Reading mode: Light Dark

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

Go to last post
Hello Philip, nice talking to you
philip said:
your correct with respect to bit depth by itself, but not in
respect to bit depth as it relates to an A/D system. A/Ds have a
fixed scale.
Click to expand...
yes, every a/d converter has a fix scale (maybe some dont but I am not aware of these) but the scale is not the same for every converter. Thus a bit is not always the same bit between the systems
philip said:
The reference system (voltage) for the A/D is almost
always a fixed scale value
Click to expand...
the reference voltage is not a scale. It's a fixed value.
philip said:
(In some cases there may be a selectable
scale for a few ranges).
Click to expand...
could be, I dont know
philip said:
Thus the A/D converter does have a dynamic range.
Click to expand...
yes but's it's not the bit system, it's the analog voltage it mesure that is received from the device (in our case a light sensor). the converter dynamic range is for the voltage input.
philip said:
A true 16-bit A/D converter does have more dynamic range than an
8-bit converter, within the scope of a specific hardware
application.
Click to expand...
not nescessarily, your 8 bit converter may be able to receive 1 to 100 volt at the input and your 16 bit may only be hable to receive .5 to 10 volt

(It's easier to stay away from 0 volt) And we have not yet talked about ability of the converter to differntiate between 2 discreet value :-)
philip said:
When you are design a system, you are specifically
assigning a conversion range for the A/D system and as such
assigning a speicific quantity to each bit.
Click to expand...
yes
philip said:
So if you replace an
8-bit A/D with a 12-bit A/D you are either resolving smaller
quantities or increasing the range that can be caputured.
Click to expand...
yes, that's exactly what I tried to say, i.e. the number of bit as nothing to do with the dynamic since it can be one or the other

but the range you can capture is dictated by the sensor, not the converter,(they both may be part of the same chip)
philip said:
Or better
still you can do both. Increase the dynamic range and resolve more
detail.
Click to expand...
there is 2 dynamic range involved here. The first is the dynamic range of the sensor. The second the dynamic range of the converter.
philip said:
--
Philip G.
http://www.cgrafx.com
Click to expand...
--
Gaetan J.
 
Wynn Aker said:
would differ with that, if I recall. The dynamic range is not the
one foot length of the ruler, rather the ability to measure the
ruler.
Click to expand...
I beg to differ, what you are talking about know is "precision" not dynamic range. Try to think of a scope instead.
...
Wynn Aker said:
That digital audio (CD) system at home? 16 bit digital, with 2 to
16th power or 96db dynamic range. Lots! Without getting off-topic
into digital audio and its relative strengths or weaknesses
Click to expand...
I prefer tubes, they are less acurate but sound sweater :-) Ever see a tv with the channel 1 ? Sorry to drift OT
...major cut here...

have fun

--
Gaetan J.
 
Sounds like you know a thing or two about audio, eh?

I think we're in fundamental disagreement, or at least misunderstanding of terms, but that's what I love about America.

All the best!

:-)
Gaetan J. said:
Gaetan J. said:
would differ with that, if I recall. The dynamic range is not the
one foot length of the ruler, rather the ability to measure the
ruler.
Click to expand...
I beg to differ, what you are talking about know is "precision" not
dynamic range. Try to think of a scope instead.
...
Gaetan J. said:
That digital audio (CD) system at home? 16 bit digital, with 2 to
16th power or 96db dynamic range. Lots! Without getting off-topic
into digital audio and its relative strengths or weaknesses
Click to expand...
I prefer tubes, they are less acurate but sound sweater :-) Ever
see a tv with the channel 1 ? Sorry to drift OT
...major cut here...

have fun

--
Gaetan J.
Click to expand...
 
Wynn Aker said:
Sounds like you know a thing or two about audio, eh?
Click to expand...
Enough to get into trouble without getting fried. Used to do small mods to my Marshall amps. Used to have a jcm800, wich got stollen with all my guitars (robber did not know that the old thing under it, the jmp, was worth 5 times more :-) and an early JMP, sweet thing it was (same as a plexy but in a different casing)
Wynn Aker said:
I think we're in fundamental disagreement, or at least
misunderstanding of terms, but that's what I love about America.
Click to expand...
these threads are fun :-)
Wynn Aker said:
All the best!

:-)
Gaetan J. said:
Wynn Aker said:
would differ with that, if I recall. The dynamic range is not the
one foot length of the ruler, rather the ability to measure the
ruler.
Click to expand...
I beg to differ, what you are talking about know is "precision" not
dynamic range. Try to think of a scope instead.
...
Wynn Aker said:
That digital audio (CD) system at home? 16 bit digital, with 2 to
16th power or 96db dynamic range. Lots! Without getting off-topic
into digital audio and its relative strengths or weaknesses
Click to expand...
I prefer tubes, they are less acurate but sound sweater :-) Ever
see a tv with the channel 1 ? Sorry to drift OT
...major cut here...

have fun

--
Gaetan J.
Click to expand...
Click to expand...
--
Gaetan J.
 
Wynn Aker said:
Wynn Aker said:
but then a bit is not a unit of mesure. It as no value in itself.
So, 0 (zero) can represent total darkness and 1 (one) total light
(white). or even the reverse. You just wont have details between
the too.

Or if you will you can mesure something using inch or millimeter
(no fractions allowed :-) with a ruler (somebody allready used this
analogy) the dynamic range is the lenght of the ruler, the bit
depth is how much unit available to "describe" what you are
mesuring.
Click to expand...
would differ with that, if I recall. The dynamic range is not the
one foot length of the ruler, rather the ability to measure the
ruler. If you have a ruler which only indicates inches, then you
only have a total of 12 discrete steps of measurement, which is a
very small dynamic range. However, if you use thousandths of an
inch, you have 12 thousand total discrete measurements, which is a
much bigger dynamic range (bigger, in fact, than the 4096 discrete
steps of a 12-bit digital measurement).
Click to expand...
Your textbooks are (probably) not wrong. However, you're missing the fact that the theoretical maximum dynamic range of a recording format isn't the same thing as the actual dynamic range of the recording itself.

The theoretical dynamic range of any digital format is defined by the number of bits. The maximum dynamic range is the difference between the smallest and greatest values you can represent. An 8-bit format can represent 8 steps of data without having to clip the data at either end or without having to compress it.

However, as regards the actual recording, the dynamic range is the difference between the smallest value you can capture and the greatest value you can capture. In other words, the length of the ruler.

Part of the confusion here is caused by the fact that the number of bits in your image format controls more than just the theoretical dynamic range. It also affects the level of quantization error. This is sometimes mistakenly thought of as an indication of greater dynamic range, but it's really something else.

Quantization is the process of converting the original analog value into a digital value. Having more bits means each value can be captured more accurately. For example, two points in the image may have slighly different brightness values. With 12-bit samples, the difference is maintained, but with 8-bit they both come out to the same value.

The improved sample resolution has a big practical impact on image quality. With RAW files, saving 12-bits per pixel allows you to do things like bring up detail from shadow areas. Small differences between brightness values are increased, which makes the detail more visible. But those small differences might not be present if you only had 8-bits per pixel to begin with. Likewise, you can pull down highlights to preserve detail that might otherwise appear washed out.

Mike
 
Wynn Aker said:
Disagree. The "dynamic range" is a measurement of the resolving
ability of a given "range."
Click to expand...
With Dynamic range, you have to start at some level of black and go to max highlight just before burn out. The problem is, where do you start measuring the black and at what level of noise? No one has an established, published spec that I know of that tells manufacturers at what point they are getting usable data past a level of noise. This is why you can see two scanners from two companies (say Polaroid and Microtek) both identical and both manufactured by Microtek show different spec's for dynamic range. The same was true when Agfa made desktop scanners (they were actually made by Microtek and had an Agfa label). Agfa was always very conservative with their specs.

So what I'm saying here (and you're welcome to disagree) is that the measurements are being made with a rubber ruler. If you can find ANY type of ISO or similar body that has an exacting specifications for dynamic range measurement of scanner data, I'd love to hear about it.

The spec's are getting so bogus now that I've seen dynamic ranges well over 4.0 which exceeds film (well perhaps not XRay or Litho but you know what I mean).
Wynn Aker said:
Remember: dynamic range is defined as
the ratio, in decibels, of the largest measureable value to the
smallest.
Click to expand...
Where's the shadow noise in the calculations here?

--
Andrew Rodney
http://www.digitaldog.net
 
and descriptions are probably getting in the way. It's obvious that at least 4 or 5 of us have EE, DSP, signal theory, etc. training. We're probably at the level where we're in "violent agreement!"

This has been a great thread! Much better than some of the nasty ones that fly around sometimes!

Best to all :-)

( big snip )

Wynn
 
Mike Fulton said:
Specifically, an image format with 8 bits per pixel has a
theoretical maximum dynamic range of 8 stops. Likewise a 12-bit
format has a theoretical maximum dynamic range of 12 stops.
Click to expand...
Exactly! We work with 8 bits per color and if you do the math, we can create 16.7 million colors. We can only see maybe 12 million and you'd be hard pressed to actually be able to output more than 70-80,000 colors in CMYK (other RGB devices will be much larger but you get the point).

You can mathematically create 16.7 million colors but you can't see a fraction on a computer display (the gamut is far too small). Humans can't even see 12 of the million colors above simultaneously.

If I have a file that's 8 bits per color, do I have 16.7 million colors? No.
Mike Fulton said:
BUT... that's only refering to the theoretical capacity of the
image format itself. How many steps it can represent without
clipping or compression. It doesn't say anything about the image
being captured or the actual real-world capabilities of the capture
device.
Click to expand...
Exactly!

--
Andrew Rodney
http://www.digitaldog.net
 
But I love good reproduction. Of sound. Yeah. That's what I meant.

I have a Classe, Parasound, AudioQuest, Thiel system (strictly CD) that I enjoy too infrequently, but when I do, I remember why I love good music.

It's funny, but one of the things that came up in this thread was saving a down-rez sound recording on a high-rez system. You still get down-rez sound. That was a great analogy.

Yes, this has been a good one.

Wynn
 
I'm totally clueless when it comes to technical discussion of scanning equipment, and I won't pretend to know what I don't know. I'll have to take your word for all of that.
Andrew Rodney said:
Wynn Aker said:
Disagree. The "dynamic range" is a measurement of the resolving
ability of a given "range."
Click to expand...
With Dynamic range, you have to start at some level of black and go
to max highlight just before burn out. The problem is, where do you
start measuring the black and at what level of noise? No one has an
established, published spec that I know of that tells manufacturers
at what point they are getting usable data past a level of noise.
This is why you can see two scanners from two companies (say
Polaroid and Microtek) both identical and both manufactured by
Microtek show different spec's for dynamic range. The same was true
when Agfa made desktop scanners (they were actually made by
Microtek and had an Agfa label). Agfa was always very conservative
with their specs.

So what I'm saying here (and you're welcome to disagree) is that
the measurements are being made with a rubber ruler. If you can
find ANY type of ISO or similar body that has an exacting
specifications for dynamic range measurement of scanner data, I'd
love to hear about it.

The spec's are getting so bogus now that I've seen dynamic ranges
well over 4.0 which exceeds film (well perhaps not XRay or Litho
but you know what I mean).
Andrew Rodney said:
Remember: dynamic range is defined as
the ratio, in decibels, of the largest measureable value to the
smallest.
Click to expand...
Where's the shadow noise in the calculations here?

--
Andrew Rodney
http://www.digitaldog.net
Click to expand...
 
Mike Fulton said:
Specifically, an image format with 8 bits per pixel has a
theoretical maximum dynamic range of 8 stops. Likewise a 12-bit
format has a theoretical maximum dynamic range of 12 stops.
Click to expand...
This assumes a linear mapping of luminance to digital values, but that's not how it's done. If it was done this way, you'd have too much precision in your high values and too little in your low values. Think about it, when you look at the pixel values in photoshop, {128,128,128} isn't one stop darker than {255,255,255}. Nor is {1,1,1} one stop darker than {2,2,2}.

Exposure is inherently logarithmic, so you want to map luminance to digital values logarithmically. Given this, you can't make any statements about the theoretical maximum dynamic range that can be expressed in any given number of bits.
-harry
 
While we are here...

Here goes a method I use to make shadow masks (and I know many people using same principles in different variations), which are very useful for Layer blending to push-process the image, shadow noise reduction, neutralizing color cast in shadows, as well as a lot of other image improvement tricks. It originates from the "wet" process, where it was quite popular for quality and very expencive jobs. Being modified for digital, it is now within the reach of everybody.

Essentially, the trick is to borrow shadow mask for RGB Background layer from K plate of CMYK image duplicate .

Start with 8-bit RGB image (some modification allow for application of same method to 16-bit as well).

Make a duplicate (Image -> Duplicate).

Convert a duplicate to Custom CMYK (Image -> Mode -> Convert to Profile -> Destination Space -> Custom CMYK -> Dot Gain =0 -> Separation Type -> UCR -> Black Ink Limit=60 - some variations may be considered here, from approx. 40 to 70, - it would affect how strongly we pull the shadows)

Go to Channels palette, highlight the K plate, select it with Ctrl-A and copy with Ctrl-C.

Return to the original image, go to Layers palette, right-click on Background and choose "Duplicate Layer". Now we need to add Layer mask (Layer -> Add Layer Mask -> Reveal All), then switch to Channels Palette and activate Background copy Mask and paste K plate into it with Ctrl-V, then invert with Ctrl-I.

Now we have a shadow mask for the layer, and to make it more useful we can adjust levels and curves for this mask, so as to emphasize shadows, as well as apply Gaussian blur to make the effect softer.

To smoothen shadow noise we can blur Background copy itself, and the mask will allow only shadows to be blurred on the composite image. We can blur pretty strong, and apply only fraction of the effect by controlling Opacity.

Making another copy of Background and using same mask, but probably with other values for blur, levels and curves adjustment, we can push shadows out or pull them (to add contrast and snap).

Instead of Background copy we can use the image exposed for shadows.
 
Mike Fulton said:
Specifically, an image format with 8 bits per pixel has a
theoretical maximum dynamic range of 8 stops. Likewise a 12-bit
format has a theoretical maximum dynamic range of 12 stops.
Click to expand...
An 8 bit per pixel format could "theoretically" have a dynamic range of 255 stops, where the brightest value is mapped to 255, and the lowest to 0.

The dynamic range is what the sensor and system is able to accurately register. The number of bits simply gives the precision.

Or to put it another way, having more bits doesn't increase your dynamic range. It simply allows you to more precisely distinguish individual color tones within that range.

If you were to linearly map a 4-stop range to an 8 bit sensor, the darkest colors would be mapped to 0-63, mid-tones to 64-190, and highlights to 191-255.

More practically, I could easily map a 12-stop sensor to an 8-bit per pixel format. It's just that each stop would be linearly mapped to just 21 values.

What this exercise DOES show is that if we do get higher dynamic range sensors, then 8 bit jpgs are going to have a rough time, as higher precision is going to be required to get smoother tones with less posterization.
 
I will try to say something here.

Start with a photo of a landscape, including a white sheet and bushes, on a sunny day. Then the cd/m2 value of that picture would have a maximum of about 1,000 cd/m2 on the sheet and a > 0 cd/m2 minimum value within the bush shadows. This is the SCENERY dynamic range.

Shooting a photo using a 8-bit camera having a 50:1 range, will result in capturing cd values from 1,000 to 20. This camera will divide the 880 cd/m2 dynamic range into 256 shades of gray. Changing the camera to a 12-bit camera, having the same 50:1 range, will result in 880 cd/m2 dynamic range divided into 4096 shades of gray. This is the CAMERA dynamic range and the CAMERA shades of gray.

Scanning a photo of the same scenery at the same time will result in a SCANNER dynamic range of, say log3. That is 1,000 shades of gray. However, these shades of gray is out of the SCENERY dynamic range of 880 cd/m2, persupposed the photo film has a dynamic range of 50:1. (Because, I think we all can agree about that even the best scanner cannot do magic with the cd/m2 range below 20 and above 1,000, as it is not even captured by the photo film.)

So we are talking about SCENERY dynamic range, CAMERA dynamic range and SCANNER dynamic range. The SCANNER dynamic range seems to be nothing but CAMERA shades of gray.

Perhaps someone else have something to say about my guess.
 
smoody said:
I love the 1Ds, but when I do blow-ups, the shadows look pretty
awful. >
Click to expand...
What kind of blow-ups do you mean ?

Are you judging it on prints or on screen ?

If you mean screen blow-ups than it might be a problem of bit-depth. If you are speaking of prints than it could be a problem of color space.

I detected color space problems (-> posterization) in my prints especially in dark areas printing on an epson or fuji frontier.

Konrad
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top