Dynamic Range Perception : Poll/Survey

[George Chen]

"A" looks clearly better to me, while "B, C, D" are similar and not as good.

I have taken 4 clips of controlled lighting static scene test
captures, thanks to imaging-resource.com. I would like for a few
people to list in order, with ratings (1) being the 'best' and (4)
being the worst, what they believe looks like preserves both
hilight AND shadow details in teh best visible balance. Each
example is identified as (A), (B), (C) and (D) at the far right of
each image. I suggest using the 'EXPAND WIDTH' java funcitno button
that this site gives you as an option over teh posted images. Feel
free to play with your monitor's contrast and brightness controls
if you can't tell enough difference to give an answer. NOTE: LCD
monitors and some 'old' crt monitors will proably not be useful for
this comparison.

I will identify A, B, C and D in a day or so after I get a few
responses. BTW, NO cheating by going to imaging-resource and trying
to find the exact matches before giving your answer. Then your
answer would be biased, now wouldn't it?

For those who might not for some reason know what I am referring
to, here is a synthetic example at two extremes to make it clear:



Now, on to the poll. These samples had only one modification made
to them from teh orginal file(s): they were all resampled to the
same physical dimensions.



Thanks in advance for your assistance.

-Chris

Though not to sceintific standards, the images used above employ
the highest level of controls that I know of from the cameras being
compared that are avaialble.

 

[George Chen]

"A" looks clearly better to me, while "B, C, D" are similar and not
as good.

Woops, I only judged the right side of the image. Are the left and right sides from the same image? If we find an inverse relationship between the ratings for the right and left sides, the differences may be due to exposure differences.

 

[WmAx]

"A" looks clearly better to me, while "B, C, D" are similar and not
as good.

Woops, I only judged the right side of the image. Are the left and
right sides from the same image? If we find an inverse relationship
between the ratings for the right and left sides, the differences
may be due to exposure differences.

Yes, right and left sides are from teh same target capture. You can not estimate dynamic range by looking at just one side; as simploy exposure differences can dictate then. The goal is too see what people think looks like it has the most visible information in the light and dark areas simultaneously. The 'sample' better-worse image set dictates an exaggerated example set.

-Chris

 

[Carl Lind]

I have taken 4 clips of controlled lighting static scene test
captures, thanks to imaging-resource.com. I would like for a few
people to list in order, with ratings (1) being the 'best' and (4)
being the worst, what they believe looks like preserves both
hilight AND shadow details in teh best visible balance. Each
example is identified as (A), (B), (C) and (D) at the far right of
each image. I suggest using the 'EXPAND WIDTH' java funcitno button
that this site gives you as an option over teh posted images. Feel
free to play with your monitor's contrast and brightness controls
if you can't tell enough difference to give an answer. NOTE: LCD
monitors and some 'old' crt monitors will proably not be useful for
this comparison.

I will identify A, B, C and D in a day or so after I get a few
responses. BTW, NO cheating by going to imaging-resource and trying
to find the exact matches before giving your answer. Then your
answer would be biased, now wouldn't it?

For those who might not for some reason know what I am referring
to, here is a synthetic example at two extremes to make it clear:



Now, on to the poll. These samples had only one modification made
to them from teh orginal file(s): they were all resampled to the
same physical dimensions.



Thanks in advance for your assistance.

-Chris

Though not to sceintific standards, the images used above employ
the highest level of controls that I know of from the cameras being
compared that are avaialble.

-------------------------------------------------------------------

--DACB uncalibrated Samtron monitor
CarlL

 

[arbon]

This morning (getting light)uncalibrated A,B,C,D..
Later in morning (day light) Calibrated B,A,C,D..
This evening ( dark) still calibrated B,A,C,D

 

[WmAx]

O.K., it seems a few people who have responded do not seem to understand the concept of dynamic range. I should not have posted this survey on a whim very late at night(stupid me) and should have prepared a basic guide and summary to instruct some people. I suppose it was stupid of me to 'assume' everyone knew what to look for. It only appears that very few misunderstood my vague explanation(s), but does anyone think that I should invalidate the survey now? Also take into consideration that this survey is still only intended to guage people's perceptions of the images on their computers/monitors.

If 5 or more people reply to this and state 'AGREE' I will cancel this survey and reintroduce the same samples later(in 1-3 days) along with better instructions. If at least 5 people do not agree by the time i have 20 people reply with their survey answers, I will continue with this survey.

-Chris

 

[WmAx]

Please specify which ranking you want to count in the final survey results. You have provided 3 unique rankings, with two different results. I need 1 ranking.

Thanks.

-Chris

This morning (getting light)uncalibrated A,B,C,D..
Later in morning (day light) Calibrated B,A,C,D..
This evening ( dark) still calibrated B,A,C,D

 

[Mike Chinnock]

O.K., it seems a few people who have responded do not seem to
understand the concept of dynamic range. I should not have posted
this survey on a whim very late at night(stupid me) and should have
prepared a basic guide and summary to instruct some people. I
suppose it was stupid of me to 'assume' everyone knew what to look
for. It only appears that very few misunderstood my vague
explanation(s), but does anyone think that I should invalidate the
survey now? Also take into consideration that this survey is still
only intended to guage people's perceptions of the images on their
computers/monitors.

If 5 or more people reply to this and state 'AGREE' I will cancel
this survey and reintroduce the same samples later(in 1-3 days)
along with better instructions. If at least 5 people do not agree
by the time i have 20 people reply with their survey answers, I
will continue with this survey.

-Chris

--
Happy shooting

Mike Chinnock
Sanyo VPC-250, VPC-350X; HP C912; Fuji S602
Sunpak 611; Kenko VC-200 HI

 

[arbon]

B,A,C,D.calibrated monitor

Thanks.

-Chris

This morning (getting light)uncalibrated A,B,C,D..
Later in morning (day light) Calibrated B,A,C,D..
This evening ( dark) still calibrated B,A,C,D

 

[Wils Wong]

O.K., it seems a few people who have responded do not seem to
understand the concept of dynamic range. I should not have posted
this survey on a whim very late at night(stupid me) and should have
prepared a basic guide and summary to instruct some people. I
suppose it was stupid of me to 'assume' everyone knew what to look
for. It only appears that very few misunderstood my vague
explanation(s), but does anyone think that I should invalidate the
survey now? Also take into consideration that this survey is still
only intended to guage people's perceptions of the images on their
computers/monitors.

If 5 or more people reply to this and state 'AGREE' I will cancel
this survey and reintroduce the same samples later(in 1-3 days)
along with better instructions. If at least 5 people do not agree
by the time i have 20 people reply with their survey answers, I
will continue with this survey.

-Chris

--
Happy shooting

Mike Chinnock
Sanyo VPC-250, VPC-350X; HP C912; Fuji S602
Sunpak 611; Kenko VC-200 HI

 

[photobob]

I have taken 4 clips of controlled lighting static scene test
captures, thanks to imaging-resource.com. I would like for a few
people to list in order, with ratings (1) being the 'best' and (4)
being the worst, what they believe looks like preserves both
hilight AND shadow details in teh best visible balance. Each
example is identified as (A), (B), (C) and (D) at the far right of
each image. I suggest using the 'EXPAND WIDTH' java funcitno button
that this site gives you as an option over teh posted images. Feel
free to play with your monitor's contrast and brightness controls
if you can't tell enough difference to give an answer. NOTE: LCD
monitors and some 'old' crt monitors will proably not be useful for
this comparison.

I will identify A, B, C and D in a day or so after I get a few
responses. BTW, NO cheating by going to imaging-resource and trying
to find the exact matches before giving your answer. Then your
answer would be biased, now wouldn't it?

For those who might not for some reason know what I am referring
to, here is a synthetic example at two extremes to make it clear:

I have a bit of a problem with your extreme examples to make it "clear." If one can assume the camera in the "better" example above can make darker darks than we see and lighter lights (not just the gray shown), I would agree that it could have more dynamic range. But the range that is shown is less. I have a C-2100 that changes contrast when you change sharpness settings. On Soft, the contrast is so bad you can get darks and brights to look like the "better" sample, but really, the darks look like mud and details are bad across the range. If I try boosting the contrast in PS, the levels split up and make strange steps of tone that make horrid prints. In this case, the dynamic range is really lower as it doesn't properly use the extremes. What is good is to get good darks that still have details (not just going black, but still close) and lights that don't wash out to blank white. If blacks look dark gray and whites light gray, one might assume more range but that would depend on there being infinite steps between, and in digital photography, there are a limited few. Thus, a good camera needs to utilize the full 0 to 255 range well, not start at 50 and end at 205.

Now, on to the poll. These samples had only one modification made
to them from teh orginal file(s): they were all resampled to the
same physical dimensions.

Considering what I said above, I'll go with ABCD as best to worst. A has more detail in the lightest part of the white (upper right corner), and a fairly decent black. The detail in the tear or flap in the middle of the black seems a bit worse, but either the lighting or the shape of what is there is somewhat different from sample to sample, and it's hard to tell which. D has really poor detail in the highlights. B and C are similar in the highlights. I wish the blacks had more texture/gradations to go on; I'm not convinced that A is the best, but it's better to lose a bit of detail at near black than at the other extreme.

Thanks in advance for your assistance.

-Chris

You're welcome.

Bob

 

[Richard Bolt]

B A C D

 

[WmAx]

I have a bit of a problem with your extreme examples to make it
"clear." If one can assume the camera in the "better" example above
can make darker darks than we see and lighter lights (not just the
gray shown), I would agree that it could have more dynamic range.

That is the intention. Any camera with inferior dynamnic range will artificially force contrasts into extremes at both ends, where typicallly those extreme values were not present in the first place. In reality, their would be degree of latitude left for both ends.

I can not comment on the second part of your post until the poll is over(if 5 people do not 'agree' to stop it before the respondents hit 20).

-Chris

 

[photobob]

I have a bit of a problem with your extreme examples to make it
"clear." If one can assume the camera in the "better" example above
can make darker darks than we see and lighter lights (not just the
gray shown), I would agree that it could have more dynamic range.

That is the intention. Any camera with inferior dynamnic range will
artificially force contrasts into extremes at both ends, where
typicallly those extreme values were not present in the first
place. In reality, their would be degree of latitude left for both
ends.

I can not comment on the second part of your post until the poll is
over(if 5 people do not 'agree' to stop it before the respondents
hit 20).

-Chris

I knew what you meant with the examples, but knew of some problem if one were to take it at face value without a bit more information. The pictures from the Sony 717 are sometimes considered dull/flat due to lack of contrast afforded by the dynamic range supposedly supplied by the 14 bit A-D conversion. And I've read for some years about Canon cameras that keep the black level up a bit from 0. I like a good black somewhere in a picture unless it is foggy, else it reminds me of the poor darkroom technique of excessive exposure of print paper and yanking it from the developer before it gets too dark and quickly putting into the stop solution. Blacks often look gray without detail. To me, black without detail is reasonable (too dark to see) but gray is not.

I'm not surprised you can't comment on part 2. One thing I forgot to mention is the small recess or crack in the middle that does go a good black in the A dark section, showing ability to go darker than the look of the general strip.

Bob

 

[Azrifel]

Acer Aspire 15inch 1024x768 32 bit LCD, manually calibrated

A
B close to C
D

On my 17 inch CRT, also calibrated B looks a bit better.

--
Sander [Fuji602 SonyP1]
http://www.azrifel.org
http://www.pbase.com/azrifel/

 

[ianR]

I'm personally not a big fan if Imaging resources test shots for comparison.

Consider this in relationship to the dynamic range experiment. Pictures of 4 of the bells in the Davebox taken at random.

Look at the reflections of the lights in the bells, particularly the top row of lights in the bottom right hand corner bell.

The intensity of light in each case is different. Some lights are brighter than others, sometimes it's brighter at the top than at the sides etc.

When evaluating the charcoal bricks for example, they are in shadow, so how Will the image bottom right be affected when evaluating for dynamic range when they will clearly be in more shadow and therefore getting less light??? We would expect the bricks to give vvery little detail and they don't to my eye. That's the 10D.

Furthermore, many people are evaluating B as having greater dynamic range but it has a very distinct colour cast to it. The charcoal is grey/brown and the duster is pink.

If you are trying to evaluate detail in black to white then this must be the worst of the bunch as it's certainly not capable of producing a black at 0 and a white at 256.

I don't think this is a very good reference for evaluating dynamic range in general but it's a lot of fun.



regards
Ian

 

[WmAx]

It should be ratehr obvious why the lights will not be in the same places on teh bells. As it would be very difficult to place the cameras focal plane at teh exact distance in the test shots, and the reflections on that bell, like a mirror will be extrememly sensitive to any tiny movement, including the slight eclipse of the border of teh dave box that cut out the top row of lights on that bell on lower right. I have never recommended using the bell for a reference becuase of these very reasons.

As for the guaze and charcaol, if you notice i used only top row of the charcaol for this comparison(the bottom has changed slighlty over teh different camera shots due to a shift in teh flexible foil/whatever that had a different shadow cast on teh bottom two pieces in previous times. the positiong of everything else has remianed constant. Therefor I only consider the top charcoal section.

As for this not being 'ideal', i made this clear too. This is simply the 'best' reference shots out their. Due to the specifiic controlled lighting conditions and for the most part, the static shooting scene. 'Best' is a relative term you know, it does not mean 'valid', or even ultimately good...just best of waht's avaiable! Would i prefer, for example, an absolute measurment or more carefully controlled testing? Certainly. But unless you can refer me to a source that tests dynamic range objectively for cameras and publishes this informaition publicly; then what is your pont about critisizing the comparison? As well, i was also very clear in a couple of the replis that the purpose of this test is for me to see what the perceptual response to the images was, including the variables introduced by their typical computers/monitors/etc.

Will the image bottom right be affected when
evaluating for dynamic range when they will clearly be in more
shadow and therefore getting less light??? We would expect the
bricks to give vvery little detail and they don't to my eye. That's
the 10D.

The main section of the two top bricks has kept a constant position and shadow(excpet for the very far right small section of the right side brick) for every dave box test I have seen. I don't understand your comment:

"We would expect the bricks to give vvery little detail and they don't to my eye. That's the 10D."

Also, the 10D? The Canon 10D was/is not one of the samples in this test.

Furthermore, many people are evaluating B as having greater dynamic
range but it has a very distinct colour cast to it. The charcoal is
grey/brown and the duster is pink.

YOu will observe identical relative perceptions if you wish to convert the samples to greyscale.

If you are trying to evaluate detail in black to white then this
must be the worst of the bunch as it's certainly not capable of
producing a black at 0 and a white at 256.

Of course not. I don't know of any published set of cameras tests that demonstrates an ideal dyamic range measurement or reference image set. But, at least for reference, this is very useful IMO.

Have a nice day.

-Chris

 

[Jakeblu]

B,A,C,D.

17in CRT Adobe callibrated.

Jakeblu

 

[photobob]

I'm personally not a big fan if Imaging resources test shots for
comparison.
Consider this in relationship to the dynamic range experiment.
Pictures of 4 of the bells in the Davebox taken at random.
Look at the reflections of the lights in the bells, particularly
the top row of lights in the bottom right hand corner bell.
The intensity of light in each case is different. Some lights are
brighter than others, sometimes it's brighter at the top than at
the sides etc.
When evaluating the charcoal bricks for example, they are in
shadow, so how Will the image bottom right be affected when
evaluating for dynamic range when they will clearly be in more
shadow and therefore getting less light??? We would expect the
bricks to give vvery little detail and they don't to my eye. That's
the 10D.
Furthermore, many people are evaluating B as having greater dynamic
range but it has a very distinct colour cast to it. The charcoal is
grey/brown and the duster is pink.

Obviously, something is different in the bottom right picture. Looks like something has at least changed positions. And my problem with the test strips was that they did not seem to be lit the same. The dark ones in particular just didn't seem to have a lot to make a good evaluation of. Even the light ones seem to have shadows or wrinkles in different places, making it harder to compare.

I noted the color casts, of course, but the test was on dynamic range only, apparently.

I picked A as best because the white was a bright one but still with detail in the brightest area and the "black" was not as dark as a couple of them but had a deeply darker crack. I just wish I could see more detail in the dark gray spread and around the crack in the middle. Since we want dynamic range for the detail it can show near an extreme, that would seem important.

Oh. KDS VS-190i 19" CRT, calibrated by me on my perceptions with various pictures and checked with various test strips, including Phil's white to black one on the start of his reviews, on which it showed all the gradations, the last one being virtually black. I had thought of fooling with the Adobe gamma but think I must be about there already so didn't bother. I've been evaluating pictures on CRTs for decades.

Bob

 

[ianR]

It should be ratehr obvious why the lights will not be in the same

Obvious yes, it's set up to give equivalent exposure but invalid yes also. If it can't reflect the lights properly because they aren't exactly lined up the same, then you can't expect them to cast exactly the same shadow. Nor can you claim 'controlled lighting' Controlled for exposure perhaps but not controlled to ensure that exactly the same light is falling on the bricks from exactly the same angle. That fact that ther is any difference invalidates this as a useful standard.

As for the guaze and charcaol, if you notice i used only top row of
the charcaol for this comparison

You mean you'll get different results if you choose different data from the same test. In any research, they'll tell you this invalidates the test.

then what is your pont about critisizing the comparison?

The point about criticizing the comparison is that people take this kind of thing seriously and base buying decisons on it. They read here that camera X has more dynamic range because person XX did a scientific test on it and proved it to be the case. If it's a bit of fun then ok, but if it's to be used to draw concrete conclusions on which people might actually spend their money, then I feel it reasonable to state the antithetic view. Is criticism not allowed? I am trying to be objective and factual and there is nothing personal about it I assure you.

clear in a couple of the replis that the purpose of this test is
for me to see what the perceptual response to the images was,
including the variables introduced by their typical
computers/monitors/etc.

My point is, it's a bit of fun but it means precisely nothing.
I assume you were trying to prove something apart from monitor perception.
There are many other, and better ways of doing that.

Will the image bottom right be affected when
evaluating for dynamic range when they will clearly be in more
shadow and therefore getting less light??? We would expect the
bricks to give vvery little detail and they don't to my eye. That's
the 10D.

The main section of the two top bricks has kept a constant position
and shadow(excpet for the very far right small section of the right
side brick) for every dave box test I have seen. I don't understand
your comment:

No, I just disagree. I don't think the light is at all the same there and neither is it the same on the white duster either. If he's adjusting the light intensity to allow for the F stop or ISO if the camera, then the amount of light on the duster is different. If it's got more or less light to reflect then it'll look diferent and you can't judge the dynamic range.

White doesn't exist as a colour remember, it is just grey scale at 100% reflectivity-and that can be more than 100%. The only way to judge the white is if exactly the same light is falling on it from exactly the same angle, and colour and that clearly isn't the case here is it. Even if the angle and colour are the same there is demeonstably more light on in some than others. The camera may expose them the same but the duster has more or less light on it.It's a really flawed test for dynamic range.

"We would expect the bricks to give vvery little detail and they
don't to my eye. That's the 10D."

Also, the 10D? The Canon 10D was/is not one of the samples in this
test.

I chose the examples at random to see if the angle of lighting, and intensity was exactly the same in each. It isn't that's all. What does it matter if that camera was included in your test? That wasn't my point.

Furthermore, many people are evaluating B as having greater dynamic
range but it has a very distinct colour cast to it. The charcoal is
grey/brown and the duster is pink.

YOu will observe identical relative perceptions if you wish to
convert the samples to greyscale.

But it isn't greyscale. You can have a dynamic range from 0 o 256 in red or green or blue so does that mean you want a camera that can handle a good dynamic range in red but makes white look pink? Either the dynamic range is good in the colours as in the original or it isn't a good result. If it's made white pink then that doesn't prove that it has a good dynamic range.

If you are trying to evaluate detail in black to white then this
must be the worst of the bunch as it's certainly not capable of
producing a black at 0 and a white at 256.

Of course not. I don't know of any published set of cameras tests
that demonstrates an ideal dyamic range measurement or reference
image set. But, at least for reference, this is very useful IMO.

Well I can't see how valid conclusions can be drawn from it so its use is rather limited IMO.

Have a nice day.

You too

-Chris

Ian
(Sorry, had to trim it)