ISO and actual sensitivity in DSLR's (D70, *istD, 20D, S3...)

Alan Browne wrote in news:d23q03$r5d$1
@inews.gazeta.pl:

Sensor. They go through a blow by blow description of the test on p
169. Basically, they're looking for the sensor data converted to JPG to
be 118/255 (R,G,B) to indicate successful grey card measurement. They
do not delve into the extremes of the curve.

OK - then sorry - they are wrong.
ISO sensitivity is not defined by 118 in the converted JPEG.


/Roland

Roland Karlsson wrote:

Alan Browne wrote in news:d23q03$r5d$1
@inews.gazeta.pl:


Sensor. They go through a blow by blow description of the test on p
169. Basically, they're looking for the sensor data converted to JPG to
be 118/255 (R,G,B) to indicate successful grey card measurement. They
do not delve into the extremes of the curve.


OK - then sorry - they are wrong.
ISO sensitivity is not defined by 118 in the converted JPEG.

Go buy the magazine where, as I state above, the entire process is
detailed, 'blow by blow' including references to the standards, and
which I'm not going to type for you here.

And yes, mid tone is, as previously discussed, R=118=G=B.

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: there's no such thing as a FreeLunch.

Roland Karlsson wrote:

Alan Browne wrote in news:d23q03$r5d$1
@inews.gazeta.pl:


Sensor. They go through a blow by blow description of the test on p
169. Basically, they're looking for the sensor data converted to JPG to
be 118/255 (R,G,B) to indicate successful grey card measurement. They
do not delve into the extremes of the curve.


OK - then sorry - they are wrong.
ISO sensitivity is not defined by 118 in the converted JPEG.


I just repeated the test as follows:

0. Shot grey card at f/5.6 using strobe set at f/5.6 incident metered
(for ISO 100) with K-M 7D.

1. RAW import of the same shot.

2. In RAW import (Adobe) set all sliders to 0; temp to 5500K.

3. Within the RAW import, 8-bit end of R,G,B = 119,117,105 (typical,
many values around there with the blue being consistently 10 or 12
levels below the R and G. (Whether this is a reflection npi of the
card or the sensors is not determined).

4. Imported that into E 3.0. R,G,B values are the same.

5. Converted to 8-bit and saved as JPG

6. Reopened. Values are as above.

Cheers,
Alan

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: there's no such thing as a FreeLunch.

Alan Browne wrote in news:d23t8b$8m1$1
@inews.gazeta.pl:

And yes, mid tone is, as previously discussed, R=118=G=B.

Hmmm ... how can that be? The sensitivity of a sensor
is defined by the saturation level.

The conversion to JPEG depends on color space and the
bias you use for representing 255. So - the mid tone
can (IMHO) be anything in the JPEG.

Moreover - the R, G and B are only equal if the system
is in perfct color balance. And that is highly unlikely.


/Roland

Thomas T. Veldhouse wrote:


If the D70 varies in sensitivity per individual camera then the D70
would be irrelavent. In fact, I am sure that they are consistant
between cameras and as such, it this is all useful information to have.
I own the D70 and will likely be working with film (Nikon N80) in the
not too distant future and it will be useful to know that my experience
with ISO settings on the D70 are likely to require a factor of 0.8 when
going to film.

....then there's the meters...

Cheers,
Alan



--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: there's no such thing as a FreeLunch.

Roland Karlsson wrote:

Alan Browne wrote in news:d23t8b$8m1$1
@inews.gazeta.pl:


And yes, mid tone is, as previously discussed, R=118=G=B.


Hmmm ... how can that be? The sensitivity of a sensor
is defined by the saturation level.

The conversion to JPEG depends on color space and the
bias you use for representing 255. So - the mid tone
can (IMHO) be anything in the JPEG.

Moreover - the R, G and B are only equal if the system
is in perfct color balance. And that is highly unlikely.

We've gone through this before. As the parameter being tested is
fidelity to the ISO setting, the mid tone is the only place the can be
measured. The toe and shoulder should vary too much between camera
models to be of any use in comparison.

As my own test shows, the three are not in perfect balance, and one
should not expect or hope that they would be. Further, and as
mentioned, and again I suggest you get a copy of the mag, the ISO table
they published and that I repeated are rounded numbers.

Cheers
Alan

Cheers,
Alan.

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: there's no such thing as a FreeLunch.

Alan Browne wrote in
:

I just repeated the test as follows:

Hmmm ... I assure you ... I am not trying to be difficult.
But ... I have a hard time to believe in this. I have added
some comments below.

0. Shot grey card at f/5.6 using strobe set at f/5.6 incident metered
(for ISO 100) with K-M 7D.

1. RAW import of the same shot.

2. In RAW import (Adobe) set all sliders to 0; temp to 5500K.

3. Within the RAW import, 8-bit end of R,G,B = 119,117,105 (typical,
many values around there with the blue being consistently 10 or 12
levels below the R and G. (Whether this is a reflection npi of the
card or the sensors is not determined).

I am sorry - I don't understand this at all.

4. Imported that into E 3.0. R,G,B values are the same.

Nor do I understand this. This might be because I have
no knowledge about Adobe RAW import. It looks like step
3 and 4 makes some modifications of the RAW data upon import.
How can we use the data for meassurements then? The data shold
be unaltered IMHO.

5. Converted to 8-bit and saved as JPG

Converting to 8 bit can be made in lots of ways. gamma?
color space? bias? This is another step where the data
is modified and therefore makes meassuremets invalid IMHO.

6. Reopened. Values are as above.

To mee it looks like the correct method is to compare
the level of white clipping and then use the definition
for ISO sensitivity for sensors.

How the image looks after converting with Adobe RAW to
JPEG seems totally bogus to me. If you don't want to
test the workflow described above of course. But most
serious photographers do not convert to JPEG.


Cheers,
Alan

Roland Karlsson wrote in
:

To mee it looks like the correct method is to compare
the level of white clipping and then use the definition
for ISO sensitivity for sensors.

Hmmm .. after some thought I withdraw this suggestion.
The clipping level might be hard to find.

The problem is even more complex really. There are two
different things you can meassure.

1. Given a standard luminosity of the picture - what
exposure time and aperture do the camera choose.

2. Given a standard luminosity of the picture - what
is the exposure on the CCD with a given exposure
time and aperture.

Case 1 meassures the light meassuring bias in the camera.

Case 2 meassures the light sensitivity of the system.

Both are two interesting results. But they are different.


/Roland

Roland Karlsson wrote:

Alan Browne wrote in
:


I just repeated the test as follows:


Hmmm ... I assure you ... I am not trying to be difficult.

Hmmm yourself.

But ... I have a hard time to believe in this. I have added
some comments below.


0. Shot grey card at f/5.6 using strobe set at f/5.6 incident metered
(for ISO 100) with K-M 7D.

1. RAW import of the same shot.

2. In RAW import (Adobe) set all sliders to 0; temp to 5500K.

3. Within the RAW import, 8-bit end of R,G,B = 119,117,105 (typical,
many values around there with the blue being consistently 10 or 12
levels below the R and G. (Whether this is a reflection npi of the
card or the sensors is not determined).


I am sorry - I don't understand this at all.

Grey := Red = Blue = Green. When high values, pale grey, when low
values, dark grey. 18% grey is 118/255.
(A result of a coefficient from the ISO docs that is 0.461 of the
maximum level. 0.461 X 255 = 117.56)

The math is in the article so get down to your local newsstand and buy
the edition.



4. Imported that into E 3.0. R,G,B values are the same.


Nor do I understand this. This might be because I have
no knowledge about Adobe RAW import. It looks like step
3 and 4 makes some modifications of the RAW data upon import.
How can we use the data for meassurements then? The data shold
be unaltered IMHO.

It is correct as long as all conversions are set to 0. (As it was).
The only variable is the color temp of the light source which I set to
that of the flash.

By the way, this would be the same for film. Wrong color temp for the
film would render results that are not useful.

5. Converted to 8-bit and saved as JPG


Converting to 8 bit can be made in lots of ways. gamma?
color space? bias? This is another step where the data
is modified and therefore makes meassuremets invalid IMHO.

Gamma goes to dislay/print, not content. Using 'info' in PS to examine
the content of a pixel is independant of gamma.



6. Reopened. Values are as above.


To mee it looks like the correct method is to compare
the level of white clipping and then use the definition
for ISO sensitivity for sensors.

How the image looks after converting with Adobe RAW to
JPEG seems totally bogus to me. If you don't want to
test the workflow described above of course. But most
serious photographers do not convert to JPEG.

Regardless of the conversions:

-the grey level was maintained exactly if the RAW conversion was held to
no changes other than color temperature. eg: the 8 bit values in the
RAW and JPG were the same.

-you can't make such a comparison at the shoulder or toe of a variety of
sensors. You would get incomparable data.


--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: there's no such thing as a FreeLunch.

Bubbabob wrote:



I've been wondering if checking the histogram is considered 'chimping'.
Is it OK if I just forgo the 'Ooh, ooh" part?

Yes. But those who know, know you're "ooh, oohing" mutely.

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: there's no such thing as a FreeLunch.