D80 histogram vs histogram on computer

I have a D80 which I am quite happy with. I am shooting RAW,
postprocessing with Capture One LE.

I have the colour space set to Adobe RGB, and try to fit the histogram
to the right without blowing highlights.

But when I then open the image in C1LE (or PS CS2, or Graphic
Converter), the histograms have loads of "space" to the right side.
What is happening?

Can anyone give me a clue?

As an aside, I came to the D80 from an FA. What I miss most is the
shutter release button with the ring around, and then the viewfinder.

TIA

Martin

 

Your images are still a bit underexposed. It is hard to get that tiny
histogram that you see on the camera to match the one which shows up on
your very large monitor.
Jim

 

Hello,
The in-camera histograms are produced from the in-camera jpeg, even if
you are shooting raw (to see this, shoot the same thing twice but with
different white balance, and notice that the colour histograms
change). Note that a jpeg is always produced and is embedded into the
raw file (that is what you are looking at when you zoom-in in-camera).
The point is that different converters (the camera itself, C1 etc)
apply different tone curves; so your histograms will look different.

What you can do is always use a particular curve in your camera (the
contrast setting; you can even upload one) and work out, by trial and
error, and for a given white balance and sharpening setting, what the
in-camera histograms look when you actually get clipping in the raw
data. But it's too much work.

Anyway, it's good to keep in mind that the histograms you see will be
very strongly affected by white balance (and contrast settings), even
when shooting raw: if you look at actual raw data, it is almost
impossible to clip the red channel without first clipping the green
one, even under tungsten light; but shoot a red rose in broad daylight
and try to get the red channel histogram not to clip in the camera...

 

So, completely normal and the best I can do apart from bracketing?

To me, it is just striking that often there is (almost) nothing on the
right-most quarter of the computer histogram, even if I try to hist if
with the camera.

Martin

 

It takes awhile, but you'll get used to the shutter
button.

However, for exposure control, after you've checked the
histogram, try moving your "photo information" display
one setting to the right, so that instead of a histogram
you have a "blink on overexposure" display, and use that
to fine tune exposures. I typically use manual
exposure, but if you like the camera to do it you can
adjust the Exposure Compensation to get it right on
target. The trick is to crank in extra exposure until
it blinks, and then back off just one step (defaults are
1/3 of an f/stop, but can be set to 1/2). It requires
"wasting" two or three shots, but developing digital film
is cheap...

You can then learn to accept blown highlights in certain
areas, and adjust on others. For example, if there is a
reflection, or an electric lamp in the scene, let it
blink, and adjust on the brightest area that should not
be blown. The single biggest advantage of the blink on
overexposure display is that you can see where it is
overexposed, which a histogram does not tell you.

Basically, if you leave it at 1/3rd f/stop steps you can
nail exposure to within 1/3rd f/stop virtually every
time. That's close.

 

I agree with acl, what you can do is turn down the contrast, saturation
and sharpening as far as possible in the menus. This will give a jpeg
histogram that is closer to the raw file but still not raw.

I also agree about the blinking overexpose mode, I switch between those
often. And do look at the 3-color histogram, in particular skies can
blow out the blue channel making an irretrievably cyan sky when the
composite histogram looks fine in some applications.

BTW the viewfinder may not look as good as your old film camera but it's
one of the better viewfinders available for cropped digital and is
excellent considering the price of a D80.

 

No, it not normal. You need to increase the exposure by either switching to
manual mode or by using exposure compensation. If the right quarter of the
histogram is blank, then you are underexposing by about 1 stop.

Jim

So, completely normal and the best I can do apart from bracketing?

To me, it is just striking that often there is (almost) nothing on the
right-most quarter of the computer histogram, even if I try to hist if
with the camera.

Martin

 

This isn't unusual. The histogram you see on the camera LCD is based
on the embedded JPEG within the RAW file. It is therefore in 8-bit
color so it's not capable of showing you all the headroom you're
getting with RAW images.

Having one full quarter of the histogram flatline sounds a bit
excessive, but depending on variations in meter calibration and the
types of scenes you're shooting it's certainly possible when combined
with the JPEG/RAW discrepancy.

It's usually OK to clip the right-hand edge of the in-camera histogram
when shooting RAW, but you have yo learn from experience how much you
can get away with.

Some people I know who shoot RAW exclusively change the default JPEG
settings on their cameras, turning contrast and saturation way down,
to make the JPEG histogram more accurately represent the RAW file.
These settings don't change what's in the RAW file, and they'll make
your JPEG files look flat & washed out (which is why only RAW-only
shooters use this technique), but you'll get a histogram that looks
closer to what's in the RAW file.

 

Back to the OP's question. I'd be interested in knowing why the upper edge
of the D80's histogram should be so different from the one shown in the
editing software, and the responses so far are not particularly satisfying.

The histogram, for the D80, and AFAIK all SLR's, is after the fact, and does
not involve SLR sensor technology. Nor should the camera's JPEG settings
have much of an effect on the histogram:. changes in gamma affect mostly
the midtones, and not the min or max values. Sharpness, and the output
color profile, likewise have little or no effect on overall brightness.

My own personal guess is that it's the raw converter's "exposure" setting
that is the culprit in this case, but it's only a guess.

Though I can understand the desire for precision, from a practical
standpoint, 25% of the histogram is only a half a stop error. Luckily in
this case it is on the safe side, since what you really want to avoid is
clipping.

 

Completely normal for any dSLR. You have to understand, that the light-meter in
any SLR design system is only a rough rough approximation from what you see, and
what you see is already an approximation of the full-frame that will be
recorded. Plus, if you change lenses from anything than other the default
"normal" lens of 55mm e.q. for which the light-metering system was optimized,
any other lens' light path will not hit those sensors correctly. Forget about it
even trying to cope with a long zoom range for any accuracy. Just learn to know
the differences between particular lenses, their focal lengths, and the
resulting true histogram that will be provided. Then mentally adjust before each
shot for that "fudge factor" that you'll have to remember for each lens used and
each focal length range on any zoom lens.

You're just finding out something about all dSLR's that most dSLR owners turn a
blind-eye too (like so many things they do).

This problem of any auto-exposure or metering system in any SLR light-path
camera has existed since the SLR design was first implemented.

Nothing new. Perfectly normal. Perfectly useless for anyone requiring more
precise metering of their subjects.

Oh, and don't keep your eye too far away from that viewfinder lens. Any light
that enters through the lens where your eye is at on that "much desired" optical
viewfinder will also upset the auto-exposure readings. This is just a few of
things they didn't want to tell you when they were going on and on about how
dSLR's are so superior to all other camera designs.

SURPRISE!

 

Have fun with your P&S kiddie toys, moron.

 

If you shoot RAW, there is no point setting the color
space to aRGB. That only changes the way the JPEG
is processed, which is part of the difference you are
seeing between the in camera histogram and the
post-processed histogram.

For RAW, you are best off setting all of your "in
camera" conversion parameters to the least possible
processing. But even then, you have to understand that
you are looking at the histogram for a JPEG processed by
the camera and then later comparing that to the
histogram for an image you've processed using different
parameters with an external RAW conversion program.

Two distinctly different images! There is no reason
they should be the same.

But if you are using the RAW file, you'll want the
histogram to be closer to what that is, so don't set it
up for any significant amount of processing on the in
camera JPEG.

True, once you've set the JPEG processing to something
less than aggressive. I'm not at all certain how much
effect one can expect to see from that, but you might
find it interesting to experiment with it and find out.

See how much difference there is when you change the
JPEG parameters for camera processing.

 

Perhaps they are not. However, to demonstrate the point, I took a
random shot with my D200, raw+jpeg. This one, in fact:
http://www.pbase.com/image/90542027
(it was meant to be used to test some noise reduction technique I
thought of-a complete disaster). I then processed the raw with
rawshooter, everything at 0, and opened both the converted file and
the camera jpeg in photoshop (the camera settings were all to 0 or
medium, ie sharpening normal, tone compensation/contrast to 0,
saturation to normal etc). I also made sure the WB of the file
converted from raw was as close as I could get it to the jpeg. So,
here are the two histograms:
http://www.pbase.com/al599/image/90541934
(the top is from the jpeg, the bottom from the raw-converted file). As
I said, I had all controls in the raw converter zeroed out.

This isn't too surprising: if you use more than one converter
regularly, it is blatantly obvious that they use completely different
curves when they convert (eg nikon's converters don't just use a
simple gamma curve, as far as I can tell etc-but I did no detailed
tests to work out the curve, although it should not be that hard with
patience).

Well, besides WB, tone compensation (ie contrast in Nikon-speak) does
have an effect, although less than in the screenshots of the
histograms I already posted (about half the change seen there). I can
show examples from my d200 if you want. Also switching bet srgb and
argb has about the same change in the in-camera histograms, at least
(and under tungsten light, with bright red objects-I suppose it
strongly depends the colours). I tried shooting them with my compact
but it couldn't focus in the low light...

 

Are you sure about that "half a stop"? That would make the D80
histogram different from most others. In Photoshop, and on Canon
cameras, the X axis is the gamma-corrected pixel value. The distance
between the midpoint and the right edge represents about 2.5 stops
exposure change, not the one stop you'd expect if the X axis represented
linear intensity. And 25% of the histogram is more than one whole stop,
not half a stop.

Dave

 

Why 2.5 stops? It depends on the curve used during conversion. I
showed this before, but here it is again:
http://www.pbase.com/al599/image/90541934
This is the exact same image converted through rawshooter (lower
histogram) and the jpeg produced by the camera (upper histogram), both
"converters" at their defaults. Obviously, rawshooter uses a curve
with a longer shoulder by default. So you can't make a blanket
statement like "the upper half is 2.5 stops", unless you specify the
curve used. Isn't this obvious?

Some raw converters let you see the exact curve used to map from
linear data to final result (raw magick, for example, and ufraw also,
I think). Others let you play with shadow contrast/highlight contrast
sliders (or they may call it shadow compression/highlight compression
etc, this is the opposite presumably). These allow tweaking the
shoulder/toe of the curve. I don't see why people assume that if you
don't touch these (or your converter doesn't have them), or you don't
modify the curve, then a simple gamma curve is applied.