Is Dynamic Range the new albatross?

[SirSeth]

I was rereading your response, and I don't understand how you are using the term "tonal relationships" or how that differs from dynamic range. I consider tones to be a color shade and a function of how a light ray is absorbed by a particular subject to product color on the subject. Having more tones is like going from a 16 count box of crayola crayons to that monster 96 count box (the one with the crayon sharpener in the side). So you have more tones or shades. Is that how you are using the term? I think of Dynamic Range this same way; basically, how many shades exist between pure white and black (including color). More shades mean there are smaller steps between similar shades (tones?), and the image you could create with the 96 box of crayons would have more dynamic range if you used all the shades. Please clarify if I have misunderstood your use of the terms.

Photographically, I believe with better dynamic range shadows are not forced to be blacked out because of a light object (or sky) in the image. Less contrast between light and dark means a more accurate exposure, but I often shoot high contrast landscape. As I understand it, Fuji handles this by basically merging two exposures captured at the same time (large pixel exposure is merged with smaller less sensitive pixel exposure all at once).

I do disagree that if a photographer doesn't use dodge and burn or layering techniques they are not making the best of their equipment. (if that is what you meant). I use these techniques effectively, but I would like not to have to. To analogize, I can get great Oly-like color from my Canon files, but I much prefer getting the colors right off the sensor with my Oly files. So much less work to get the pictures I envision when looking through the viewfinder.

Cheerio,
Seth

--
What if the hokey pokey really is what it's all about?

--
wallygoots.smugmug.com
wallygoots.blogspot.com

 

[Jay Turberville]

While I agree that at base ISO a good compact can take images that
are often indistinguishable from a DSLR, things tend to unravel
pretty quickly as gain increases.

Right. And that's because they have less dynamic range to begin with. In the case of your camera, I'm guessing that its is below 7 at the .25 Imatest threshold since the pixel pitch is smaller than that of the Sony 7Mp 1/1.8" sensor that does 7 stops at .25, and because Panasonic has a reputation for having more noise given a small sensor speck similar to a Sony. It might be as low as 6 stops the way the Simon complains. Assuming that, at ISO 400, you are down to about 4 stops of good quality DR and pusing into 5 or more stops means lots of noise.

At ISO100, with a well illuminated scene, my LX-1 (8mp Panny sensor)
is capable of taking excellent images. But by ISO400 anything at
mid-grey or darker is drowning in noise in a way that simply doesn't
happen with the E500.

Right. But the thing worth noting here is that with 6-7 stops of DR, excellent images can be taken.

Here's a 100pc crop from the LX-1 in which the usuable DR has
shrunk alarmingly

I don't think its is "allarming." Your camera may be operating with a two stop deficit. Its like starting at ISO 400 with an E-500. And frankly, going up two stops to ISO 1600 on an E-500 isn't exactly a joyride.

Admittedly the Panny compacts are notoriously noisy (as Simon
complains in almost every review), but I think most compacts have the
same problem.

Just not to the same degree. Otherwise Simon would be complaining similarly. And 5Mp on a 2/3" sensor has a greater fundamental potential than 8Mp on a smaller piece of silicon. Though being older designs, they may not be as highly optimized.

--
Jay Turberville
http://www.jayandwanda.com

 

[850_Dan]

I may well be wrong, but that image seems PP to me.

 

[Jay Turberville]

I was rereading your response, and I don't understand how you are
using the term "tonal relationships"

In the context of the above statement, I'm talking about how tone levels of different areas in the image compare to each other. For instance, in Bootstrap's original image, the sky colors were actually a fairly dark tone. When I altered the tone curve to bring up the somewhat darker tones of the train, I also ended up messing with those relatively dark sky tones as well. That tonal relationship has been set at some point. It may have been via a polarizing filter or it may have been through Bootstrap's post processing. But if I'm going to raise the brightness of the train, I'll have to do soemthing to "protect" the sky.

or how that differs from dynamic
range.

Camera sensors are linear capture devices. If I assume (and I know this may not be the actual case) that the sky was rendered dark on the original exposure (because that's how dark it was or because a polarizing filter darkened it), then DR isn't really the issue. If a camera with more or less DR was used, the tonal relationships between the train and sky tones would have been the same. In other words, the darker train isn't a DR issue, it is one of where the subject values fell on the scales of tones that are recorded. More DR would simply mean that there is more detail availble in the shadows (assuming the exposure was pushing to near saturation of the clouds).

I consider tones to be a color shade and a function of how a
light ray is absorbed by a particular subject to product color on the
subject. Having more tones is like going from a 16 count box of
crayola crayons to that monster 96 count box (the one with the crayon
sharpener in the side). So you have more tones or shades. Is that how
you are using the term?

I think in grayscale with photography. With "range of tonal values" I was referring to the subject's brightness limits, not to how finely divided the tonal gradations are. More DR means you get to record a greater 'range' of tonal values, not necessarily numerically more tones. The total number of tones comes from the number of sampling bits which doesn't necessarily equat to DR.

I think of Dynamic Range this same way;
basically, how many shades exist between pure white and black
(including color).

No. Dynamic range is not the number of shades. It is the range of photographic stops. It refers to the range of brightness in the original scene that you can usefully record, not to pure white or black. What's the brightest and darkest thing that I can record and still hold detail without too much noise? Now, what's the range in brightness (in stops) between those two values? That's dynamic range in a practical photgraphic sense.

More shades mean there are smaller steps between
similar shades (tones?), and the image you could create with the 96
box of crayons would have more dynamic range if you used all the
shades.

otally different concept. You could have a box of 96 crayons that were all subtle shades of middle or light tones. Imagine a box of pastels. Or you could have a box of 16 crayons that went from very white to very black with no color. The 16 crayons are analogous to more dynamic range. The 96 crayons merely provide a more diverse pallete, but very limited dynamic range. Ideally we want both!!

Photographically, I believe with better dynamic range shadows are not
forced to be blacked out because of a light object (or sky) in the
image.

Dynamic range with electronic sensors is limited by pixel saturation on the bright end and by noise in the shadows on the dark end. If your sensor has 8 stops of dynamic range and you light object is 8 stops brighter than your shadows, then you will be able to nicely record those two tones and the tones in-between. If the light object is 10 stops brighter, then you will have to decide which is more important, the shadows or the light object. Either that, or you need to use a grad filter, two exposures, or get a camera with more dynamic range.

Less contrast between light and dark means a more accurate
exposure, but I often shoot high contrast landscape.

More or less dynamic range doesn't mean a more or less accurate exposure. It merely means you can capture a greater 'range' of tones (not to be confused with a greater 'numerical count' of tones) from your subject.

As I understand
it, Fuji handles this by basically merging two exposures captured at
the same time (large pixel exposure is merged with smaller less
sensitive pixel exposure all at once).

I think that's right.

I do disagree that if a photographer doesn't use dodge and burn or
layering techniques they are not making the best of their equipment.
(if that is what you meant). I use these techniques effectively, but
I would like not to have to.

IMO, the use of such techniques is part of the art of making a photograph. The exposure is just the beginning. Part of the art is the creative altering of tonal relationships. We use polarizers, grad filters, fill lights and more to get the look we want. In the 'darkroom' we use dodging, burning, masks and color controls. With traditional black and white film, photographers would use red and yellow filters to darken skies. To me, fiddling with the tonal relationships is a part of the difference between taking a picture and making one.

To analogize, I can get great Oly-like
color from my Canon files, but I much prefer getting the colors right
off the sensor with my Oly files. So much less work to get the
pictures I envision when looking through the viewfinder.

I'm an advocate of doing it the easy way if the results are as good. Why not?

--
Jay Turberville
http://www.jayandwanda.com

 

[A Aquacelli]

I may well be wrong, but that image seems PP to me.

Poorly PP'd.

http://www.sound-effect.com/sounds1/animal/Dogs/barks2.wav

 

[knoblock]

I think latitude would be similar to the linear part of the sensor response as in film. If a scene did not have a full range of tones from black to white, like a concrete sidewalk, you can slide the exposure up and down the linear slope a bit and still retain detail across the range of values in the scene. So if you underexpose or overexpose you still get a usable image. I think you're right about the RAW development, since you would have to move the exposure slider to get an acceptable image from this if the captured tones were too light or dark. This is how development and printing would make use of the latitude in film. The image can be printed lighter or darker or middle without losing any of the original tones. Or select the contrast of the paper to extend the tones to fill black to white, similar to using levels.

For a scene with deep shadows and bright highlights, the range of tones in the scene would be nearly the same as those the film was capable of recording, and the latitude would be small.

I never thought about it, but I assume a film with limited latitude is just one that has a range close to that of most normal scenes, and therefore cannot stand much under or over exposure.

So latitude would be the dynamic range of the film. I always thought dynamic range came from audio technology and was applied to sensors. Dynamics is a term from music, meaning loud to soft.

--
http://www.flickr.com/photos/knoblock/ Equipment in plan.

Film will only become art when its materials are as inexpensive as pencil and paper. -- Jean Cocteau

 

[John Cozijn]

So latitude would be the dynamic range of the film. I always thought
dynamic range came from audio technology and was applied to sensors.
Dynamics is a term from music, meaning loud to soft.

which is why it is measured in decibels, when talking about sensors.

But that is generally how people are using the term here. What is meant I think is the number stops between a blocked highlight and a blown hightlight, and then qualified by what is "usable" at the bottom end.

And it is this last component that makes for a great deal of subjectivity and variation that cannot be captured in a tiff of step wedge because it really does depend on the nature of the image.

John

 

[John Cozijn]

But that is generally how people are using the term here.

That should have been:
But that is not generally how people are using the term here.

 

[Rriley]

and then there is the Fuji F30

http://www.digitalcamerainfo.com/content/Fujifilm-FinePix-F30-Digital-Camera-Review/Testing--Performance.htm

The Fujifilm FinePix F30 turned in an excellent performance. We've noted a handful of compact cameras lately with very high scores in the Low Quality measure at ISO 100, but they tend to be unimpressive at High Quality, and their scores look worse at higher ISOs. The F30 turns this phenomenon on its head: its 10.3 EV score at ISO Low Quality is good, but not outstanding, while its High Quality scores match up well with any camera short of a good DSLR.



--
Riley

real men get zippo haircuts

 

[marf]

... of different cameras ?

To my knowledge the DR is not fundamentally sensor size linked, but its possible that some sensors are better and some sensors are not good, and usually the cheap sensors are bad.

But there should be a comparison table somewhere.
--
regards
Martin

 

[fldspringer]

... of different cameras ?

To my knowledge the DR is not fundamentally sensor size linked, but
its possible that some sensors are better and some sensors are not
good, and usually the cheap sensors are bad.

But there should be a comparison table somewhere.
--
regards
Martin

There is where the problem lies. The Raw has to be converted. That requires software. Software handles different RAW files differently. Hard to make comparisins for that reason.

 

[Louis_Dobson]

To boil it down to one line, anything more than 4/3rds is wasted. Yes, I'll go along with that.

What bothers me is that I don't think there have been or will be technology jumps that let you get away with a smaller well size. What has been going on up to now is smarter ways of hiding the problem (Canon with their "plastic" look, which is NR at the sensor level, pure and simple, and everyone else with their signposted NR, which gets them slagged), and clever tricks to use more of the silicon for wells (NMOS for Oly, and whatever Canon have just done, and that's been done now, not much more to come).

One buys into a system, not this year's camera. Hence my interest in the D3 - I agree with you that for 8-10MP 4/3rds is optimal. I worry about where we go next. Oly have thought about this, and have a patent on a system that takes two exposures, ie realtime HDR. Well, OK, but more silicon is a simpler solution...
--
http://www.flickr.com/photos/acam

 

[Jay Turberville]

To my knowledge the DR is not fundamentally sensor size linked, but
its possible that some sensors are better and some sensors are not
good, and usually the cheap sensors are bad.

But there should be a comparison table somewhere.

There is where the problem lies. The Raw has to be converted. That
requires software. Software handles different RAW files differently.
Hard to make comparisins for that reason.

Actually, the same raw software handles the different raw files virtually the same way. The only difference would be slight differences in color correction. But since the goal is to give a neutral white balance of the same scene (black and white step wedge), that difference actually tends to level the playing field.

The only argument against standard processing that I can imagine really is that a particular algorithm might favor a particular sensor. That bit of favoring seems like it would be minor for CFA sensors. We'd certainly be better off than the mish-mash of testing we have now.

I do my tests by running the files through dcraw so that it applies no noise reduction, no sharpening, provides 16 bit files with no gamma correction, and good white balance. This minimizes the variables and provides us with a fairly level playing field and gives a nice picture of what the sensor actually delivers. You can see the linear output and even pick up on the slight bit of lens flare that is seen in the darker shadows that causes the response to become slightly non-linear.



--
Jay Turberville
http://www.jayandwanda.com

 

[SirSeth]

That makes more sense. And now I see that both more tones and greater dynamic range are desirable future image quality improvements. Not that I will refuse to use all the tools the digital darkroom provides me to make pictures with my current equipment. I actually would rather add lenses than cameras anyway, but I suspect that I will also continue to admire the shots taken with the S5 and wish for such advantages to be added to the list of advantages I enjoy in Olympus.

So the MP race slows down and high ISOs are cleaner with every new iteration of cameras. Meanwhile everyone has stabilization of some kind, most are adopting live view in some form, and most have given at least a crack at dust reduction, and there are more sealed bodies these days as well. It seems really exclusive features are fading. (ie Minolta's AS, Canon's IS, Oly's dust busting and live view). Everyone has to offer everything to be competitive. But let image quality be the next differentiator--and DR will be part of that. That's been FujiFilm's plan all along. How else do they get such high prices for their cameras?

Cheerio,
Seth

--
What if the hokey pokey really is what it's all about?
--
wallygoots.smugmug.com
wallygoots.blogspot.com

 

[Amin Sabet]

Its an interesting puzzle, but I don't think losing a stop of DR is
going have much practical effect on anybody's photography.

Jay, I don't agree. Shooting RAW gives about a stop of latitude or less on some of my cameras, and I really appreciate that stop. It can make a real difference.

The simple fact is that there is almost no DR testing done on
compacts. The better compacts deliver about a stop more DR than
you'd expect based on the simple pixel math that most people do.

Although I haven't done any testing, my experience has been consistent with that.

--
http://aminphoto.blogspot.com

 

[Jay Turberville]

Its an interesting puzzle, but I don't think losing a stop of DR is
going have much practical effect on anybody's photography.

Jay, I don't agree. Shooting RAW gives about a stop of latitude or
less on some of my cameras, and I really appreciate that stop. It
can make a real difference.

I goofed in my typing. I was referring to the approximate one-third stop that seems likely with the E-410/510. I agree that a full stop is significant - especially if it is a one stop loss starting at 8 stops of DR. But that said, losing even a full stop isn't a disaster. I've done quite well with cameras with a stop less DR than my E-system cameras in bright Arizona daylight.

--
Jay Turberville
http://www.jayandwanda.com

 

[Mikael Risedal]

Read this to understand DR, a good explanation at http://www.qpcard.se/BizPart.aspx?tabId=75 by Lars Kjellberg in English.

 

[Bootstrap]

post processed. There was no disclaimer that it wasn't.
--
Troll Whisperer
Bill Turner

 

[Bootstrap]

I may well be wrong, but that image seems PP to me.

Poorly PP'd.

If your posted image above is any indication, were I you, and thankfully I'm not, I wouldn't be making any claims or judgements as to what constitutes well or poorly post processed images.
--
Troll Whisperer
Bill Turner