight that I want to go out of range and be bruned out. I want to keep a dark, nice tonal range in the picture keeping the colors nice and deep. If you had to shoot this what changes would you make in-camera? What would you save for post?

 

Thanks...

[Michael Nash]

The scence will be a low key scene, with one decent size highlight that I want to go out of range and be bruned out. I want to keep a dark, nice tonal range in the picture keeping the colors nice and deep. If you had to shoot this what changes would you make in-camera? What would you save for post?

 

Thanks...

 

 

Another way of looking at is this:

 

Master Black (Pedestal) = brightness or value of the darkest black of the image. Raise it and black turns gray; lower it and you crush deep shadow information into black.

 

Black Stretch (Black Gamma) = brightness or value of shadows, approximately 0-30% luminance (that range being adjustable in some cameras).

 

Master Gamma = brightness or value of the midtones. Raise or lower the value and the scene tends to appear overall brighter or darker, while highlight and shadow information stays relatively the same.

 

Keep in mind that as you adjust any/all of these values, you're "reshaping" the gamma curve, in other words the transition of values throughout the luminance range. For example, lowering the black stretch can make not only darker shadows but also a smooth or "compressed" transition of shadow into black. Raising the black stretch on the other hand can make shadow detail brighter and more visible, but transitions abruptly into a "clipped" or "crushed" looking black instead of shadows rolling smoothly into black. The same principle applies to the relationship between any two adjoining values like black stretch-to-gamma, gamma-to-knee and so on. So it's not just about the luminance of the vlaue ranges, it's also about the transition from one value range into another.

 

For the look you describe I would tend to stick with fairly "normal" pedestal, black stretch, and gamma values, and then perhaps massage the contrast a little in post. If I did anything in camera I might lower the master gamma a touch and raise the black stretch a touch (but only a touch), to flatten out the mid-to-shadow contrast a little. But really, most video cameras tend to have a pretty nice curve in that range right out of the box.

 

If you can adjust the master chroma I'd experiement with raising that. It's generally easier to pull out chroma than it is to put it back in. And for burnt-out highlights, you could turn the knee (DCC) off completely.

 

Just keep in mind that whatever you do in camera and goes to tape is what you're stuck with. Attempting to raise shadows that were lowered too much in camera will result in noise and compression artifacts, and any clipped or crushed values cannot be recovered once they're on tape that way. So play it a little safe. ;)

[Alessandro Machi]

Keep in mind that as you adjust any/all of these values, you're "reshaping" the gamma curve, in other words the transition of values throughout the luminance range. For example, lowering the black stretch can make not only darker shadows but also a smooth or "compressed" transition of shadow into black. Raising the black stretch on the other hand can make shadow detail brighter and more visible, but transitions abruptly into a "clipped" or "crushed" looking black instead of shadows rolling smoothly into black. The same principle applies to the relationship between any two adjoining values like black stretch-to-gamma, gamma-to-knee and so on. So it's not just about the luminance of the vlaue ranges, it's also about the transition from one value range into another.

 

One additional variable to consider is the television monitor one is viewing the image on. Ideally a monitor should be perfectly set up using SMPTE color bars with the Pluge bars properly adjusted to help see how contrast values are being displayed, and the monitor should be shielded from ambient lights. If one does not have this luxury, I think it's safe to just slightly lift/stretch the blacks knowing that they can be pulled down later.

 

Just keep in mind that whatever you do in camera and goes to tape is what you're stuck with. Attempting to raise shadows that were lowered too much in camera will result in noise and compression artifacts, and any clipped or crushed values cannot be recovered once they're on tape that way. So play it a little safe. ;)

 

That's why I think it's safer to err on the side of lifting/stretching the blacks just a bit because I know later I can lower them. However, for the first time in a long time, I found myself actually doing the opposite. I was shooting a nighttime shot and I was trying to make an unwanted background white roof gutter drain disappear and I noticed I could in fact crush my signal just a bit without losing critical information in the foreground. I normally set my gamma at the plus one position, but in this instance I was able to drop it to minus 3, something I rarely do. I was able to make this alteration because the actor had recently shaved his head and I didn't have to worry about crushing his dark hair into the background darkness.

[Yaron Y. Dahan]

Another way of looking at is this:

 

Master Black (Pedestal) = brightness or value of the darkest black of the image. Raise it and black turns gray; lower it and you crush deep shadow information into black.

 

Black Stretch (Black Gamma) = brightness or value of shadows, approximately 0-30% luminance (that range being adjustable in some cameras).

 

Master Gamma = brightness or value of the midtones. Raise or lower the value and the scene tends to appear overall brighter or darker, while highlight and shadow information stays relatively the same.

 

Just keep in mind that whatever you do in camera and goes to tape is what you're stuck with. Attempting to raise shadows that were lowered too much in camera will result in noise and compression artifacts, and any clipped or crushed values cannot be recovered once they're on tape that way. So play it a little safe. ;)

 

Thanks michael. thats exatly the information I was looking for. I actually did some camera tests last week, and sort of figured it out at the end.

 

I think for the situation I described above, I will play around with the M black a bit, because IVe so far been unsatisfied with the way blacks look on the regular "Straight line" Gamma, and the M Black seems (to my eye at least) to somewhat duplicate the filmic "foot" of the gamma.

 

Thanks again for the help, and for the tip regarding Chroma.

[Nadav Hekselman]

Hi yaron,

 

I did some test the last two days, also inspired by this thread.

 

I took a DSR-500 out on my porch and shot myself with diffrent exposure and settings of the m.black, gamma, and stretch...

 

I gotta tell you, in my opinion, you might want to play a bit with those settings but only to get the most information from the picture and to expand the latitude as much as you can, and leave the real work for post.

 

Video burns out so fast and lose information in highlights .. its realy frustrating.... i rated the camera at 400 asa and the highlights got washed out only 1 stop over exposed... grrr

 

knowing that, i recommend you to expose the highlights correctly and according to that judge the mid to low range of the gamma in the picture you get. then crush the blacks or reduce the gamma according to that.. but just a bit, so you can work on it later in post. if its a very contrasty day and when you expose the highlights you get no exposure in the black areas at all... you can add some stretch to salvage information and avoiding the ugly fony video black.

 

 

hope it helps in anyway...

 

Nadav Hekselman

[Alessandro Machi]

Hi yaron,

 

I did some test the last two days, also inspired by this thread.

 

I took a DSR-500 out on my porch and shot myself with diffrent exposure and settings of the m.black, gamma, and stretch...

 

I gotta tell you, in my opinion, you might want to play a bit with those settings but only to get the most information from the picture and to expand the latitude as much as you can, and leave the real work for post.

 

Video burns out so fast and lose information in highlights .. its realy frustrating.... i rated the camera at 400 asa and the highlights got washed out only 1 stop over exposed... grrr

 

knowing that, i recommend you to expose the highlights correctly and according to that judge the mid to low range of the gamma in the picture you get. then crush the blacks or reduce the gamma according to that.. but just a bit, so you can work on it later in post. if its a very contrasty day and when you expose the highlights you get no exposure in the black areas at all... you can add some stretch to salvage information and avoiding the ugly fony video black.

hope it helps in anyway...

 

Nadav Hekselman

 

 

It just dawned on me we could use an additional word description here. Normally crushing the blacks means the lower end of the signal is compressed for a rich black, but I think the way you mean it in your description is "crunch" the blacks in the mid range so that they all are adjustable later on.

[Nadav Hekselman]

It just dawned on me we could use an additional word description here. Normally crushing the blacks means the lower end of the signal is compressed for a rich black, but I think the way you mean it in your description is "crunch" the blacks in the mid range so that they all are adjustable later on.

 

Well, actually, when I said gamma i was refering to the mid parts and "cursh the blacks" meant the lower parts.