bits per channel, f-stop range, & shadow detail?

[smoody]

I am looking for a solution to give me better shadow detail. I'd like to minimize posterization that happens in shadows (as we all would, no doubt). The 1Ds and Kodak Pro Back produce 12bit per channel output. The Leaf digital backs are at 14bits and 16bits per channel. I've only played with a Leaf back once and did not have a chance to do any testing.

My question is...do the higher bits-per-channel of digital backs help in capturing shadow detail more accurately? Or maybe the 12 stops of range?

I love the 1Ds, but when I do blow-ups, the shadows look pretty awful. I'm assuming it's limitations of the sensor technology and can't be fixed, but I'd like to know if that's the case.

Thanks,

Smoody

 

[Ranjan]

I am looking for a solution to give me better shadow detail. I'd
like to minimize posterization that happens in shadows (as we all
would, no doubt). The 1Ds and Kodak Pro Back produce 12bit per
channel output. The Leaf digital backs are at 14bits and 16bits
per channel. I've only played with a Leaf back once and did not
have a chance to do any testing.

If u are shooting stills then one way to increase the shadow details is layer blending wherte u shoot 3 shots -1 0 +1 exposure then blend the layer in PS.

My question is...do the higher bits-per-channel of digital backs
help in capturing shadow detail more accurately? Or maybe the 12
stops of range?

I guess bits-per-channel & 12 stops of range are related to each other, its like faster ASA higher contrast, slower ASA lesser contrast.

I love the 1Ds, but when I do blow-ups, the shadows look pretty
awful. I'm assuming it's limitations of the sensor technology and
can't be fixed, but I'd like to know if that's the case.

Dont know if its a limitation of sensor technology, but surely it can be overcome in the time to come.

--
Ranjan
Professional photographer.
http://www.photosig.com/go/users/view?id=11993

 

[philip]

There are two main components involved. The sensor and the A/D converter.

Both have to be matched to work effectively. The dynamic range of the sensor obviously will affect the ability of the system to capture shadow detail. As was indicated, multiple shots can be combined to improve the shadow detail, by layering images taken at different exposure settings. This of course doesn't help much for non-still photography.

The A/D converter is the heart of the image translation process from sensor (analog) to image processing and storage (digital). There are a number of factors that affect image quality here. Bits (which again relates to dynamic range) and Liniarity/noise (which affects the usefullness of the bits). The greater the number of bits of the converter, (8-bit, 10-bit, 12-bit, 14-bit, 16-bit, the greater the dynamic range, and conversly the better the shadow detail. The usefulness of those bits however, is affected by the linearity of the converter (how faithfully does the output data track the input values) and how much noise is present in the electornics. As the nosie floor increases, it effectively destroyes the usability of the lower bits, so a 14-bit converter may in fact only be able to resolve 12-bits of useful data.

So, theoretically a higher bit-depth converter should have greater dynamic range, but you really need to know what the noise and linearity characteristic of the A/D converter subsystem before you can properly evaluate two systems.

Even though one may be 14-bits of data, it may only really be 12-bits of useful data.

hope this helps.

regards,

pg

--
Philip G.
http://www.cgrafx.com

 

[Gaetan J.]

So, theoretically a higher bit-depth converter should have greater
dynamic range, but you really need to know what the noise and
linearity characteristic of the A/D converter subsystem before you
can properly evaluate two systems.

Even though one may be 14-bits of data, it may only really be
12-bits of useful data.

good explanation but you are mixing up 2 different things. Dynamic and bits are not related. You can sample a sensor with any number of bits you want. The only difference will be the number of steps between 2 values.

So if you have more bits you will eliminate posterization.

Sensor dynamic on the other hand will dictate at what intensity you will loose details. (either way, shadows and highlights).

Linearity can be handled by software

--
Gaetan J.

 

[Michael Long]

Are you doing 12-bit conversions?

Or are you shooting JPG and also correcting after the fact in PS?

 

[smoody]

Yes, I'm shooting raw and then converting to 16 bit tiff. It shows up in the raw processing application (CaptureOne -- with both 1Ds and 10D). It's also visible in Kodak Pro Back images (in a full-sized sample I dowloaded from their site). Both Canon and the Kodak Back have 12 bits per channel color resolution.

I tend to suspect that it's an artifact of the sensor technology because it's not as if 12 bits per channel doesn't give you enough colors to handle shadow transitions from black to very dark colors. I haven't seen it in any Sigma SD9 sample images yet, but I haven't found any images with deep shadows w/o a lot of noise (no disrespect to SD9 users -- I suspect that foveon technology will one day make me reconsider my investment in equipment unless does an about face in the next couple of years).

• • • • • • • • • • • • • • -- - - - - - - - - - - - SMoody

http://www.pbase.com/smoody
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

 

[philip]

So, theoretically a higher bit-depth converter should have greater
dynamic range, but you really need to know what the noise and
linearity characteristic of the A/D converter subsystem before you
can properly evaluate two systems.

Even though one may be 14-bits of data, it may only really be
12-bits of useful data.

good explanation but you are mixing up 2 different things. Dynamic
and bits are not related. You can sample a sensor with any number
of bits you want. The only difference will be the number of steps
between 2 values.

So if you have more bits you will eliminate posterization.

Well.. yes.. but it also is a direct correlation to dynamic range as well. Think of a digital audio system. The more bits you have the more subtle the changes in audio levels you have, but also the more broad the range of levels you can have and still maintain a smooth enough transistion from descrete step to descrete step. Thus there is a correlation between number of bits and dynamic range.

Sensor dynamic on the other hand will dictate at what intensity you
will loose details. (either way, shadows and highlights).

On the other hand, here you are absolutely correct. If the sensor dynamics don't allow for sufficent range of light capture, the rest of the system is pointless, and thus this is a critical component any imaging system.

Linearity can be handled by software

Linearity can be handled by software as long as it is sufficiently algorithmic. If it dances around to randomly from step to step, just like noise, you will lose low order bits, as the data they represent is meaningless.

--
Gaetan J.

--
Philip G.
http://www.cgrafx.com

 

[Iliah Borg]

I am looking for a solution to give me better shadow detail. I'd
like to minimize posterization that happens in shadows (as we all
would, no doubt). The 1Ds and Kodak Pro Back produce 12bit per
channel output. The Leaf digital backs are at 14bits and 16bits
per channel. I've only played with a Leaf back once and did not
have a chance to do any testing.

My question is...do the higher bits-per-channel of digital backs
help in capturing shadow detail more accurately? Or maybe the 12
stops of range?

As Ranjan noted, for stills you can blend bracketed shots. The step and number of shots vary, depending on the camera. Making posters 5 years ago w/Kodak DC290 (it was just an experiment, but pretty much a successful one), I was shooting 5 frames per subject, -2, -1, 0, +1, +2. The posters were 6x3'. With my current D1 I'm still making 5 shots

exposed for Zones 8, 5, 3, 2, 1 and blending them. I found this to be the easiest way to go for me.

Also, with stills you can use scanning backs for MF/LF, they seem to be not so expensive now. You can even modify an ordinary scanner to make a scanning back. I've done this several times since 90-s, and made scanning for several cartoon films using these devices. Basically any MF scanner with moving sensor is a good thing to start with.

New Leaf scanners also look wonderful, for example Leaf Valeo full-frame (I'm not sure what they mean by full-frame , having 16 bits ADC in 11Mpix (4056 x 2684, ISO 25-200) model and 14 bits ADC in 6Mpix (3150 x 2100, ISO 125-250) model, claiming to produce 12 and 11 stops respectfully.

http://www.creo.com/global/products/digital_photography_leaf/leaf_valeo/default.htm
And I'm afraid to ask for the price

 

[Ricki Boswell-Challand]

Iliah,

I've done similar for both digital images and scanned transparencies in photoshop, but I've never been happy with the results (at least not without a lot of manual work).

Which way have you found to be the most reliable and least effort way of achieving good blends?

thanks,
Ricki

http://www.mountainhippies.com
--

Why is it, the further I get from civilisation, the more civilised things become?

As Ranjan noted, for stills you can blend bracketed shots. The step
and number of shots vary, depending on the camera. Making posters 5
years ago w/Kodak DC290 (it was just an experiment, but pretty much
a successful one), I was shooting 5 frames per subject, -2, -1, 0,
+1, +2. The posters were 6x3'. With my current D1 I'm still making
5 shots
exposed for Zones 8, 5, 3, 2, 1 and blending them. I found this to
be the easiest way to go for me.

 

[Gaetan J.]

So, theoretically a higher bit-depth converter should have greater
dynamic range, but you really need to know what the noise and
linearity characteristic of the A/D converter subsystem before you
can properly evaluate two systems.

Even though one may be 14-bits of data, it may only really be
12-bits of useful data.

good explanation but you are mixing up 2 different things. Dynamic
and bits are not related. You can sample a sensor with any number
of bits you want. The only difference will be the number of steps
between 2 values.

So if you have more bits you will eliminate posterization.

Well.. yes.. but it also is a direct correlation to dynamic range
as well. Think of a digital audio system. The more bits you have
the more subtle the changes in audio levels you have, but also the
more broad the range of levels you can have and still maintain a
smooth enough transistion from descrete step to descrete step. Thus
there is a correlation between number of bits and dynamic range.

no, there is no relation/correlation. There is only a relation/correlation if you decide to use a finite qte for a bit. But since this is not the case ...

--
Gaetan J.

 

[Andrew Rodney]

My question is...do the higher bits-per-channel of digital backs

help in capturing shadow detail more accurately? Or maybe the 12
stops of range?

More bit depth does NOT provide more dynamic range. They are completely different. Higher bit depth will allow you to edit a file with the range it has and greatly reduce posterization. But you'll gain no more shadow detail.

Dynamic range is the range of tones between black and white. So more dynamic range allows you to alter shadows without blowing highlights. But bit depth doesn't do anything in this regard.

Think of a stair case that's 50 feet tall. The length is the dynamic range. The number of actual steps in the staircase is the bit depth. If you have a staircase that's 75 feet and has 75 steps, you have more tones than the 50 foot staircase with 75 steps. Having more steps doesn't make the staircase any larger, it only breaks up the steps finer.
--
Andrew Rodney
http://www.digitaldog.net

 

[Ranjan]

More bit depth does NOT provide more dynamic range. They are
completely different. Higher bit depth will allow you to edit a
file with the range it has and greatly reduce posterization. But
you'll gain no more shadow detail.

Dynamic range is the range of tones between black and white. So
more dynamic range allows you to alter shadows without blowing
highlights. But bit depth doesn't do anything in this regard.

Think of a stair case that's 50 feet tall. The length is the
dynamic range. The number of actual steps in the staircase is the
bit depth. If you have a staircase that's 75 feet and has 75 steps,
you have more tones than the 50 foot staircase with 75 steps.
Having more steps doesn't make the staircase any larger, it only
breaks up the steps finer.
--
Andrew Rodney
http://www.digitaldog.net

Hi Andrew, Very well explained
--
Ranjan
Professional photographer.
http://www.photosig.com/go/users/view?id=11993

 

[MarkMcCombs]

that has nice shadow detail, density, etc. It was taken by Isaac Alongi with a 1DS. I'm not sure what your problem is. He shoots in jpg mode, also. here's his site... http://www.isaacalongi.com/

Mark

 

[smoody]

I'm visiting my parents at the moment, so I don't have access to my full size 1Ds files, but here is a 100% crop taken from the Kodak Pro Back sample gallery on the Kodak website. I don't expect Kodak to put images on their site that highlight this problem with digital sensor technology, but you can see it slightly here where the shadow goes from black to gray. As I said, it happens on all of my digital SLRs.

I don't think that I have a problem. I think it's more likely that the photographer you mentioned probably has a technique to smooth this out. Did you have any really dark shadows in you images? Oh, and it isn't as much of a problem at 20x30, but crank it up to 40x60 and it creeps into the interpolation in pretty obvious ways.

Here's the 100% crop. I only use LCD monitors w/DVI connections and it might be more on screen with this configuration. If you look at the diagonal shadow line across the image, it goes from black to purple/noise to near black to gray (don't take my color names literally -- the names are for illustrative purposes).

• • • • • • • • • • • • • • -- - - - - - - - - - - - SMoody

http://www.pbase.com/smoody
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

 

[smoody]

Andrew, thanks for clearing that up for me!

• • • • • • • • • • • • • • -- - - - - - - - - - - - SMoody

http://www.pbase.com/smoody
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

 

[smoody]

I was having a chat with someone close to Leaf and they told me there's a Leaf single-shot back with a 22+ megapixel sensor on the way that is very close to medium format in size . I think their problem at the moment is that their 11megapixel sensor is the same size as the 1Ds sensor so it's hard to tell the difference when you're just playing around with the two.

Of course, everyone expects this will happen sooner or later, but hopefully it'll happen sooner (and be priced at a point where it'll matter to me ;-)

• • • • • • • • • • • • • • -- - - - - - - - - - - - SMoody

http://www.pbase.com/smoody
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

 

[Iliah Borg]

Ricki,

Which way have you found to be the most reliable and least effort
way of achieving good blends?

Some preparation to be done. First I've shoot calibrated gray scale (you can get it from Stouffer, for example) covering full 7 stops, bracketing 1/3 eV + - from the correct exposure (0.1D is 1/3eV, so is the difference between steps in the scale). Second, I've made curves in a way that they cut the noisy steps to 10 in shadows and blown-up highlights to 245 and counted steps reproduced (Shift-Pencil while making curves editing if you need straight lines). After that 5 shots overlapping full 3 stops (9 steps) were stacked in layers with their appropriate curves. I chaged opacity of each layer in a way to have correct eyedropper readings of the total pyramid.

Actually your second question about civilisation seems to be much more interesting

 

[philip]

Actually there is a direct corration, because sensors are non-linear (this is why you DON'T get posterization in the mid-tones and high-lights. You can see this affect in all scanner systems. The lower the bit depth of the scanner the more posterization there is in the shadow detail. 8-bit/ch scanners do not handle shadow detail as well has 10-bit/ch or 12-bit/ch scanners.

And yes I realize this is also a function of the sensor, but in the real wolrd they are a matched pair, you can't separate the two.

Since we don't have specifications for the sensor dynamics/characteristics, the only thing that one can evaluate is the complete system specs. Thus, we look for imaging systems with higher numbers of bits. A 12-bit conversion system should produce better shadow detail than an 8- or 10-bit system.

Actually in many cases a 12-bit system is actaully a 14- or 16-bit converter, in which the last 2-bits are dropped because the noise and/or linearization errors are great enough to invalidate their usefulness.

The staircase analogy is great but only describes a linear system, in reality sensors are non-linear, think of a staircase where the height of the steps gets progressively larger. If we start at the top of the stairs (hightlights), the steps are very close together, as we progress down the staircase (moving towards shadows) the steps get further and further apart. If we don't start with enough bits, we end up with insufficient steps at the bottom of the staircase. So from a design standpoint our choices are to reduce the height of the stairs (decreasing the dynamic range of the system) or we take progressively larger steps (which results in posterization).

no, there is no relation/correlation. There is only a
relation/correlation if you decide to use a finite qte for a bit.
But since this is not the case ...

--
Gaetan J.

--
Philip G.
http://www.cgrafx.com

 

[Michael Long]

Oh, and it isn't as much of a problem at 20x30, but crank it up to
40x60 and it creeps into the interpolation in pretty obvious ways.

Just what are you expecting out of 35mm format? There was a time not too long ago I would have been VERY happy with a clean, sharp, non-grainy 16x20 print using 35mm Velvia.

At 40" you've got maybe 85-90 DPI of original data (after cropping to 3x4 aspect ratio). There's no way I'd expect NOT to see posterization issues at that size.

 

[Gaetan J.]

--
Gaetan J.