contrast in p&s digital cameras vs film

acl wrote:


Well I wasn't trying to make a philosophical point; I was trying to
say that the fact that the response at some point in the capture
process is linear doesn't directly result in "flat" or whatever
output. And I drew an analogy to illustrate what I mean, namely, that
film also seems to react linearly before development, so any
difference in the results can't be due to that (we ignore the
difference in high intensities).

Anyway, my response to the OP would be that canon compact cameras seem
to do better than others (that I tried, ie minolta and fuji) in terms
of colour. Or he can get a photoshop plug in or a program to take an
image and simulate some type of film (most are very bad though). Or
process it himself in photoshop, gimp, paintshop etc (I think
paintshop has automatic tools for this too). In short, play with the
various tools and techniques until it finally works like he wants.


It's been very interesting reading all the posts in this thread. Gives
me some information that I wasn't very clear about (had only heard it
mention here and there) and it has prompted me to know about it and
experiment with the digital pics. I use gimp, and tried a few curve
setting last night and liked the improvements.

Thanks,
-HS

On Oct 14, 10:37 pm, Doug McDonald
wrote:
Roger N. Clark (change username to rnclark) wrote:



It seems like film low level response might be linear,
but I can't see why chemically regarding the characteristic curve
it would change to a curve in the toe. So I don't know
what film does at the latent image stage. But does that
matter, as the latent image can not be viewed and must be
chemically processed to see, unlike digital camera raw data.
I don't think you can develop film, say Fuji Sensia,
or Tri-X and not get a curving toe.

Most B&W negative film is quite linear in the toe, and only
in the toe.

But you are talking about developed film, yes?

Remember that it has fog. To see the linearity you
have to measure the transmission in the toe region, correct for
the possibility of overlapping grains if the transmission drops much below
85% (after compensation for surface reflection) and then
subtract off the background. Just before CCDs obsoleted film,
astrophotographers took multiple short exposures where the
transmission of the negative in regions of maximum broad field
density (i.e. excluding stars) was about 20 to 50%, scanned
them and added them together. This produced lower limiting
broad field density than using one long exposure, simply
because one no longer lost information to overlapping grains.
But it was very very ticklish business.

Doug McDonald

On Oct 14, 10:53 pm, "H.S." wrote:
acl wrote:

Well I wasn't trying to make a philosophical point; I was trying to
say that the fact that the response at some point in the capture
process is linear doesn't directly result in "flat" or whatever
output. And I drew an analogy to illustrate what I mean, namely, that
film also seems to react linearly before development, so any
difference in the results can't be due to that (we ignore the
difference in high intensities).

Anyway, my response to the OP would be that canon compact cameras seem
to do better than others (that I tried, ie minolta and fuji) in terms
of colour. Or he can get a photoshop plug in or a program to take an
image and simulate some type of film (most are very bad though). Or
process it himself in photoshop, gimp, paintshop etc (I think
paintshop has automatic tools for this too). In short, play with the
various tools and techniques until it finally works like he wants.

It's been very interesting reading all the posts in this thread. Gives
me some information that I wasn't very clear about (had only heard it
mention here and there) and it has prompted me to know about it and
experiment with the digital pics. I use gimp, and tried a few curve
setting last night and liked the improvements.


Hi, two things:
a) Don't believe what is said in this newsgroup without confirming it
somehow (Roger and Doug in this thread are reliable sources, but many
others aren't-don't make the mistake of equating a confident attitude
with actual knowledge :] )
b) About processing in general, take a look at these
http://www.ronbigelow.com/articles/articles.htm
I haven't read most of them, but they seem quite good. He has some on
curves, too. There are also numerous internet resources on what curves
do.

On Oct 14, 9:55 pm, "Roger N. Clark (change username to rnclark)"
wrote:
acl wrote:
On Oct 14, 7:00 pm, "Roger N. Clark (change username to rnclark)"
wrote:
(It typically takes a couple of photons to excite
a grain, but once done, the grain is no longer sensitive to
light; photons must hit other grains to contribute to the image.)

Yes but until this (=you start getting many photons/grain during your
exposure) occurs, the response is linear. No?

It seems like it might be, but I'm not sure. See below.





The standard curve applied in digital cameras is still linear
at the low end. See Figure 8a in the web page I gave.
So the shadows in a digital camera image shows linear
response when raw linear conversion, jpeg (without excessive
contrast added), or raw conversion with a standard curve applied.
That is different than shadows in film (print or slide) which
has the toe.

Well as I tried to say twice, that, ie
So the shadows in a digital camera image shows linear
response when raw linear conversion, jpeg (without excessive
contrast added), or raw conversion with a standard curve applied.
has nothing to do with the fact that sensors respond linearly to
light, but is a result of the curve applied. No? Which is why I say
that how you convert is important.

I would say conversion is a post processing step so anything
can apply. One need not actually convert the raw data,
simply split the data out and get 3 images, one in red,
a second in green, and a third in blue, with no modification
or manipulation of the data.

It seems like film low level response might be linear,
but I can't see why chemically regarding the characteristic curve
it would change to a curve in the toe. So I don't know
what film does at the latent image stage. But does that
matter, as the latent image can not be viewed and must be
chemically processed to see, unlike digital camera raw data.
I don't think you can develop film, say Fuji Sensia,
or Tri-X and not get a curving toe.

Roger

OK, the best reference I found that's easily accessible to me (ie
accessible online from where I am) is this
http://archimede.mat.ulaval.ca/rcs/c...0206&nbre= 17
which gives a summary of the relevant points (dead time during which
the activated site repels other electrons, giving rise to high-
intensity reciprocity failure, and decay time of the Ag atom, giving
rise to low-intensity reciprocity failure). Do you, Doug, or anybody
else, have any references where I can find out estimates of these
times, and more details (up to date, preferably)?

acl wrote:
On Oct 14, 9:55 pm, "Roger N. Clark (change username to rnclark)"
wrote:
acl wrote:
On Oct 14, 7:00 pm, "Roger N. Clark (change username to rnclark)"
wrote:
(It typically takes a couple of photons to excite
a grain, but once done, the grain is no longer sensitive to
light; photons must hit other grains to contribute to the image.)
Yes but until this (=you start getting many photons/grain during your
exposure) occurs, the response is linear. No?
It seems like it might be, but I'm not sure. See below.





The standard curve applied in digital cameras is still linear
at the low end. See Figure 8a in the web page I gave.
So the shadows in a digital camera image shows linear
response when raw linear conversion, jpeg (without excessive
contrast added), or raw conversion with a standard curve applied.
That is different than shadows in film (print or slide) which
has the toe.
Well as I tried to say twice, that, ie
So the shadows in a digital camera image shows linear
response when raw linear conversion, jpeg (without excessive
contrast added), or raw conversion with a standard curve applied.
has nothing to do with the fact that sensors respond linearly to
light, but is a result of the curve applied. No? Which is why I say
that how you convert is important.
I would say conversion is a post processing step so anything
can apply. One need not actually convert the raw data,
simply split the data out and get 3 images, one in red,
a second in green, and a third in blue, with no modification
or manipulation of the data.

It seems like film low level response might be linear,
but I can't see why chemically regarding the characteristic curve
it would change to a curve in the toe. So I don't know
what film does at the latent image stage. But does that
matter, as the latent image can not be viewed and must be
chemically processed to see, unlike digital camera raw data.
I don't think you can develop film, say Fuji Sensia,
or Tri-X and not get a curving toe.

Roger

OK, the best reference I found that's easily accessible to me (ie
accessible online from where I am) is this
http://archimede.mat.ulaval.ca/rcs/c...0206&nbre= 17
which gives a summary of the relevant points (dead time during which
the activated site repels other electrons, giving rise to high-
intensity reciprocity failure, and decay time of the Ag atom, giving
rise to low-intensity reciprocity failure). Do you, Doug, or anybody
else, have any references where I can find out estimates of these
times, and more details (up to date, preferably)?

I think some of the film data sheets may indicate some of these
times. It varies from film to film. I'm mostly familiar
with the low light end (even made reciprocity curves on
some films, hypered in N2 and normal, as an undergraduate in
physics). The question might be what is the 1/e loss time?
It can vary from a few minutes on bad films to hours for
the good films (like Kodak 103aF used in astronomy).
I don't remember how the manufacturers varied this effect
(chemically in the making of the emulsion).

Roger

Doug McDonald wrote:

Most B&W negative film is quite linear in the toe, and only
in the toe. Remember that it has fog. To see the linearity you
have to measure the transmission in the toe region, correct for
the possibility of overlapping grains if the transmission drops much below
85% (after compensation for surface reflection) and then
subtract off the background. Just before CCDs obsoleted film,
astrophotographers took multiple short exposures where the
transmission of the negative in regions of maximum broad field
density (i.e. excluding stars) was about 20 to 50%, scanned
them and added them together. This produced lower limiting
broad field density than using one long exposure, simply
because one no longer lost information to overlapping grains.
But it was very very ticklish business.

Doug McDonald

Doug,
I know of only one person who stacked scanned film astrophotos.
Did a great job, by the way, and the response need not
be linear for it to work, especially the same subject exposed
the same amount of time. People do that now with jpegs.
But scanned film brought out more faint detail than one could
print anyway, so it was/is a win/win situation.

Roger

On Oct 14, 9:07 am, acl wrote:
On Oct 14, 9:05 am, "Roger N. Clark (change username to rnclark)"



wrote:
H.S. wrote:
Hello,

I was wondering if anybody can comment on the quality of contrast in the
pictures taken with a point and shoot digital camera compared to the one
obtained on 35mm film.

I have seen pictures from Canon A520 and a few older models. I can
usually pick them out from the ones scanned from negatives quite easily.
The ones taken with digital cameras usually appear to have less contrast
and less amount of colors (less saturated colors?).

Anybody know of any recent digital cameras in which the colors are
better and we can have more contrast? I know this may appear to be a bit
vague, but related comments are welcome.

The main reason is that film has a toe in its characteristic
curve and digital cameras are linear.

Is that so? My reading of the Gurney-Mott business (a couple of years
ago and very superficial) led me to conclude that the formation of the
latent image on film is also a linear process (as long as you're not
in the region of low or high intensity reciprocity failure). The
characteristic curve comes when you develop it. So I don't see how
this is any different from digital capture, where the detection is
linear and you then apply a curve. I'm not an expert and just scanned
the paper quickly, so may have completely misunderstood it.

I'd say the difference the OP see is because of the processing, which
I think deserves a lot more attention than people seem to think. I
know this is also what you say below, but I think saying that one is
linear and the other isn't is not accurate (as it refers to different
stages of capture/development in each case). But maybe I am wrong.

They are also linear
at the low end after the "gamma" tone curve is applied.
Learn to use curves to add an s-curve response to give a
film-like response. You can boost contrast in camera, but
post processing gives more control over shadow and highlight detail.

Depends on the film. Some are quite linear, others are quite S-
curved. Look at the characteristic curves of various films.

On Oct 15, 3:33 pm, Don Stauffer in Minnesota
wrote:
On Oct 14, 9:07 am, acl wrote:



On Oct 14, 9:05 am, "Roger N. Clark (change username to rnclark)"

wrote:
H.S. wrote:
Hello,

I was wondering if anybody can comment on the quality of contrast in the
pictures taken with a point and shoot digital camera compared to the one
obtained on 35mm film.

I have seen pictures from Canon A520 and a few older models. I can
usually pick them out from the ones scanned from negatives quite easily.
The ones taken with digital cameras usually appear to have less contrast
and less amount of colors (less saturated colors?).

Anybody know of any recent digital cameras in which the colors are
better and we can have more contrast? I know this may appear to be a bit
vague, but related comments are welcome.

The main reason is that film has a toe in its characteristic
curve and digital cameras are linear.

Is that so? My reading of the Gurney-Mott business (a couple of years
ago and very superficial) led me to conclude that the formation of the
latent image on film is also a linear process (as long as you're not
in the region of low or high intensity reciprocity failure). The
characteristic curve comes when you develop it. So I don't see how
this is any different from digital capture, where the detection is
linear and you then apply a curve. I'm not an expert and just scanned
the paper quickly, so may have completely misunderstood it.

I'd say the difference the OP see is because of the processing, which
I think deserves a lot more attention than people seem to think. I
know this is also what you say below, but I think saying that one is
linear and the other isn't is not accurate (as it refers to different
stages of capture/development in each case). But maybe I am wrong.

They are also linear
at the low end after the "gamma" tone curve is applied.
Learn to use curves to add an s-curve response to give a
film-like response. You can boost contrast in camera, but
post processing gives more control over shadow and highlight detail.

Depends on the film. Some are quite linear, others are quite S-
curved. Look at the characteristic curves of various films.


I was talking about the latent image. See he
... led me to conclude that the formation of the
latent image on film is also a linear process...

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

The main reason is that film has a toe in its characteristic
curve and digital cameras are linear. They are also linear
at the low end after the "gamma" tone curve is applied.
Learn to use curves to add an s-curve response to give a
film-like response. You can boost contrast in camera, but
post processing gives more control over shadow and highlight detail.


I think that *most* P&S users do NOT post process their images and often go
directly from SD card to upload to the processor. Those that do post
processing are probably mostly limitted to red-eye removal or cropping. I
think a relatively few P&S users do true post processing as you describe.

--
Thomas T. Veldhouse

Ninety percent of the politicians give the other ten percent a bad reputation.
-- Henry Kissinger

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

film-like response. You can boost contrast in camera, but

er ... how can one do that?