A2 histograms: live vs playback

I find my A2's playback histograms are very different from the live histograms. I set the exposure with the "Expose to the right" method so that a live histogram's right highlight is at its max but does not clip. After taking the shot, the playback histogram will look dramatically different in one or more ways:

1. Sometimes the playback histograms will show over exposure, i.e. highlight clipping, and sometimes they will show under exposure, i.e. highlight not at the max. I think how they differ depends on the image's tonal ranges, but I have not confirmed it yet.

2. The amount of over/under exposure in the playback histograms can be as much as 1 full stop.

3. The playback histograms' tonal distribution, i.e. contour, is very different from the live histograms'.

My assumption is that the playback histogrrams reflect what have been captured. Then the question is how can I use the live histogram to set an accurate ETTR exposure?

A related question is that the live histogram would look one way during the setting of an exposure, then changes very briefly to look a different way when the shutter release is pressed half way down (to lock focus and exposure), and then changes back again when the shutter release is pressed all the way down. What is going on? What is the significance, if any, of the brief live histogram display change?

TIA.

An experienced film shooter with a new A2.

hpycmpr said:

A related question is that the live histogram would look one way
during the setting of an exposure, then changes very briefly to
look a different way when the shutter release is pressed half way
down (to lock focus and exposure), and then changes back again when
the shutter release is pressed all the way down. What is going on?

Click to expand...

hpycmpr said:

What is the significance, if any, of the brief live histogram
display change?

Click to expand...

Those last comments are the clue to how the live histogram works.....

It operates from the EVF/LCD output, and NOT from the image sensor directly. Presumably this is to cut down the data that are processed to a volume that can be processed in real time.. instant by instant.

In effect you are getting a readout that is very much less 'data-rich' than the post shot histogram.

Now, quite WHY they are so very different on occasion, is a bit of puzzle to me too...

.... because they undoubtedly are!!

This does mean the live histo is less use than it might be from an exposure metering point of view.... [and I have noticed that the difference between the two histograms in the A2 is greater than in my much older Dimage 7ug, for instance]

All I can add is that I share your puzzlement, and believe it is because of the naturally more limited accuracy of the viewfinder image that is being employed to produce it.

Hey, all cameras have their positive and negative aspects. I hope this quirk in the A2 does not spoil your enjoyment of what is...

...one heck of an all-in-one camera package. ;-)

Off topic...

Did you know the A2 can communicate wirelessly with KM/Sony flashguns?
Did you know there is excellent free software for USB computer operation?
--
Regards,
Baz

hpycmpr said:

I find my A2's playback histograms are very different from the live
histograms. I set the exposure with the "Expose to the right"
method so that a live histogram's right highlight is at its max but
does not clip. After taking the shot, the playback histogram will
look dramatically different in one or more ways:

1. Sometimes the playback histograms will show over exposure, i.e.
highlight clipping, and sometimes they will show under exposure,
i.e. highlight not at the max. I think how they differ depends on
the image's tonal ranges, but I have not confirmed it yet.

Click to expand...

After almost three years with the A2, I have to admit I never check the histograms post-shooting. I do expose to the right a lot. If I had encoutered a lot of problems with clipped highlights/shadows, I probably would have looked into the matter and started checking my captured images, however I have had no such problems.

hpycmpr said:

2. The amount of over/under exposure in the playback histograms can
be as much as 1 full stop.

Click to expand...

I take it this is an estimation? If this is the case, I'd say that Houston, we have a problem!

hpycmpr said:

3. The playback histograms' tonal distribution, i.e. contour, is
very different from the live histograms'.

Click to expand...

I wouldn't worry about that too much.

hpycmpr said:

My assumption is that the playback histogrrams reflect what have
been captured. Then the question is how can I use the live
histogram to set an accurate ETTR exposure?

Click to expand...

I think the answer lies in your next question.

hpycmpr said:

A related question is that the live histogram would look one way
during the setting of an exposure, then changes very briefly to
look a different way when the shutter release is pressed half way
down (to lock focus and exposure), and then changes back again when
the shutter release is pressed all the way down. What is going on?
What is the significance, if any, of the brief live histogram
display change?

Click to expand...

Now this I have experienced. When the shutter button is half-pressed, I suppose, as Barrie has said, that the readout must come from the sensor. So now you know what to do to get an accurate histogram before shooting - press halfway.

That said, I rely on the live histogram and it works for me.

hpycmpr said:

TIA.

An experienced film shooter with a new A2.

Click to expand...

--
Gideon



PAW - Week 31

Thank you for confirming that you are seeing the same thing.

When shooting transparencies on my film camera, I use spot metering in manual mode for precise exposures. I would spot meter the highlight details, and adjust A or S until the exposure indicator shows just under 2 stop overexposure. This almost always gets me the correct exposure.

With the A2, I was hoping to do something similar since ETTR is about the same as how I meter exposures for transparencies. But A2 has no exposure indicator, so I thought that the next best thing is the live histogram. But if the live histogram cannot be relied upon for this purpose, I will have to find another way, such as:

1. Shoot raw and underexpose (somewhat) based on the live histogram, and rely upon raw conversion to recover the highlight. But this is just the opposite of ETTR!

2. Shoot a test shot with ETTR based on the live histogram. Check the playback histogram, guestimate and adjust exposure and shoot again. Repeat until the playback histogram shows ETTR.

3. Bracket the exposure.

2 and 3 only work when shooting speed is not an issue. Would like to hear how others deal with this problem.

So far I really like A2's design from an ergonomic point of view, but the numerous features and settings will take me a while to get familiar with. Until then, I won't bother with checking the image quality, etc.

Thanks for the off topic references. I'm aware of Minolta's wireless flash, but that is pretty low on my priority right now.

Where is the "excellent free software for USB computer operation"?

An experienced film shooter getting educated on A2.

Barrie Davis said:

hpycmpr said:

A related question is that the live histogram would look one way
during the setting of an exposure, then changes very briefly to
look a different way when the shutter release is pressed half way
down (to lock focus and exposure), and then changes back again when
the shutter release is pressed all the way down. What is going on?

Click to expand...

Barrie Davis said:

What is the significance, if any, of the brief live histogram
display change?

Click to expand...

Those last comments are the clue to how the live histogram
works.....

It operates from the EVF/LCD output, and NOT from the image sensor
directly. Presumably this is to cut down the data that are
processed to a volume that can be processed in real time.. instant
by instant.

In effect you are getting a readout that is very much less
'data-rich' than the post shot histogram.

Now, quite WHY they are so very different on occasion, is a bit of
puzzle to me too...

.... because they undoubtedly are!!

This does mean the live histo is less use than it might be from an
exposure metering point of view.... [and I have noticed that the
difference between the two histograms in the A2 is greater than in
my much older Dimage 7ug, for instance]

All I can add is that I share your puzzlement, and believe it is
because of the naturally more limited accuracy of the viewfinder
image that is being employed to produce it.

Hey, all cameras have their positive and negative aspects. I hope
this quirk in the A2 does not spoil your enjoyment of what is...

...one heck of an all-in-one camera package. ;-)

Off topic...

Did you know the A2 can communicate wirelessly with KM/Sony flashguns?
Did you know there is excellent free software for USB computer
operation?
--
Regards,
Baz

Click to expand...

The amount of exposure difference between the live and playback histograms seems to depend on the tonal range in the image. But I have not experimented enough to confirm that. The "1 full stop" difference is obviously my unscientific estimation. For example, if an image is shot with ETTR based on the live histogram (i.e. highlight barely extended all the way to the right, but w/o clipping), then the playback histogram will show one of two things:

1. The highlight can be "severely" clipped.

or,

2. The highlight can be "way" off to the left (about 1/8 of the histogram scale).

I use "1 full stop" to describe these extreme situations. The difference is NOT ALWAYS this dramatic, but there is ALWAYS some. Again, I think that the image's tonal range has something to do with it, and those with very high contrast seem to result in the biggest differences. If you should make such a comparison, I would appreciate knowing if we both need to call Houston, or just me.

Gideon01 said:

hpycmpr said:

2. The amount of over/under exposure in the playback histograms can
be as much as 1 full stop.

Click to expand...

Click to expand...

Gideon01 said:

I take it this is an estimation? If this is the case, I'd say that
Houston, we have a problem!

Click to expand...

Oops. Almost missed this important answer.

With the shutter button half-pressed, the live histogram only changes VERY briefly. Are you relying on this brief live histogram? This may be the solution, and may explain why your exposures have no problems. I'll definitely give it a try and report back. Thanks.

Gideon01 said:

hpycmpr said:

A related question is that the live histogram would look one way
during the setting of an exposure, then changes very briefly to
look a different way when the shutter release is pressed half way
down (to lock focus and exposure), and then changes back again when
the shutter release is pressed all the way down. What is going on?
What is the significance, if any, of the brief live histogram
display change?

Click to expand...

Click to expand...

Gideon01 said:

Now this I have experienced. When the shutter button is
half-pressed, I suppose, as Barrie has said, that the readout must
come from the sensor. So now you know what to do to get an accurate
histogram before shooting - press halfway.

That said, I rely on the live histogram and it works for me.

Click to expand...

I recall last time the histogram issue came up that someone said that the live histogram was only calculated using the green pixels of the EVF.

That would explain why the difference between the live and recorded histograms are different.

If so the difference should be very large when framing something dominantly blue or red.

--
Regards,
Gary Pearson

http://www.flickr.com/photos/garypearson/

Hi to All,

I was just about to post exactly the same as Gary has posted about the green channel issue, ...and I think it was Greg (Staxis) who called our attention to this fact.

--
Cheers, Feri

'I can look at a fine photograph and sometimes I can hear music. Ansel Adams.'

hpycmpr said:

When shooting transparencies on my film camera, I use spot metering
in manual mode for precise exposures. I would spot meter the
highlight details, and adjust A or S until the exposure indicator
shows just under 2 stop overexposure. This almost always gets me
the correct exposure.

Click to expand...

You could still do someting similar with the A2 in aperture priority mode. If you set the exposure compensation to -2, you can use the AEL set to 'Spot Toggle' to find the exposure you want. I too have a slide shooting background and I also used to meter in manual mode with the spot meter, but I used to use an 18% grey card as my target and I would aim for half a stop underexposure. This got it right for me almost all the time.
--
Greg

When you've got a moment, have a look at my newly updated site:
http://www.wrightphotos.co.uk
also http://www.wrightphotos.co.uk/FromeInFocus

Winner of the South West Rural section of the BBC's Picture of Britain Competition.

Ferenc MOGOR said:

I was just about to post exactly the same as Gary has posted about
the green channel issue, ...and I think it was Greg (Staxis) who
called our attention to this fact.

Click to expand...

Thanks for crediting me with this one. It was during me series about exposing to the right. I've now put the main points of this on my website:
http://www.wrightphotos.co.uk/ExpsRt/EposRt.htm

Don't forget that colour space you set and the WB that you use also affect the histogram.
--
Greg

When you've got a moment, have a look at my newly updated site:
http://www.wrightphotos.co.uk
also http://www.wrightphotos.co.uk/FromeInFocus

Winner of the South West Rural section of the BBC's Picture of Britain Competition.

As a former film shooter ,to my suprise,i also ran into the problem of finding correct exposure with the A2.

I never had any problem on all of my film Minoltas to find correct exposure, either in auto mode or complete manual or A/S mode i was never more then 1/3 to 1/2 EV off ,which is stil a lot on Velvia50 and makes the shot, more or less, useless.

On film cameras i had a average of 1-2, maybe at most 3, bad exposed frames on a 36 roll.

With the A2 i got 75% ,if i am lucky on a sunny day, the moment i went indoors the succes rate droped even more.

After discovering the fact i tried to fall back to my old method which is , although more work, foolproof; spotmetering in manual mode and using my knowledge how to estimate lightvalues of different materials or even use a graycard.

But that is not possible with the A2, it does not have a metering indicator in manual mode like the ones found on analog cameras: a bargraph which tells you how your exposure setting compares to 18% gray measured in EV/stops.
It has the histogram which appeares to do the same thing but it does not.

After discussing this in a post some time ago i realised what the problem is.
The A2, as all digi non-reflex cams, does not have a light meter.
Reflex cams do have a(or more then one) light meter.

The A2 uses the data coming from the image chip to calculate the lighting and unfortunatly the (a) imagechip is not well suited as a lightmeter,especialy in low lighting and/or high contrast scenes it has problems measuring light values which are on the edge (or over) the range of sensitivity of the image chip, in fact it cannot measure at all values which lie more the a few EV beyond the avarage of what it is metering.

The sensitivity of the chip depends on the gain to the chip which is controlled in such a way that a "reasonable" live preview is seen on the evf/mon.

The gain of the chip is applied to ALL pixels and therefore the whole chip is setup to measure only light which falls in a certain range; this range is certainly not from EV2 to EV20, my gues is that is somewhere in the region of 4-5 EV,at most.

The chip is perfectly capable of registering light values from EV 2 to EV20 but not in 1 shot/preview(!).

If the camera is aimed at a scene with ie an avarage of 10 EV it is capable of registering light between ie EV7.5 and EV12.5, everything above or below that is not registered.

You can observe this behaviour in a high contrast scene when you raise the exposure compensation, up to a certain amount there is a change on the histogram moving to the right but after aprox + 1.3 there is no more change: you ask the camera to make a measurement of something which lies beyond the sensitivity of the chip and thus it can not compute these values.

This is a fundamental difference with reflex cameras which use real light metering cells.

The 4-8-16 or even 32 cells are all independent of each other(!) and thus the camera can measure between EV2 and EV20 in 1 scene and therefore the camera can make a more realistic computation on how to adjust gain,contrast,brightness etc. to make a good shot.

You can see it as "minus" but i see it more a "feature" which is inevitable linked to the design of the camera type; you have a live preview with histogram but the price you pay is not having real light measurement but instead you have a "computation" from a live preview which works within limits.
This can be an advantage or a disadvantage depending on the scene.

Over time this will improve as image chips become more sensitive and the "problem" will also be less on bigger sized chips( or less megapixels) but for now it is the reality.

That said i still think it is a hell of a piece of typical minolta enginering and curse the unfortunate day that they pulled the plug and sold out to the devil.

To all the transparency shooters using spot metering in manual mode,

Thank you, thank you, thank you! Your presence and comments allow us to discuss how to achieve something equivalent on the A2 on the same wavelength.

Your transparency metering methods are either identical to mine, or very similar. Those are the best ways I know of to expose the chromes accurately. I recall one lab manager commented on my slides, saying, "Your slides are always so accurately exposed. Most of our customers don't get it right, and many blame it on us."

In addition to spot metering the highlight details to prevent blowouts as I described earlier, I also spot meter skin tones as an alternative, exactly like how robert576 does it:

http://forums.dpreview.com/forums/readflat.asp?forum=1024&message=21533104

My A2 metering holy grail is to figure out how to achieve these same results.

Is A2's live histogram based on only the green channel officially documented somewhere? Or is it objectively proven to be the fact?

In either case, that will explain why there is such a big difference between the live and playback histograms. But only if the playback histogram is rgb composite.

Is histogram based on a single channel a common practice in the lower end digicams? If I recall correctly, some DSLRs not only have rgb composite histograms, but also include all three rgb individual channel based histograms. That will satisfy those who absolutely do not want clippings in any channels.

Staxis said:

Ferenc MOGOR said:

I was just about to post exactly the same as Gary has posted about
the green channel issue, ...and I think it was Greg (Staxis) who
called our attention to this fact.

Click to expand...

Thanks for crediting me with this one. It was during me series
about exposing to the right. I've now put the main points of this
on my website:
http://www.wrightphotos.co.uk/ExpsRt/EposRt.htm

Don't forget that colour space you set and the WB that you use also
affect the histogram.
--
Greg

When you've got a moment, have a look at my newly updated site:
http://www.wrightphotos.co.uk
also http://www.wrightphotos.co.uk/FromeInFocus

Winner of the South West Rural section of the BBC's Picture of
Britain Competition.

Click to expand...

Thanks for suggesting this alternative. I'll definitely try it out. By "spot toggle", do you mean moving the spot metering point from the center? BTW, how big is A2's spot metering area? How effective is A2's spot metering in low light, i.e. at what EV can it support?

So far I have yet to play with moving either the focus point or the spot metering point. One reason is that on A2, theses two points seem to be tied together. In many situations, one would like to focus on one point, but meter at a different one. Hope I can do that on the A2.

Staxis said:

hpycmpr said:

When shooting transparencies on my film camera, I use spot metering
in manual mode for precise exposures. I would spot meter the
highlight details, and adjust A or S until the exposure indicator
shows just under 2 stop overexposure. This almost always gets me
the correct exposure.

Click to expand...

You could still do someting similar with the A2 in aperture
priority mode. If you set the exposure compensation to -2, you can
use the AEL set to 'Spot Toggle' to find the exposure you want. I
too have a slide shooting background and I also used to meter in
manual mode with the spot meter, but I used to use an 18% grey card
as my target and I would aim for half a stop underexposure. This
got it right for me almost all the time.
--
Greg

When you've got a moment, have a look at my newly updated site:
http://www.wrightphotos.co.uk
also http://www.wrightphotos.co.uk/FromeInFocus

Winner of the South West Rural section of the BBC's Picture of
Britain Competition.

Click to expand...

robber576, thanks for the great explanation, and an apology for getting your name wrong in a previous post.

What are your metering methods to make the best use of this A2 "feature"? In particular, how do you preserve the highlight details, and expose the skin tones correctly? Compared to how you meter transparencies, what is your exposure success rate on the A2 using these methods?

TIA.

I have a solution. I don't use the live histogram. I use the EVF only. If it looks good in the EVF the print will look good. The EVF is a live preview of what the image will look like with exposure changes and everything.
--
http://www.greatphotos.org

I've tried this a few times: I take a picture, then, without noving, flip between the live view and the image I have just taken. They are not identical.

I just tried it agin before posting. The picture I shot looked a little overexposed, whilst the live view looked OK.

--
Gideon



PAW - Week 31

hpycmpr said:

Thanks for suggesting this alternative. I'll definitely try it out.
By "spot toggle", do you mean moving the spot metering point from
the center? BTW, how big is A2's spot metering area? How effective
is A2's spot metering in low light, i.e. at what EV can it support?

Click to expand...

Not quite. The AutoExposureLock button has four possible settings. I have mine set on Spot Toggle (via a menu). This means that when I press the AEL button it will take a spot reading and lock the exposure to that value until I press AEL again. This gives you an instant sopt meter reading whenever you want it.

hpycmpr said:

So far I have yet to play with moving either the focus point or the
spot metering point. One reason is that on A2, theses two points
seem to be tied together. In many situations, one would like to
focus on one point, but meter at a different one. Hope I can do
that on the A2.

Click to expand...

They are tied together but there is no reason why you can't do this. I could do it with my Minolta film SLRs even though they only had one focus point in the middle of the frame and the spot meter always read from the centre of the frame as well.
--
Greg

When you've got a moment, have a look at my newly updated site:
http://www.wrightphotos.co.uk
also http://www.wrightphotos.co.uk/FromeInFocus

Winner of the South West Rural section of the BBC's Picture of Britain Competition.

hpycmpr said:

Is A2's live histogram based on only the green channel officially
documented somewhere? Or is it objectively proven to be the fact?

Click to expand...

I recall seeing it documented somewhere but I could not give you a reference. It is fairly easy to check by using a mixture of red, blue and green targets.

hpycmpr said:

In either case, that will explain why there is such a big
difference between the live and playback histograms. But only if
the playback histogram is rgb composite.

Click to expand...

hpycmpr said:

Is histogram based on a single channel a common practice in the
lower end digicams? If I recall correctly, some DSLRs not only have
rgb composite histograms, but also include all three rgb individual
channel based histograms. That will satisfy those who absolutely do
not want clippings in any channels.

Click to expand...

Some DSLRs do indeed have histograms for each channel. Equally, some don't.

I use the 'expose right' method a lot. To nail it with the A2 - or my Sony A100 - I will establish a base exposure and then bracket, just like with slide film. Bracketing using digital equipment is a lot cheaper than with transparency stock. You could also ensure that you have a little space at the right hand end ofthe histogram to allow for errors - then it would be 'expose nearly right'.

--
Greg

When you've got a moment, have a look at my newly updated site:
http://www.wrightphotos.co.uk
also http://www.wrightphotos.co.uk/FromeInFocus

Winner of the South West Rural section of the BBC's Picture of Britain Competition.

Are you confirming that you are seeing the same thing as I do?

Gideon01 said:

I've tried this a few times: I take a picture, then, without
noving, flip between the live view and the image I have just taken.
They are not identical.
I just tried it agin before posting. The picture I shot looked a
little overexposed, whilst the live view looked OK.

--
Gideon



PAW - Week 31

Click to expand...