10D ISO ratings a lie?

<JPS.RemoveThis@no.komm> wrote in message
news:q6qng05shdolr5ncs27p3qpentd8m28sdm@4ax.com...
 > Has anyone tried using "Sunny f16" in manual mode with a 10D on a sunny
 > day?
 >
 > When I do, the pictures are under-exposed by over a stop. When I check
 > exposure of a flat, grey surface or blue sky in the camera, and compare
 > it to my Sekonic meter, there is about 1.3 stops difference.
 >
 > This casts a shadow on the integrity of the camera's "low noise at high
 > ISO", when ISO 800 is really ISO 320, no?

How are you making the comparison?
What are you "metering"?

If you're using sunny 16, then you shouldn't be metering at all.
Are you metering from a grey card?
If so, there are a number of factors that can throw this off somewhat.

What flat, "grey" surface are you referring to?
18% grey card?
What?<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

<JPS.DeleteThis@no.komm> wrote in message
news:q6qng05shdolr5ncs27p3qpentd8m28sdm@4ax.com...
 > Has anyone tried using "Sunny f16" in manual mode with a 10D on a sunny
 > day?
 >
 > When I do, the pictures are under-exposed by over a stop. When I check
 > exposure of a flat, grey surface or blue sky in the camera, and compare
 > it to my Sekonic meter, there is about 1.3 stops difference.
 >
 > This casts a shadow on the integrity of the camera's "low noise at high
 > ISO", when ISO 800 is really ISO 320, no?

I just went outside and tried mine at 1:00PM under bright sunlight, and
little to no haze.
It did indeed turn into a "Sunny 11" before it exposed properly.
Clearly under-exposed on my 10D.

The under-exposure was shot at 1/90th, f16, ISO 100.

My conclusion is based upon photoshop histograms, camera histograms, and
visual inspection on my properly calibrated monitor.

It does seem to be significantly under-sensitive in this case.<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

In message <i4SOc.3316$go.2741@fed1read07>,
"Skip M" <shadowcatcher.TakeThisOut@cox.net> wrote:

 >Hmm, you may have a point about "low noise." But our 10D is about 1/3 stop
 >over by its own meter, as it is, and about 1 1/3 over on flash, metered by
 >our Sekonic. I'll check the daylight readings when it gets home. The
 >combination of ours and yours sounds like a consistency problem, at least...

The "problem" I refer to should not be affected by the camera's metering
at all. There is no metering involved at all in shooting in manual mode
with "sunny f16" settings, and no relation to your auto-exposure or
metering anomalies. Taking the camera's stated ISO literally, there is
about 5 stops of highlight headroom in the red channel RAW data, and 4.5
stops in the green and blue channels.

My camera's metering is pretty good (as far as filling the histogram is
concerned), even autoflash (unless I'm close to the subject and using a
high ISO and better beamer, in which case it may wash out). But if what
I say is universal, then the camera is actually working at ISO 320 or so
when it says 800.

--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS.TakeThisOut@no.komm>
  ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

On 2004-07-31, JPS.RemoveThis@no.komm <JPS.RemoveThis@no.komm> wrote:

 > Has anyone tried using "Sunny f16" in manual mode with a 10D on a sunny
 > day?

Well if you look at
<a style='text-decoration: underline;' href="http://www.dpreview.com/reviews/canoneos10d/page20.asp," target="_blank">http://www.dpreview.com/reviews/canoneos10d/page20.asp,</a> You'll see
that Phil's exposure here is bascially Sunny F16 (Well, F11 @ 1/200)
Do you think that these images are underexposed by a stop? Could be
slightly underexposed, but the day looks a bit hazy to me (the shadows
are less distinct than I like for "Sunny")

Actually, the conventional wisdom is the opposite: that the Canon
10D and 300D are actually a little more sensitive than the ISO
indicates at least compared to other DSLRs:

1/2 stop over the D70 here at all ISOs:
<a style='text-decoration: underline;' href="http://www.dpreview.com/reviews/nikond70/page14.asp" target="_blank">http://www.dpreview.com/reviews/nikond70/page14.asp</a>

1/3rd when Sunny and outdoors
<a style='text-decoration: underline;' href="http://www.dpreview.com/reviews/nikond70/page21.asp" target="_blank">http://www.dpreview.com/reviews/nikond70/page21.asp</a>

and 2/3 stop 10D over the SD10 here:
<a style='text-decoration: underline;' href="http://www.dpreview.com/reviews/sigmasd10/page13.asp" target="_blank">http://www.dpreview.com/reviews/sigmasd10/page13.asp</a>

 > This casts a shadow on the integrity of the camera's "low noise at high
 > ISO", when ISO 800 is really ISO 320, no?

Not until you eliminate other sources of error. It could be your
camera. It could be your lens. You also don't say how the images are
processed or how you are actually measuring the camera exposure. You
should not draw a sweeping conclusion from one sample.

--
Erik<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

Well, sunny 16 is just a guideline, I've never really tried it with digital,
and with film, only as a starting point for bracketing. As with any rule of
thumb, there are exceptions. I thought, misreading your post slightly, that
you were referring go metered exposures, not the rule of thumb...nevermind.

--
Skip Middleton
<a style='text-decoration: underline;' href="http://www.shadowcatcherimagery.com" target="_blank">http://www.shadowcatcherimagery.com</a>
<JPS.DeleteThis@no.komm> wrote in message
news:j2tng0togqh3fnan98j6klksvfvm4rf3lo@4ax.com...
 > In message <i4SOc.3316$go.2741@fed1read07>,
 > "Skip M" <shadowcatcher.DeleteThis@cox.net> wrote:
 >
  > >Hmm, you may have a point about "low noise." But our 10D is about 1/3
stop
  > >over by its own meter, as it is, and about 1 1/3 over on flash, metered
by
  > >our Sekonic. I'll check the daylight readings when it gets home. The
  > >combination of ours and yours sounds like a consistency problem, at
least...
 >
 > The "problem" I refer to should not be affected by the camera's metering
 > at all. There is no metering involved at all in shooting in manual mode
 > with "sunny f16" settings, and no relation to your auto-exposure or
 > metering anomalies. Taking the camera's stated ISO literally, there is
 > about 5 stops of highlight headroom in the red channel RAW data, and 4.5
 > stops in the green and blue channels.
 >
 > My camera's metering is pretty good (as far as filling the histogram is
 > concerned), even autoflash (unless I'm close to the subject and using a
 > high ISO and better beamer, in which case it may wash out). But if what
 > I say is universal, then the camera is actually working at ISO 320 or so
 > when it says 800.
 >
 > --
 >
 > <>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
 > John P Sheehy <JPS.DeleteThis@no.komm>
  > ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

In message <R4TOc.16942$DZ.2025698@twister.tampabay.rr.com>,
"E. Magnuson" <edjpgcom DeleteThis @yahoo.com> wrote:

 >On 2004-07-31, JPS DeleteThis @no.komm <JPS DeleteThis @no.komm> wrote:
 >
  >> Has anyone tried using "Sunny f16" in manual mode with a 10D on a sunny
  >> day?
 >
 >Well if you look at
 >http://www.dpreview.com/reviews/canoneos10d/page20.asp, You'll see
 >that Phil's exposure here is bascially Sunny F16 (Well, F11 @ 1/200)
 >Do you think that these images are underexposed by a stop? Could be
 >slightly underexposed, but the day looks a bit hazy to me (the shadows
 >are less distinct than I like for "Sunny")

To tell you the truth, I really don't have much confidence in Phil
Askey's precision. He may have pushed this in software, or he may have
wrote the wrong values in the article.

 >Actually, the conventional wisdom is the opposite: that the Canon
 >10D and 300D are actually a little more sensitive than the ISO
 >indicates at least compared to other DSLRs:

 >1/2 stop over the D70 here at all ISOs:
 >http://www.dpreview.com/reviews/nikond70/page14.asp

I'm not sure you understand my discovery. What I'm saying is that the
ISO values that the camera claims it is using may actually be
exaggerated by over 2x.

 >1/3rd when Sunny and outdoors
 >http://www.dpreview.com/reviews/nikond70/page21.asp
 >
 >and 2/3 stop 10D over the SD10 here:
 >http://www.dpreview.com/reviews/sigmasd10/page13.asp
 >
  >> This casts a shadow on the integrity of the camera's "low noise at high
  >> ISO", when ISO 800 is really ISO 320, no?
 >
 >Not until you eliminate other sources of error. It could be your
 >camera. It could be your lens. You also don't say how the images are
 >processed or how you are actually measuring the camera exposure. You
 >should not draw a sweeping conclusion from one sample.

I am not drawing from one sample. I have suspected this for a long
time, and I finally tested it today. The sun was high and unobstructed
by clouds; I set the camera to ISO 100, manual mode, f16, and 1/100. I
took a picture from my window of the parking lot below, where the trees
and the cars were in direct sunlight, and the only thing in the top half
of the histogram was a bright white car. I tried again later with two
other lenses, and they all gave the same exposure for the same subject
areas. I took one shot with each of three lenses of the same scene, and
only bright white objects were in the top half of the histogram, and the
images were equally dark. As I said in the original post, I checked my
Sekonic L-558 meter, and the camera exposed at 1/60 (f16, Av mode) where
the Sekonic said 1/125. After I did this I remember that images I took
with the Sekonic and with sunny f16 in the past were always dark.

The camera's firmware *definitely* knows that the camera is actually a
lower ISO than it says, because auto-exposure never gives me these dark
images in sunlight.
--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS DeleteThis @no.komm>
  ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

In message <f9TOc.12966$8G6.1906@fed1read04>,
"Mark M" <mjmorgan2.RemoveThis@goaway.cox.unless.no.crap.net> wrote:

 ><JPS.RemoveThis@no.komm> wrote in message
 >news:q6qng05shdolr5ncs27p3qpentd8m28sdm@4ax.com...
  >> Has anyone tried using "Sunny f16" in manual mode with a 10D on a sunny
  >> day?

  >> When I do, the pictures are under-exposed by over a stop. When I check
  >> exposure of a flat, grey surface or blue sky in the camera, and compare
  >> it to my Sekonic meter, there is about 1.3 stops difference.

  >> This casts a shadow on the integrity of the camera's "low noise at high
  >> ISO", when ISO 800 is really ISO 320, no?

 >How are you making the comparison?

There is no "the" comparison.
I mentioned more than one comparison.

In one, I'm comparing sunny f16 exposures to automatic exposure. Sunny
f16 images taken in direct sunlight (with distinct shadows) are much
darker than auto-exposed images. A clean white car in direct sunlight
is inside the histogram with about a stop to spare. The rest of the
image is in the left half of the histogram.


 >What are you "metering"?

In another comparison, in Av mode, the 10D wants over double the
exposure that my Sekonic meter suggests is necessary, when pointed at
pure blue sky, or a blank white wall.

 >If you're using sunny 16, then you shouldn't be metering at all.

I wasn't metering the sunny f16 images. I took shots with various
lenses, at ISO 100, 1/100 and f16 of directly sunlit scenes.

 >Are you metering from a grey card?

Not this time, but I do remember in the past getting a dark grey card
after metering it with the Sekonic.

 >If so, there are a number of factors that can throw this off somewhat.

No relevance here. Doesn't matter if it is black, white, grey, glossy,
or matte. I'm comparing a Sekonic exposure meter to the one in the
camera, with the same flat surface with narrow histogram.

 >What flat, "grey" surface are you referring to?
 >18% grey card?
 >What?

Blank white wall. Doesn't really matter, as long as it is illuminated
evenly. I was comparing the camera's metering to the Sekonic's to
*verify* that my sunny f16 episode was not a fluke.

--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS.RemoveThis@no.komm>
  ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

In message <mnTOc.12968$8G6.3269@fed1read04>,
"Mark M" <mjmorgan2 RemoveThis @goaway.cox.unless.no.crap.net> wrote:

 >I just went outside and tried mine at 1:00PM under bright sunlight, and
 >little to no haze.
 >It did indeed turn into a "Sunny 11" before it exposed properly.
 >Clearly under-exposed on my 10D.
 >
 >The under-exposure was shot at 1/90th, f16, ISO 100.
 >
 >My conclusion is based upon photoshop histograms, camera histograms, and
 >visual inspection on my properly calibrated monitor.
 >
 >It does seem to be significantly under-sensitive in this case.

.... but autoexposure corrects for it, so Canon knows that the stated ISO
is not the actual one.

I'm willing to bet that most digitals are doing this to some degree, and
no company is reporting real ISO, to keep up with the other
manufacturers' ISO-fudging. It will be interesting if people with other
cameras would try this as well.

If everyone is doing it, then the comparisons between digitals still
stand, to some degree, but the comparisons to film are exaggerated in
favor of the digitals.
--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS RemoveThis @no.komm>
  ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

In article <q6qng05shdolr5ncs27p3qpentd8m28sdm.TakeThisOut@4ax.com>, <JPS.TakeThisOut@no.komm> wrote:
 >Has anyone tried using "Sunny f16" in manual mode with a 10D on a sunny
 >day?
 >
 >When I do, the pictures are under-exposed by over a stop. When I check
 >exposure of a flat, grey surface or blue sky in the camera, and compare
 >it to my Sekonic meter, there is about 1.3 stops difference.

Interesting. I just went back and had a look at some of my holiday shots
taken in Zion national park, Utah, from last summer. Many of those were
taken in bright sunlight in the early afternoon. Looking at the histograms,
things like white T-shirts and white cloud-detail (not shot into the Sun,
obviously) do seem to fall very close to the edge of the histogram. For
example, the one I have in front of me (100 ISO, 1/750, f/5.6, so would be
1/90 at f/16, close enough for government work) is *just* showing
me the blowout warning on the white highlights in Photoshop CS's raw
convertor. Qualatatively, the picture does look very slightly over-exposed
as well. If I pull it back half a stop, it looks about right, and the
exposure-clipping almost goes away as well (just a few pixels left).

So I'd say that's about right - certainly nothing like the stop-or-more out
that you seem to be seeing.<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

"Chris Brown" <cpbrown.TakeThisOut@ntlworld.no_uce_please.com> wrote in message
news:ooltt1-psb.ln1@narcissus.dyndns.org...
 > In article <q6qng05shdolr5ncs27p3qpentd8m28sdm.TakeThisOut@4ax.com>, <JPS.TakeThisOut@no.komm>
wrote:
  > >Has anyone tried using "Sunny f16" in manual mode with a 10D on a sunny
  > >day?
  > >
  > >When I do, the pictures are under-exposed by over a stop. When I check
  > >exposure of a flat, grey surface or blue sky in the camera, and compare
  > >it to my Sekonic meter, there is about 1.3 stops difference.
 >
 > Interesting. I just went back and had a look at some of my holiday shots
 > taken in Zion national park, Utah, from last summer. Many of those were
 > taken in bright sunlight in the early afternoon. Looking at the
histograms,
 > things like white T-shirts and white cloud-detail (not shot into the Sun,
 > obviously) do seem to fall very close to the edge of the histogram. For
 > example, the one I have in front of me (100 ISO, 1/750, f/5.6, so would be
 > 1/90 at f/16, close enough for government work) is *just* showing
 > me the blowout warning on the white highlights in Photoshop CS's raw
 > convertor. Qualatatively, the picture does look very slightly over-exposed
 > as well. If I pull it back half a stop, it looks about right, and the
 > exposure-clipping almost goes away as well (just a few pixels left).
 >
 > So I'd say that's about right - certainly nothing like the stop-or-more
out
 > that you seem to be seeing.

Perhaps Canon decided to take into consideration the aspct of digital that
leads to the need of exposing for highlights, since these cannot be
recovered after blow-out (at least not much...though RAW files can be pulled
back a bit). Still, it seems that photographers should be able to rely on
ISO and sunny 16 principles.<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

In message <ooltt1-psb.ln1.DeleteThis@narcissus.dyndns.org>,
Chris Brown <cpbrown.DeleteThis@ntlworld.no_uce_please.com> wrote:

 >In article <q6qng05shdolr5ncs27p3qpentd8m28sdm.DeleteThis@4ax.com>, <JPS.DeleteThis@no.komm> wrote:
  >>Has anyone tried using "Sunny f16" in manual mode with a 10D on a sunny
  >>day?
  >>
  >>When I do, the pictures are under-exposed by over a stop. When I check
  >>exposure of a flat, grey surface or blue sky in the camera, and compare
  >>it to my Sekonic meter, there is about 1.3 stops difference.
 >
 >Interesting. I just went back and had a look at some of my holiday shots
 >taken in Zion national park, Utah, from last summer. Many of those were
 >taken in bright sunlight in the early afternoon. Looking at the histograms,
 >things like white T-shirts and white cloud-detail (not shot into the Sun,
 >obviously) do seem to fall very close to the edge of the histogram. For
 >example, the one I have in front of me (100 ISO, 1/750, f/5.6, so would be
 >1/90 at f/16, close enough for government work) is *just* showing
 >me the blowout warning on the white highlights in Photoshop CS's raw
 >convertor. Qualatatively, the picture does look very slightly over-exposed
 >as well. If I pull it back half a stop, it looks about right, and the
 >exposure-clipping almost goes away as well (just a few pixels left).
 >
 >So I'd say that's about right - certainly nothing like the stop-or-more out
 >that you seem to be seeing.

Very interesting. I wonder if the amplifier gain or sensor sensitivity
doesn't vary from camera to camera, and they tune the auto-exposure for
the camera? Your 10D could start at ISO 100, and mine might say that
it's ISO 100, but is really ISO 40. That could explain why I get so
much less noise than the camera is supposed to have. The other night, I
took a picture of a young buck at "ISO 3200" in Tv mode @ 1/100 (to
hand-hold the 400mm IS lens), and I had to boost the "exposure" in
CaptureOne by 2.5 stops to get good tonal range, making it effectively
ISO 18,100 @ 8.5 bits. The image is a bit noisy, but clears up almost
completely when shrunk to 800*533 (the inset is original pixels). The
inset has only the 2.5 stop exposure boost from Capture One; the larger
image has about another half-stop boost in PS, so the large image as it
is presented is effectively about ISO 25,600, if the 3200 can be
believed.

<a style='text-decoration: underline;' href="http://www.pbase.com/image/32020085" target="_blank">http://www.pbase.com/image/32020085</a>

Does that look too good for ISO 25,600 @ 8bits linear RAW? Maybe it's
really only ISO 10,240 @ 8bits linear RAW).
--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS.DeleteThis@no.komm>
  ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

In message <QzWOc.13079$8G6.1437@fed1read04>,
"Mark M" <mjmorgan2.TakeThisOut@goaway.cox.unless.no.crap.net> wrote:

 >Perhaps Canon decided to take into consideration the aspct of digital that
 >leads to the need of exposing for highlights, since these cannot be
 >recovered after blow-out (at least not much...though RAW files can be pulled
 >back a bit). Still, it seems that photographers should be able to rely on
 >ISO and sunny 16 principles.

Exposure is independent of actual ISO; even with this over-rated ISO in
my 10D, JPEG will still blow out without exposure compensation in
auto-exposure modes. Canon is not using this under-exposure for
anything; they don't even intend for you to go there.
--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS.TakeThisOut@no.komm>
  ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie? 

 >From: JPS.TakeThisOut@no.komm

 >The other night, I
 >took a picture of a young buck at "ISO 3200" in Tv mode @ 1/100 (to
 >hand-hold the 400mm IS lens), and I had to boost the "exposure" in
 >CaptureOne by 2.5 stops to get good tonal range, making it effectively
 >ISO 18,100 @ 8.5 bits. The image is a bit noisy, but clears up almost
 >completely when shrunk to 800*533 (the inset is original pixels). The
 >inset has only the 2.5 stop exposure boost from Capture One; the larger
 >image has about another half-stop boost in PS, so the large image as it
 >is presented is effectively about ISO 25,600, if the 3200 can be
 >believed.
 >
 >http://www.pbase.com/image/32020085
 >
 >Does that look too good for ISO 25,600 @ 8bits linear RAW? Maybe it's
 >really only ISO 10,240 @ 8bits linear RAW).
 >--

Looks to me like the pic is over-exposed. I'd like to see it without the +2.5
boost from C1.

Has your monitor been calibrated?
If not, this might explain why your images appear off.<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: 10D ISO ratings a lie?