[simons-photography]

I've notice like with my old fuji that the histogram is divided into 4 parts, does anybody know how much each quarter represents in stops ?

after a few test shots at 2 stops apart I guess about 2 stops per quarter, its handy to know so as to make it easier to make accurate exposure compensations when judgig the information in the prteview histogram

[falconeye]

Hi simons,

don't you risk to suffer from forum addiction, judging by the frquency by which you start threads?
just joking, never mind.

Histogram. Easy question. Say, available light goes from 0...255 (full histogram).
If you expose a 1 stop less, half available light, i.e. now goes from 0...127 (2 quarters = 1 stop).
If you expose another 1 stop less, quarter available light, i.e. now goes from 0...63 (3 quarters = 2 stops).
(3.5 quarters = 3 stops) ...and so on...

The scale is linear, not logarithmic, therefore not in f-stops.

[simons-photography]

yea I'm a forum addict, especially when I spend 500 £ on a great camera and run into a forum full of people that know more about my camera than the manual (and me for the moment)

the stop scale is linear it sort of converts the logaritmic scale of our eyes to linear. what you describe is saying that the right half of the scale is 127, the next quarter is 64, the next eigth 32 ect.

I presume that the histograms scale represents the whole dinamic range of the camera (CCD) so really I'm asking whats the dinamic range of the camera (K10D) ? 8 stops (EV) ? that would make each 1/4 devision 2 stops (EV) = my guess perhaps 10 stops making each 1/4 division 2.5 stops I'm just interested in knowing roughly what it is as I rely on the histogram in part in making exposure ajustments, of course if i over expose I will get the highlight blown warning but I'm trying to advoid that.

thanks for all your help and support people.

[creampuff]

There is a treasure trove of information in many previous threads.
My suggestion to those new to the forum is to slowly go through the past threads as many of the topics and information contained therein have been discussed to death.

However my observation is that many of the trivial queries could have been answered by first reading throughly the camera manual first or reading through previous threads. Instead it is perhaps easier to be spoon fed by starting a new thread instead of doing a simple search.

[Lowell Goudge]

You need to check very closely what the histogram bars in the camera mean.

the best way is to do a test, in manual, with a lens set to about F9 and exposure set to normal, with no compensation (camera set to 1/2 stop increments)

Without changing the shutter at all, take a series of shots at each F stop, from wide open to stopped down all the way. (use a neutral object like a block wall)

Look at the histograms and also look at the average grey scale value of the histogram in your photo editor.

As others have noted a histogram is 0-155, BUT the grey scale value is not linear.

between about 25 and 230 is between 4-6 stops depending on contrast setting of your JPEG (and 5 stops in RAW)

There are a couple of stops compressed very non linearly into the 0-25 and 230-255 range.

The change of 1 stop does not result in a 50% change in thehistogram value, more likely less than 1 of the 4 divisions of the camera.

Most people consider a histogram good if the histogram is non zero over the entire range, but just approaching 0 at both ends. (this is a matter of opinion and some shots may be deliberately not following this exposure)

[falconeye]

that would make each 1/4 devision 2 stops (EV) = my guess perhaps 10 stops making each 1/4 division 2.5 stops

It is really dumb simple. Maybe read the paragraph about histogram in your Photoshop manual or what other image processing manual. Those are all plain old 0...255 histograms.

If all your histogram data are found in the left half, it is time to add 1EV (or more), if it is clipped to the right, remove 1EV (more or less). Roughly speaking.

Until you understood all this perfectly, completeley forget about the camera's dynamic range.


Only then it will be time to dive into the brave new world of 16Bit histograms so few people have ever seen

[falconeye]

between about 25 and 230 is between 4-6 stops depending on contrast setting of your JPEG (and 5 stops in RAW)

What you are saying here is that the back-LCD histogram isn't the 8Bit histogram of the RAW. This is interesting and I must confess I've never heard about. Do you have a link or search phrase to follow it up? Google or the forum search returned nothing specific for me.

[Lowell Goudge]

What you are saying here is that the back-LCD histogram isn't the 8Bit histogram of the RAW. This is interesting and I must confess I've never heard about. Do you have a link or search phrase to follow it up? Google or the forum search returned nothing specific for me.

What I am saying is that the histogram does not represent a 50% change in value for 1 F stop.

I just took 4 shots of my screne, with my *istD, and SMC-105mm F2.8. Each shot was at 1/125 but I stepped doen in 1/2 stop increments on the apature.

The back histogram on the first shot (correct exposure) was just to the left of middle. each successive shot moved further to the left, such that after 2 full stops (i.e. f5.6) I hade moved 1 division out of 4.

This is also how the histogram behaves in a photo editor.

As I explained previously, the distribution of an image and dynamic range is NOT done in a log finction,

The histogram is linea over the central part, in terms of histogram value vs f-stops.

each f-stop has an average value of 45 greyscale either in raw or with contrast set to neutral. it necomes non-linear at the high and low end, with 1 stop compressed into about a range of 17 and the next to a range of about 7. therefore overall, depending on contrast setting, you have a dynamic range of between 8 and 10 stops, with 4-6 very well behaved, an additional 1 on each side that is not too bad, and the last one whioch is really grainy and poor.

[jeffkrol]

I've notice like with my old fuji that the histogram is divided into 4 parts, does anybody know how much each quarter represents in stops ?

after a few test shots at 2 stops apart I guess about 2 stops per quarter, its handy to know so as to make it easier to make accurate exposure compensations when judgig the information in the prteview histogram

Only the middle is linear I believe... The first and last "quarters" have more "stops" in each then the middle 2. Or maybe not....
http://www.adobe.com/digitalimag/pdfs/linear_gamma.pdf
Luminous Landscape Forum > Interpreting Histograms

[simons-photography]

I just took 4 shots of my screne, with my *istD, and SMC-105mm F2.8. Each shot was at 1/125 but I stepped doen in 1/2 stop increments on the apature.

The back histogram on the first shot (correct exposure) was just to the left of middle. each successive shot moved further to the left, such that after 2 full stops (i.e. f5.6) I hade moved 1 division out of 4.


As I explained previously, the distribution of an image and dynamic range is NOT done in a log finction,

yes thats about what it is for me and thats what I thought

[falconeye]

This is also how the histogram behaves in a photo editor.

Well, I understand everything about a histogram in a photo editor.
A photo editor histogram is the so-called images's N-Bit Histogram and has a linear scale, where the f-stop scale is logarithmic. There are no "middle parts" behaving differently.

I see two choices:

1. The sensor doesn't capture luminosity linearly (a bad thing to be compensated by tone mapping -- so one should know about).
2. The Pentax back LCD histogram isn't a photo editor's 8-Bit histogram.

I will investigate this and report my findings as soon as I have them. Thanks for the responses so far.

[simons-photography]

[*] The sensor doesn't capture luminosity linearly (a bad thing to be compensated by tone mapping -- so one should know about).

I think the sensor is linear but our eyes are not (same as our ears hence the decibel scale) hence the stops scale

[simonkit]

Another great feature of the K10D is it displays a full RGB histogram (use "up" arrow to select)
- I always use this option as its easy to clip highlights in a colour channel & not notice on the standard histogram


simon

[jeffkrol]

Well, I understand everything about a histogram in a photo editor.
A photo editor histogram is the so-called images's N-Bit Histogram and has a linear scale, where the f-stop scale is logarithmic. There are no "middle parts" behaving differently.

I see two choices:

1. The sensor doesn't capture luminosity linearly (a bad thing to be compensated by tone mapping -- so one should know about).
2. The Pentax back LCD histogram isn't a photo editor's 8-Bit histogram.

I will investigate this and report my findings as soon as I have them. Thanks for the responses so far.

Don't forget that a RAW histogram is fairly worthless. At the best you could get a RAW hist w/ gamma correction and the tone curve applied. That is somewhat closer than the 8bit jpg histogram that you see. And the reason "highlights" are somewhat more recoverable in RAW.
Only the jpg is blown sometimes.

[simons-photography]

yes I noticed that I was able to recover some blownout highlights in raw processing I am now wondering how much margin there is, why is part of the sensors range being used "off the scale" ?