How many megapixcels equivalent to 35 mm in quality

On Sun, 15 Jun 2008 16:44:38 -0700, Scott W wrote:

> >On Jun 15, 1:25Â*pm, Stefan Patric <n...@thisaddress.com> wrote:
> >
> > The prickly problem is to choose a color negative film (I'm going to
> > be
> > making color prints for evaluation) that comes closest to matching the
> > gamma of the 4 or 5 stop range centered around 18% grey of the digital
> > camera's MTF curve. Â*Since I can't change the film's parameters, it
> > probably will require that the digital camera's internal settings for
> > contrast, saturation, sharpness, etc. be adjusted to match the film as
> > closely as possible.
>
> Your not trying to say that a digital is limited to 4 or 5 stops are
> you?

No, of course not, but your average film exposure response curve has a
toe and a shoulder, and is only straight about +- 2.5 stops either side
of neutral grey. Digital response curves are basically straight through
their entire effective range having little toe and shoulder. I merely
want to match approximately the straight line portion of the film curve
to the corresponding area on the digital curve, so that the contrast
range between fully detailed shadows and highlights, the basic range I'm
concerned with, are the same. More than likely, if I pick a "low"
contrast color negative film like Kodak Professional Portra 160NC and
shoot outside on a "Standard Contrast" day, the default settings for
contrast and saturation on any digital camera will be sufficient, but
I'll test it to be sure.

> > Actually, it's the lens that's the major limiting factor as far as
> > image
> > resolution is concerned not the film. Â*Film can and always has been
> > able
> > to resolve more even if the lens can't deliver it. Â*However, I'm more
> > concerned with practical, real world results than strictly scientific
> > testing. Â*Besides, I don't have any resolution charts. Â*And my tests
> > won't involve scanning the film.

> Even a cheap lens will out resolve film, at least if it is stopped
> down and you are viewing the center of the field. I use a Canon 350D,
> which has a pixel spacing of right around 4000 pixels/inch. I get far
> sharper images at the pixel level with the digital then I ever got
> from a 4000 ppi film scan. And I have gotten sharper prints by first
> scanning a negative and printing from that compared to a wet print. I
> can also use a 1.4x teleconverter and still get a sharp image at the
> pixel level, that would be close to scanning film at 5600 ppi, 4000ppi
> film soft but 5600 ppi film scan are super soft, when viewed at the
> pixel level.

A topic for another day and another thread.

> Now I can tell you are going to tell me it is the scanning that is at
> fault, but I have seen just one film that really does show the
> resolution of a decent lens, Gigabit film. Here is a scan that Max
> Perl did using this film.
> http://www.terrapinphoto.com/jmdavis..._crop_1000.jpg
> A very impressive scan.
> So how does the scam equipment do with a color film, not even close to
> as good.
> http://www.terrapinphoto.com/jmdavis..._crop_1000.jpg
> If the same scanner can pickup great detail with Gigabit film it
> should be able to do the same with color film, if the detail was
> really there. Clearly the color film is not coming close to capturing
> all the detail that the lens produces.

I'm not really concerned with scanning or which is better, digital or
film, or whether scanned film will produce better prints than prints made
directly from film. I am merely trying to determine to my own
satisfaction based on my criteria: How many megapixels does it take to
equal professional 35mm color negative film of equal speed? Some say 6
to 8; others say 18 to 24; a few say 40; others say 3. Since no one has
every made such a test exactly the way I intend to, that I can find
anyway, the results will be interesting, maybe, even enlightening.

And FWIW, I'm neither film bigot nor a digital fanatic. I'm a commercial
photographer and use both, whichever fulfills the client's requirements.
Many times I use both. After all, cameras as well as all that other
photo paraphernalia are just tools.


Stef

Stefan Patric

On Mon, 16 Jun 2008 20:10:43 +0000, Dave Martindale wrote:

> Stefan Patric <> writes:
>
>>The prickly problem is to choose a color negative film (I'm going to be
>>making color prints for evaluation) that comes closest to matching the
>>gamma of the 4 or 5 stop range centered around 18% grey of the digital
>>camera's MTF curve. Since I can't change the film's parameters, it
>>probably will require that the digital camera's internal settings for
>>contrast, saturation, sharpness, etc. be adjusted to match the film as
>>closely as possible.
>
> But isn't that a biased test? You're taking the "film look" as a
> standard, and attempting to adjust the digital camera to match it as
> closely as possible. That's likely to compromise some or all of the
> advantages of the digital image (e.g. higher contrast at low spatial
> frequencies) compared to film, while leaving film at its optimum.

No. If A=B and B=C, then A=C. If scene contrast matches film's contrast
range, and film's contrast range equal digital's contrast range, then
scene contrast matches for digital's contrast range. I'll just wait for
a day with "Standard Contrast." The digital camera's default setting
should be just fine and "match" the gamma of that straight line, 5 stop
range in film.

> It seems to me that if you're doing subjective tests, you should adjust
> the digital camera for subjectively best images - not images that match
> film as closely as possible.

If the shooting conditions are properly chosen, the best output from the
digital camera should be the defaults, except for sharpness, which I
would try to optimize--highest sharpness with the least noise. Most
digitals always have low sharpness defaults, since it is assumed that the
image will be reprocessed.

>>> So how will you create an equitable measure of "equivalent quality"?
>>> Without one, you can't have an equitable test.
>
>>You're thinking too much like a scientist. Think like an artist: If it
>>looks the same or better or worse, then it is.
>
> Ok, then you probably should call your test "subjective" rather than
> "equitable". And the results you get are going to depend a lot on the
> content of the image.

The evaluation will be partially subjective--i.e. which print looks
better?--but will also be empirical--which print has recorded fine
details sharper? It is the parameters of recording the initial images,
and producing the test prints that I will try to make as equitable as
possible.

>>I'm merely trying to determine to my satisfaction at what point silver-
>>based, color prints directly made from each type of camera, without any
>>"post" other than overall color and density adjustments, are for
>>practical purposes indistinguishable as to which system was used to
>>produce the image. At that point, there is equivalence.
>
> And I'd argue that this will be very elusive. Unless you deliberately
> cripple the digital images, for many subjects the higher contrast at
> lower spatial frequency of the digital image will make it look "better"
> even while its limiting resolution is far below that of the film you're
> comparing against. The digital pixel count at which "as good or better"
> happens will depend on the image content, print size, and viewing
> distance. So I don't think you'll ever get a single number that's very
> useful for predicting "equivalent quality".

Elusiveness has never been a deterrent.

> You might be able to establish a set of numbers, each for a pretty
> restricted domain. For example, "For landscape photography where prints
> are 2 ft x 4 ft and the viewing distance is 2 ft, an X megapixel digital
> camera is subjectively equivalent to film Y recording a 24x36 mm image".

I'm just trying to get an "in the ballpark" number. Nothing more.


Stef

Stefan Patric