Measurung dynamic range...

Bart van der Wolf schrieb:
>
> "Volker Hetzer" <firstname.lastname.TakeThisOut@ieee.org> wrote in message
> news:eavk36$sa7$1@nntp.fujitsu-siemens.com...
>> Hi!
>> I'd like to measure the dynamic range of my digicam.
>> How would I go about it?
>> Are there special targets for this?
>
> The T4110 is ideal for the purpose:
> <http://www.stouffer.net/TransPage.htm#transmission%20step>
I have contacted them and will have a look at the other links you provided.

Thanks a lot for helping!
Volker

--
For email replies, please substitute the obvious.
 >> Stay informed about: Measurung dynamic range... 

Volker Hetzer wrote:
> Hi!
> I'd like to measure the dynamic range of my digicam.
> How would I go about it?
> Are there special targets for this?
>
> Lots of Greetings!
> Volker
> --
> For email replies, please substitute the obvious.

Really measuring dynamic range is quite a difficult job. Part of the
problem is that there are two different definitions of dynamic range.
A second is that one must have a camera and software to use RAW file
format.

One quick and dirty test is to shoot a pic with lens cap on.

Using a photo editor, poke around and sample what appears to be lighter
areas- noise. Take a bunch of pixel brightness values (using color
picker). Average these values by squaring each value and summing, then
take square root of total. This will give you the noise level.
 >> Stay informed about: Measurung dynamic range... 

stauffer.TakeThisOut@usfamily.net wrote:

>
> Really measuring dynamic range is quite a difficult job.

In reading my post I see I really left out a lot, and didn't explain
myself very well. The problem is accurate radiometry/photometry.
Accurately measuring light is not a trivial problem- the instruments
are expensive and need frequent calibration. The big problem is
getting black references. White references are easy to do, and don't
need super accuracy. The problem is, what is the brightness, or even
reflectivity of a black reference.

To simplify the math, lets look at Black and White, say an 8 bit grey
scale. There can be up to 256 values in the image. But the best
available black inks or paints have a reflectivity of about 2-3%. If
we use a single even illumination, then a black and white printed chart
has a dynamic range of 50:1 or less. Thus, we can't use a printed
chart to test a camera with a dynamic range of 250:1 or 500:1.

I am working on a low cost "light trap" using a pringles can and
self-adhesive black felt. If this works out I intend to write an
article on it for a camera magazine, along with some stuff on measuring
dynamic range and flare performance. I intend to use this first on my
scanner, but will use it for some tests on my cameras too.
 >> Stay informed about: Measurung dynamic range... 

stauffer RemoveThis @usfamily.net wrote:

> stauffer RemoveThis @usfamily.net wrote:
>
>
>>Really measuring dynamic range is quite a difficult job.
>
>
> In reading my post I see I really left out a lot, and didn't explain
> myself very well. The problem is accurate radiometry/photometry.
> Accurately measuring light is not a trivial problem- the instruments
> are expensive and need frequent calibration. The big problem is
> getting black references. White references are easy to do, and don't
> need super accuracy. The problem is, what is the brightness, or even
> reflectivity of a black reference.
>
> To simplify the math, lets look at Black and White, say an 8 bit grey
> scale. There can be up to 256 values in the image. But the best
> available black inks or paints have a reflectivity of about 2-3%. If
> we use a single even illumination, then a black and white printed chart
> has a dynamic range of 50:1 or less. Thus, we can't use a printed
> chart to test a camera with a dynamic range of 250:1 or 500:1.
>
> I am working on a low cost "light trap" using a pringles can and
> self-adhesive black felt. If this works out I intend to write an
> article on it for a camera magazine, along with some stuff on measuring
> dynamic range and flare performance. I intend to use this first on my
> scanner, but will use it for some tests on my cameras too.
>
Procedures for Evaluating Digital Camera
Sensor Noise, Dynamic Range, and Full Well Capacities
http://www.clarkvision.com/imagedetail/evaluation-1d2

Gets around these problems, but you need access to the raw
data. Other tests typically use transmission targets.

A good way to make a light trap is a cone painted with
glossy black paint. The reflections from the painted surface
go deeper into the cone. At the bottom of the cone, have
a deep hole ~1mm diameter. A long pipe also works, as in
Figure 1, 2 at:

Dynamic Range and Transfer Functions of Digital Images
and Comparison to Film
http://www.clarkvision.com/imagedetail/dynamicrange2

Roger
 >> Stay informed about: Measurung dynamic range... 

Roger N. Clark (change username to rnclark) wrote:

> A good way to make a light trap is a cone painted with
> glossy black paint. The reflections from the painted surface
> go deeper into the cone.
So the camera looks into the pointy end?
Otherwise it seems to me that the light gets reflected back.

Lots of Greetings!
Volker
 >> Stay informed about: Measurung dynamic range... 

stauffer.RemoveThis@usfamily.net wrote:
> stauffer.RemoveThis@usfamily.net wrote:
>
>> Really measuring dynamic range is quite a difficult job.
>
> In reading my post I see I really left out a lot, and didn't explain
> myself very well. The problem is accurate radiometry/photometry.
> Accurately measuring light is not a trivial problem- the instruments
> are expensive and need frequent calibration. The big problem is
> getting black references. White references are easy to do, and don't
> need super accuracy. The problem is, what is the brightness, or even
> reflectivity of a black reference.
I've got a light trap from basiccolor, so that's not much of a problem
for me. (Can't recommend it though, for EUR70 you get some cheap wood
and plastic stuff.)


> To simplify the math, lets look at Black and White, say an 8 bit grey
> scale. There can be up to 256 values in the image. But the best
> available black inks or paints have a reflectivity of about 2-3%. If
> we use a single even illumination, then a black and white printed chart
> has a dynamic range of 50:1 or less. Thus, we can't use a printed
> chart to test a camera with a dynamic range of 250:1 or 500:1.
I can do that by putting the stuff against a window in a darkened room or
even a box. Not sure yet.
What I will probably need is a large translucent grey slide too,
to embed the wedge in it and check different exposure settings like
center weighted or matrix or generally measuring exposure against the grey
and then compensating to get maximum range.

Lots of Greetings!
Volker
 >> Stay informed about: Measurung dynamic range... 

Volker Hetzer wrote:
> Roger N. Clark (change username to rnclark) wrote:
>
>> A good way to make a light trap is a cone painted with
>> glossy black paint. The reflections from the painted surface
>> go deeper into the cone.
>
> So the camera looks into the pointy end?
> Otherwise it seems to me that the light gets reflected back.

Yes. Think of an ice-cream cone. Look at the inside,
and paint it glossy black. A machined metal (like aluminum)
works very well.

Roger
 >> Stay informed about: Measurung dynamic range... 

On Mon, 07 Aug 2006 12:27:11 -0700, "Roger N. Clark (change username
to rnclark)" <username.DeleteThis@qwest.net> wrote:

>Volker Hetzer wrote:
>> Roger N. Clark (change username to rnclark) wrote:
>>
>>> A good way to make a light trap is a cone painted with
>>> glossy black paint. The reflections from the painted surface
>>> go deeper into the cone.
>>
>> So the camera looks into the pointy end?
>> Otherwise it seems to me that the light gets reflected back.
>
>Yes. Think of an ice-cream cone. Look at the inside,
>and paint it glossy black. A machined metal (like aluminum)
>works very well.
>
>Roger

Wouldn't machined (or even moreso, polished) metal reflect much more
light?
Why did you pick glossy black?
--
Bill Funk
replace "g" with "a"
 >> Stay informed about: Measurung dynamic range... 

Bill Funk wrote:
> On Mon, 07 Aug 2006 12:27:11 -0700, "Roger N. Clark (change username
> to rnclark)" <username.TakeThisOut@qwest.net> wrote:
>
>
>>Volker Hetzer wrote:
>>
>>>Roger N. Clark (change username to rnclark) wrote:
>>>
>>>
>>>>A good way to make a light trap is a cone painted with
>>>>glossy black paint. The reflections from the painted surface
>>>>go deeper into the cone.
>>>
>>>So the camera looks into the pointy end?
>>>Otherwise it seems to me that the light gets reflected back.
>>
>>Yes. Think of an ice-cream cone. Look at the inside,
>>and paint it glossy black. A machined metal (like aluminum)
>>works very well.
>>
>>Roger
>
>
> Wouldn't machined (or even moreso, polished) metal reflect much more
> light?
> Why did you pick glossy black?

You need something to absorb the light. Each reflection
absorbs ~95% of the light, and the remaining gets reflected
further into the cone. If you use flat black, each reflection
scatters some light out of the cone. A polished aluminum
surface would reflect most light so most light would never be absorbed,
and eventually, after reaching the bottom, would reflect back out.
Flat black gives you both controlled absorption and controlled
reflection.

Roger
 >> Stay informed about: Measurung dynamic range... 

On Mon, 07 Aug 2006 16:12:21 -0700, "Roger N. Clark (change username
to rnclark)" <username.RemoveThis@qwest.net> wrote:

>Bill Funk wrote:
>> On Mon, 07 Aug 2006 12:27:11 -0700, "Roger N. Clark (change username
>> to rnclark)" <username.RemoveThis@qwest.net> wrote:
>>
>>
>>>Volker Hetzer wrote:
>>>
>>>>Roger N. Clark (change username to rnclark) wrote:
>>>>
>>>>
>>>>>A good way to make a light trap is a cone painted with
>>>>>glossy black paint. The reflections from the painted surface
>>>>>go deeper into the cone.
>>>>
>>>>So the camera looks into the pointy end?
>>>>Otherwise it seems to me that the light gets reflected back.
>>>
>>>Yes. Think of an ice-cream cone. Look at the inside,
>>>and paint it glossy black. A machined metal (like aluminum)
>>>works very well.
>>>
>>>Roger
>>
>>
>> Wouldn't machined (or even moreso, polished) metal reflect much more
>> light?
>> Why did you pick glossy black?
>
>You need something to absorb the light. Each reflection
>absorbs ~95% of the light, and the remaining gets reflected
>further into the cone. If you use flat black, each reflection
>scatters some light out of the cone. A polished aluminum
>surface would reflect most light so most light would never be absorbed,
>and eventually, after reaching the bottom, would reflect back out.
>Flat black gives you both controlled absorption and controlled
>reflection.
>
>Roger

If you're really after a light trap, wouldn't a baffle, where any
reflections send the light to something that absorbs light be better
than any reflective surfaces?
--
Bill Funk
replace "g" with "a"
 >> Stay informed about: Measurung dynamic range... 

Bill Funk wrote:
> On Mon, 07 Aug 2006 16:12:21 -0700, "Roger N. Clark (change username
> to rnclark)" <username DeleteThis @qwest.net> wrote:
>
>
>>Bill Funk wrote:
>>
>>>On Mon, 07 Aug 2006 12:27:11 -0700, "Roger N. Clark (change username
>>>to rnclark)" <username DeleteThis @qwest.net> wrote:
>>>
>>>
>>>
>>>>Volker Hetzer wrote:
>>>>
>>>>
>>>>>Roger N. Clark (change username to rnclark) wrote:
>>>>>
>>>>>
>>>>>
>>>>>>A good way to make a light trap is a cone painted with
>>>>>>glossy black paint. The reflections from the painted surface
>>>>>>go deeper into the cone.
>>>>>
>>>>>So the camera looks into the pointy end?
>>>>>Otherwise it seems to me that the light gets reflected back.
>>>>
>>>>Yes. Think of an ice-cream cone. Look at the inside,
>>>>and paint it glossy black. A machined metal (like aluminum)
>>>>works very well.
>>>>
>>>>Roger
>>>
>>>
>>>Wouldn't machined (or even moreso, polished) metal reflect much more
>>>light?
>>>Why did you pick glossy black?
>>
>>You need something to absorb the light. Each reflection
>>absorbs ~95% of the light, and the remaining gets reflected
>>further into the cone. If you use flat black, each reflection
>>scatters some light out of the cone. A polished aluminum
>>surface would reflect most light so most light would never be absorbed,
>>and eventually, after reaching the bottom, would reflect back out.
>>Flat black gives you both controlled absorption and controlled
>>reflection.
>>
>>Roger
>
>
> If you're really after a light trap, wouldn't a baffle, where any
> reflections send the light to something that absorbs light be better
> than any reflective surfaces?

Yes, that is what you want, but there is no surface that
absorbs all light. So you must make a trap that any
light that does get reflected does so in a direction
away from your view. The is what the cone accomplishes.
The angle of the cone must be steep enough for any
reflected light to get reflected further into the cone.
If, for example, light was reflected 5 times, and the
reflectance was 0.02, then the light reaching the bottom
would be 0.02*0.02*0.02*0.02*0.02 times the original
light level. The more specular the reflection of the
glossy paint, the less light will be scattered back
to you. In practice, it works very well.

Roger
 >> Stay informed about: Measurung dynamic range... 

"Roger N. Clark (change username to rnclark)" <username.DeleteThis@qwest.net> writes:

>Yes, that is what you want, but there is no surface that
>absorbs all light. So you must make a trap that any
>light that does get reflected does so in a direction
>away from your view. The is what the cone accomplishes.
>The angle of the cone must be steep enough for any
>reflected light to get reflected further into the cone.
>If, for example, light was reflected 5 times, and the
>reflectance was 0.02, then the light reaching the bottom
>would be 0.02*0.02*0.02*0.02*0.02 times the original
>light level. The more specular the reflection of the
>glossy paint, the less light will be scattered back
>to you. In practice, it works very well.

If the sides are steep enough and clean enough that the reflection
always goes further towards the bottom of the trap, it doesn't matter
how reflective the walls are. No black paint is needed.

For a demonstration, buy a pack of double-edged razor blades. Look at
the cutting edges, and note that they have shiny ground and polished
metal just behind the cutting edge. Now get the blades very clean,
wiping off any excess oil that they were packaged in, and stack them in
a pile. Carefully line up all the blades in the stack and clamp them
in that position with a C clamp, a nut and bolt, or whatever you want.

Now look at one of the two edges of the stack of blades that is made up
of cutting edges. It will be very, very black. The cutting edges
stacked beside each other form very deep narrow "V" shapes, and
virtually all light that strikes these edges of the stack falls into one
of these V slots (since the cutting edges themselves are so thin that
they have near-zero area). And even though the reflectivity of the
metal is high, probably 80 or 90%, each time the light reflects some of
the energy is absorbed and the rest is sent deeper into the V, never to
return back into the world.

One of my professors had assembled a stack like this, and liked to show
it to people. Apparently a V-groove absorber like this is good for
absorbing laser beams that have substantial power, since the device is
all metal and the dumped energy just heats the metal. You don't want to
get it hot enough to warp the metal, but it can handle more power
without damage than black paint would.

Dave
 >> Stay informed about: Measurung dynamic range... 

Since we've got a good discussion going on light traps, I'll try to
describe the one I am working on based on a pringles can. I clean out
the can, and save the top to make a baffle. I cut off the bottom (but
save it to put back on later). I do this to make inserting the lining
and cone easier. I am using a sheet of stickybacked artificial black
felt from a craft store (Michaels in my case). I first line the inside
of the can with a rectangle of the felt. Then I cut a cone out of the
remainder of felt and glue it to the bottom of the can. I then glue and
tape the bottom back on the can. I paint the underside of the top of
the can flat black. I glue a piece of my whitest card stock to the top
surface of the top, with rubber cement. I then cut a one inch square
in the top, and blacken the edge of the cutout with a black magic
marker, and replace it on top of can. The white card gives me a white
reference.

The first one I made was okay, but the workmanship was lacking, from
too many cut and try operations. I will be photographing the building
of the second one, with thoughts of sending the manuscript to one of
the photo magazines.
 >> Stay informed about: Measurung dynamic range...