An ISO Quality question

In article ,
David J Taylor wrote:
Philip Homburg wrote:
Which means that quantisation noise corresponds to 0.5 * 40e3 / 2**12
= 4.88
electrons. At 25 electrons or more, shot-noise exceeds quantisation
noise.

Yes, in the lighter areas of the picture 12 bits is enough, agreed. But
where you see the quantisation noise is in the dark shadow regions.
Remember that the eye's response isn't linear, but more like log.

Except that according to this calculation, quantisation noise exceeds
photon shot noise only at output levels 0, 1, and 2. At level 3 you have
3 * 40e3/4095 = 29.3 electrons, which gives a photon shot noise of 5.41
electrons. Which is more than 0.5 LSB.


--
That was it. Done. The faulty Monk was turned out into the desert where it
could believe what it liked, including the idea that it had been hard done
by. It was allowed to keep its horse, since horses were so cheap to make.
-- Douglas Adams in Dirk Gently's Holistic Detective Agency

Philip Homburg wrote:
In article ,
David J Taylor

wrote:
Philip Homburg wrote:
Which means that quantisation noise corresponds to 0.5 * 40e3 /
2**12 = 4.88
electrons. At 25 electrons or more, shot-noise exceeds quantisation
noise.

Yes, in the lighter areas of the picture 12 bits is enough, agreed.
But where you see the quantisation noise is in the dark shadow
regions. Remember that the eye's response isn't linear, but more
like log.

Except that according to this calculation, quantisation noise exceeds
photon shot noise only at output levels 0, 1, and 2. At level 3 you
have 3 * 40e3/4095 = 29.3 electrons, which gives a photon shot noise
of 5.41
electrons. Which is more than 0.5 LSB.

Yes, but level 3 corresponds to the sensor output of 87, so there is quite
a lot lost. People have measured this (but not me).

David

DoN. Nichols wrote:

Not a consideration on the Cannon, but my Nikon D70 has a
minimum ISO of 200, and this prevents me from shooting at the highest
magnification range of the old 200mm Medical Nikkor (with the built-in
ring flash). It does the exposure calculation based on an ISO (actually
marked as ASA) setting ring, and a magnification ratio ring, resulting
in setting of the aperture ring.

The lens is capable of going down to a 3X magnification ratio,
but the closest that I can get at present is a 2/3X magnification ratio
because of the exposure situation.

I really need to modify the power pack to reduce the flash power
enough for those closer shots.


ND filter.

--
-- 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: Remove FreeLunch.

In article ,
Alan Browne wrote:
DoN. Nichols wrote:

Not a consideration on the Cannon, but my Nikon D70 has a
minimum ISO of 200, and this prevents me from shooting at the highest
magnification range of the old 200mm Medical Nikkor (with the built-in
ring flash). It does the exposure calculation based on an ISO (actually
marked as ASA) setting ring, and a magnification ratio ring, resulting
in setting of the aperture ring.

The lens is capable of going down to a 3X magnification ratio,
but the closest that I can get at present is a 2/3X magnification ratio
because of the exposure situation.

I really need to modify the power pack to reduce the flash power
enough for those closer shots.


ND filter.

As I already mentioned elsewhere in this thread, this lens is a
bit weird. Start off with a 200mm f5.6 lens, but mount it so it is
fixed focus, at a 1:15 ratio (10' 11.9" or 3.35 meters.

Now -- supply a set of close-up lenses which must screw into the
front of the lens -- defined as 1/8X, 1/6X, 1/4X, 1/2X, 1X & 2X.

To get the intermediate values, stacking specific CU lens pairs
is defined:

1/3X = 1/4 + 1/6

2/3X = 1/2 + 1/4

1.5X = 1 + 1/2

3 X = 2 + 1

The only place where I need the reduced light is in the range
from 1 to 3.

Note that the 2X CU lens has a rather pronounced front element
curvature, and thus also has no front female filter thread.

This means that the ND filter must be placed between the prime
lens and the first CU lens, or between the two CU lenses for the 2X or
3X settings. I'm not sure about stacking that many elements, and flare
would be lower with the ND filter as the frontmost element, which is not
an option when the 2X CU lens is in use.

Also -- the filter thread is a bit weird. IIRC, I measured it
at something like 37.6 mm, so I'm not at all sure that I could get such
a filter, except as a custom, or with a reducing ring, which would mean
that I would need the opposite converter ring to continue on to the 2X
element.

So -- overall, I think that I would get better shots with a
modification to the flash power supply.

If you want to see what this is like, here is one which just
closed on eBay:

Auction #: 7536353068

Full URL: http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7536353068

Thanks,
DoN.
--
Email: | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---

In article ,
DoN. Nichols wrote:
Not a consideration on the Cannon, but my Nikon D70 has a
minimum ISO of 200, and this prevents me from shooting at the highest
magnification range of the old 200mm Medical Nikkor (with the built-in
ring flash). It does the exposure calculation based on an ISO (actually
marked as ASA) setting ring, and a magnification ratio ring, resulting
in setting of the aperture ring.

What may work is to set the shutter speed sufficiently high that only part
of the flash output is captured . This works on the D1 (though I didn't try
it with a medical Nikkor), and I think that it is also supposed to work with
a D70.


--
That was it. Done. The faulty Monk was turned out into the desert where it
could believe what it liked, including the idea that it had been hard done
by. It was allowed to keep its horse, since horses were so cheap to make.
-- Douglas Adams in Dirk Gently's Holistic Detective Agency

In message ,
(DoN. Nichols) wrote:

In article , wrote:
In message ,
(DoN. Nichols) wrote:

In article , wrote:

[ ... Medical Nikkor with D70 & ISO 200 minimum snipped ... ]

Have you tried shooting with the manual flash for 100 when the camera
was set to 200?

I did while shooting jpegs, and there were blown highlights.

That might not be a problem at all, if you're shooting RAW. 100% matte
reflectance would probably not clip; only specular highlights, which
should be subdued by the ring, anyway (and which would generally even
out the lighting, in your favor).

Hmm ... it is worth a try. But I think that I would prefer
getting the right amount of light to start with.

I consider the ISO numbers on the camera to be pretty arbitrary. My
first DSLR, the Canon 10D, had a reputation for blowing out JPEGs unless
the contrast parameter was set to -2, or you used negative exposure
compensation. Upon further investigation, I found out that the camera
was actually metering for ISO 64, when it was set to ISO 100. I also
found out that I could use an external incident meter, or spot-meter on
a grey card, and expose for ISO 40, and no white that was purely matte
would blow out.

O.K. Note that the chart which comes with the lens suggest that
for 2:1 reproduction ratio, a maximum ASA (now ISO) of 50 is suggested
for color, and 25 for B&W. (This is an old lens system, of course.)
The chart is incomplete, but it looks like it will require an ASA of 25
(or perhaps 32) with the full 3:1 ratio, so I may have to cut the
illumination anyway. I can't see pushing it that far down.

I suggest that you try putting some magenta filters cut out from clear
magenta sheets of file folders or something. I use "Flomo" folders; two
sheets thick of their magenta folder over my Canon 550EX gives the flash
almost the exact color that is native to the camera's channel
sensitivities. The noise is less chromatic when you do this, and weaker
to boot, so the shadows are much quieter. A neutral density on the lens
will cut all three color channels equally, but ideally, for the D70, you
want to cut the green the most, and the red the least. If the lighting
is mixed with daylight or tungsten ambient, then a filter on the lens
may do the trick. I shoot in daylight with a Hoya FLD filter, which
doesn't get white balance perfectly native, but gets it closer. A
double-strength FLD would probably be perfect for the 20D, 10D, D70, or
any other DSLR with the same general color channel sensitivity. I
haven't experimented with the Tiffen magenta filters or the Cokins yet
(of course, the Cokins wouldn't work with your specialty lens).

If you want to see what's really going on, get a matte white target, and
take a picture at each ISO with it, at the magnifications in question,
in RAW, and then load the RAW files into IRIS which will show the exact
RAW values, and you can calculate how much headroom you have at a given
ISO. Do this both with and without a magenta filter over the flash.
--


John P Sheehy

In message coh.net,
(Philip Homburg) wrote:

In article ,
David J Taylor wrote:
Philip Homburg wrote:
Which means that quantisation noise corresponds to 0.5 * 40e3 / 2**12
= 4.88
electrons. At 25 electrons or more, shot-noise exceeds quantisation
noise.

Yes, in the lighter areas of the picture 12 bits is enough, agreed. But
where you see the quantisation noise is in the dark shadow regions.
Remember that the eye's response isn't linear, but more like log.

Except that according to this calculation, quantisation noise exceeds
photon shot noise only at output levels 0, 1, and 2. At level 3 you have
3 * 40e3/4095 = 29.3 electrons, which gives a photon shot noise of 5.41
electrons. Which is more than 0.5 LSB.

If you have noise that fills a bell curve 20 RAW levels wide, you can
still see images that add only 2 or 3 RAW levels of signal to the noise.
There is no concrete threshold; the image simply gets less and less
detailed and more noisy as the noise-to-signal ratio increases.
--


John P Sheehy

In article , wrote:
In message ,
(DoN. Nichols) wrote:

[ ... Medical Nikkor flash overexposure on Nikon D70 ... ]

O.K. Note that the chart which comes with the lens suggest that
for 2:1 reproduction ratio, a maximum ASA (now ISO) of 50 is suggested
for color, and 25 for B&W. (This is an old lens system, of course.)
The chart is incomplete, but it looks like it will require an ASA of 25
(or perhaps 32) with the full 3:1 ratio, so I may have to cut the
illumination anyway. I can't see pushing it that far down.

I suggest that you try putting some magenta filters cut out from clear
magenta sheets of file folders or something. I use "Flomo" folders; two
sheets thick of their magenta folder over my Canon 550EX gives the flash
almost the exact color that is native to the camera's channel
sensitivities.

Now *this* makes sense. Filtering at the flash output, instead
of in the stack of close-up lenses at the input makes a *lot* more
sense here.

I'm not familiar with the "Flomo" folders. Would I find them at
my local "Staples" or some similar office supply store?

The next trick would be to find some form of neutral density
filter material in a similar cuttable form.

The noise is less chromatic when you do this, and weaker
to boot, so the shadows are much quieter. A neutral density on the lens
will cut all three color channels equally, but ideally, for the D70, you
want to cut the green the most, and the red the least. If the lighting
is mixed with daylight or tungsten ambient, then a filter on the lens
may do the trick.

As explained elsewhere, a filter on the lens seems to not be an
option, but the exposure problem only occurs in the extreme close-ups,
between 0.66:1 and 3:1 ratios, and there, even with the filtering, the
flash will predominate. The lens stops down to about f45, and that is
where we are with the extreme close-ups.

I shoot in daylight with a Hoya FLD filter, which
doesn't get white balance perfectly native, but gets it closer. A
double-strength FLD would probably be perfect for the 20D, 10D, D70, or
any other DSLR with the same general color channel sensitivity. I
haven't experimented with the Tiffen magenta filters or the Cokins yet
(of course, the Cokins wouldn't work with your specialty lens).

Hmm ... perhaps make a mount at the *rear* of the lens? If
there is enough room so the mirror won't hit it?

If you want to see what's really going on, get a matte white target, and
take a picture at each ISO with it, at the magnifications in question,
in RAW, and then load the RAW files into IRIS which will show the exact
RAW values, and you can calculate how much headroom you have at a given
ISO. Do this both with and without a magenta filter over the flash.

O.K. IRIS is probably not an option, as I am using unix
machines, not Windows or Macs, but the capability exists in other
programs which I *do* have.

And this all started out in answering a question as to why you
would want to shoot at ISO 100 (if possible). :-)

Thanks much,
DoN.
--
Email: | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---

In article coh.net,
Philip Homburg wrote:
In article ,
DoN. Nichols wrote:
Not a consideration on the Cannon, but my Nikon D70 has a
minimum ISO of 200, and this prevents me from shooting at the highest
magnification range of the old 200mm Medical Nikkor (with the built-in
ring flash). It does the exposure calculation based on an ISO (actually
marked as ASA) setting ring, and a magnification ratio ring, resulting
in setting of the aperture ring.

What may work is to set the shutter speed sufficiently high that only part
of the flash output is captured . This works on the D1 (though I didn't try
it with a medical Nikkor), and I think that it is also supposed to work with
a D70.

I'll try that. That sounds like a good option, as does the
suggestion of cutting out filters to fit over the ring flash instead of
over the lens itself.

Thanks,
DoN.


--
Email: | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---

In message ,
(DoN. Nichols) wrote:

I'm not familiar with the "Flomo" folders. Would I find them at
my local "Staples" or some similar office supply store?

They are common at "99 cent" stores here in the NYC area. Staples
stores only carries Staples folders, and I haven't seen anything
equivalent there.
--


John P Sheehy