How many megapixcels equivalent to 35 mm in quality

David J. Littleboy wrote:
> "Paul Furman" <paul-@-edgehill.net> wrote:
>> frederick wrote:
>>> I'm not sure - 5d vs APS-C. I've been using both (5d and D300). My
>>> comparison of 17-40l @ 18mm vs Sigma 10-20 @ 12mm, the D300/10-20 won for
>>> corner performance, edge performance similar.
>> Frederick, how does low light/noise performance compare? Dynamic range is
>> another area I'd expect full frame to excel with.
>
> The physics has it that you'd expect a 1-stop difference (both in noise and
> DR (which are, of course, the same thing<g>)). Is that enough to write home
> about?

Yes, thanks, and it wasn't actually obvious to me that dynamic range &
low light performance are the same thing. So in good light or with a
tripod, does full frame give a noticeable improvement in dynamic range?
I've heard it's only noticeable in high ISO shots but should be also
good for brightening shadows. My D200 ISO 100 shadows can get annoyingly
noisy so I would notice & appreciate improvement there. Does full frame
look like ISO 50?

> My guess is that it would be hard to provide the extra 50% resolution the
> D300 requires at the sensor plane* at wide f stops. Thus there is a double
> hit (more noise, softer images) when you try to do low-light with 12MP APS-C
> vs. FF. But you are still miles and miles ahead of P&S.
>
> *: For the same 12MP image, the D300 pixels are a lot closer together. So
> the image hitting the sensor must have higher resolution to achieve the same
> sharpness.

If the lens is sharp, full frame takes advantage of more good glass for
capturing detail and it sounds like 12MP is getting close to where that
matters, but maybe just marginally.

>>> In general though, at 19x13", there's no significant difference, D300/5d.
>
> Which is why Canon needs an 18MP 5DII and a 20/2.8 II.
>
> David J. Littleboy
> Tokyo, Japan
>
>


--
Paul Furman
www.edgehill.net
www.baynatives.com

all google groups messages filtered due to spam

Paul Furman

David J. Littleboy wrote:
> Paul Furman wrote:
>> Yes, thanks, and it wasn't actually obvious to me that dynamic range & low
>> light performance are the same thing. So in good light or with a tripod,
>> does full frame give a noticeable improvement in dynamic range? I've heard
>> it's only noticeable in high ISO shots but should be also good for
>> brightening shadows. My D200 ISO 100 shadows can get annoyingly noisy so I
>> would notice & appreciate improvement there. Does full frame look like ISO
>> 50?
>
> I shoot the 5D at ISO 200 most of the time and am very happy with the
> shadows. If I shoot at ISO 400 _and_ underexpose, I'm unhappy with the
> shadows. If I hold my nose, expose correctly at ISO 3200, and make a 12x18
> print, my jaw hits the floor.
>
> The bad news is that ISO 100 is no better than ISO 200 for noise/DR. That
> requires the DIGIC III.

I don't understand, the processor it doesn't have, improves low ISO?

--
Paul Furman
www.edgehill.net
www.baynatives.com

all google groups messages filtered due to spam

Paul Furman

On Sat, 31 May 2008 06:59:18 +0900, David J. Littleboy wrote:

> "Stefan Patric" <> wrote:
>> On Sat, 31 May 2008 00:10:23 +0900, David J. Littleboy wrote:
>>>
>>> People who have actually looked, find that 12.7MP (i.e. the 5D) looks
>>> very similar to _645_ slide film, and 24x36mm is nowhere close to
>>> either.
>>>
>>> http://www.ales.litomisky.com/projec...sus%20Digital%
>> 20Shootout%20(Hasselblad,%2035mm,%20Canon%205D).ht m
>>>
>>> http://www.shortwork.net/equip/review-1Ds-SQ-scantech/
>>
>> Point of Order: The links you site as well as all the others sited in
>> this thread, and all those I've read, but not mentioned here, do NOT
>> truly compare digital to film. Instead they compare digital to
>> digitally scanned film, and therein lies the caveat.
>
> No caveat required. People who have compared the wet projection prints
> they've made to prints made from scans find that scanning makes as good
> or better prints. Partly because the projection system itself degrades
> the image, but mostly because the last generation of scanners (Nikon
> 5000, Nikon 9000) were quite good.

You missed the point: The question is how many megapixels (effective,
that is) does it take to be equivalent to film. To answer this question,
one must compare film prints directly, warts and all, with digital
prints, preferably produced from the file directly out of the camera, no
post permitted. To scan the film, first, do post on that digital image,
then make the print doesn't really answer the question: That was the
caveat I was referring to.

>> Any advantages film might or
>> does have would be negated either partly or wholly by the scanning
>> process. In reality, you are comparing a digital "original" to a copy
>> of an original. Is that an equitable comparison? I think not.
>
> I think so. I shot a lot of medium format, scanned it on a Nikon 8000,
> and checked the film carefully with a 20x loupe and a 60x microscope.
> The scan gets by far the vast majority of the valid photographic
> information that's actually on the film.

Did you print the film scan without doing any post work on it? Probably,
not. You have to do post to try to improve the scanned image to get a
digital file suitable for printing. That means that the scan degraded
the film image to a certain degree, maybe only a little or perhaps a lot.

Now, I'm not implying that today's scanners don't obtain all the info in
a film image. After all a 4000 dpi scan is more resolution that the
average 35mm lens-film combo can record under "normal" conditions.
However, the scan does degrade the original. It can't help but, since
scanning involves lenses and transmitted light, which obey the laws of
optics--scattering, diffusion, etc.--causing contrast to increase and
details to soften requiring post work to correct.

Of course, all this doesn't answer the original question.

> The "scanning doesn't get all the information" is a common claim by film
> fans, but it's nothing more than hopeless grasping at straws. A 5D file
> simply captures more useful photographic information than 24x36mm of
> film does. (This might be different for gigabit film, but its a slow,
> high contrast B&W film that's not practical for actual photography.)

Maybe, a few years ago, but today's scanner have higher resolution than
the films they scan. I've always been an advocate of gross
oversampling. You can always discard what you don't want or need.

>> A true test of digital vs film would be direct out of the camera to
>> photographic emulsion 20X prints with only overall color balance and
>> density corrections permitted, and only during the printing process;
>> then second 20X prints where any manner of manipulation is permitted to
>> make the best possible prints.
>
> 20x enlargements from film (any film that you'd actually want to use)
> are really really ugly. Even the best films are either grossly soft,
> grossly grainy, or both at 13x. If you care about the quality of your
> prints, you make 8x and smaller enlargements.

That's why 20X: It's well beyond what is accepted as film (and digital)
capability. Makes it easier to pick the better. Or the worse.

>> I have never found such a test in all my searches: The film is always
>> scanned.
>
> Again, lots of people who are now scanning pulled out their wet
> projection prints and found that the scans were as good, or more often,
> simply better.

With post processing you can make almost anything look better than the
original. Women have been doing that for centuries with makeup: analog
post processing.

> What I'm saying is that the differences between desktop scanning, drum
> scanning, and the best wet projection printing are tiny compared to the
> differences between different film formats. You just aren't going to
> make 35mm look like 645 by buying a Leitz enlarger and Leitz enlarging
> lenses. At 12 x 18, 645 is an 8.2x enlargement, and at 12x18, 35mm is a
> 13x enlargement. If you care about print quality, the difference is
> large with any of the better printing technologies.

I go by the old rule I learned in college from one of my physics
professors, who was also a professional photographer: "At normal viewing
distances any print that has at least 4 lppm resolution will appear sharp
to the unaided human eye." For your "good" 35mm lenses, this works out
to about a 13X enlargement maximum, but in my film days, I used to make
around 18X enlargements (16x20 paper) that other pros would swear I had
shot with medium format.

(For the curious: Nikon FM cameras, Nikkor AI lenses, heavy tripod,
mirror locked up, Ilford Pan F rated at 25, developed in HC-110 at twice
the B dilution, Beseler 45M enlarger anchored so sturdily that an atomic
bomb blast couldn't induce vibration, Aristo cold light head, 50mm Apo
Rodagon, home-made full frame, nega-flat carrier, home-made vacuum paper
easel, expensive grain focuser, Edwal No Scratch or nose grease. Don't
move and hold breath during exposure. )

> But the 5D produces images that are in the 645 class. And there's no way
> to make 35mm look that good.
>
>> Maybe, I'll do it myself. I have a friend who has a Canon 5D. Now all
>> I need to do is find someone with a Canon film camera, so the same lens
>> can be used. Or I could just use my 30 year old, all manual Nikons for
>> the film part, but that might bias the test toward film due to the vast
>> superiority of old manual Nikkors over today's plastic, loosie-goosie
>> AF lenses.
>
> You are wasting your time. Again, the 5D produces images that are
> roughly on par with 645 film. If I shoot TMX100 in my (6x7) Mamiya 7, I
> see slightly higher resolution for sharp architectural detail than I get
> with the 5D. But there's no way to make 645 beat out the 5D in any
> visually significant way.

I've never considered the quest for enlightenment a waste of time.


Stef

Stefan Patric

In article <>,
"David J. Littleboy" <> wrote:

> "Stefan Patric" <> wrote:
>
> >The question is how many megapixels (effective,
> > that is) does it take to be equivalent to film.
>
> It's been answered. Years ago. The answer is 6 to 8. See the Normal Koren
> references.
>
> My point here is exactly one. The claim that 35mm is equivalent to 12MP is
> off the wall.

David-

Most of what you say agrees with my experience. However, the claim that
film equivalent Megapixels is 6 to 8 seems just as off-the-wall as 12.
I've scanned slides to 8 Megapixels, that were clearly pixel-limited
when enlarged to the point where pixels were evident.

Fred

Fred McKenzie

On Jun 2, 11:52 am, Fred McKenzie <fm...@aol.com> wrote:
> In article <PKWdnUQuI9TC0t7VnZ2dnUVZ_vedn...@giganews.com>,
> "David J. Littleboy" <davi...@gol.com> wrote:
>
> > "Stefan Patric" <n...@thisaddress.com> wrote:
>
> > >The question is how many megapixels (effective,
> > > that is) does it take to be equivalent to film.
>
> > It's been answered. Years ago. The answer is 6 to 8. See the Normal Koren
> > references.
>
> > My point here is exactly one. The claim that 35mm is equivalent to 12MP is
> > off the wall.
>
> David-
>
> Most of what you say agrees with my experience. However, the claim that
> film equivalent Megapixels is 6 to 8 seems just as off-the-wall as 12.
> I've scanned slides to 8 Megapixels, that were clearly pixel-limited
> when enlarged to the point where pixels were evident.
>
> Fred

Indeed. The slower speed transparency films were among the best
performers in resolution. Also, 8Mp would not suffice for Plus-X
developed in dilute developer.

Don Stauffer in Minnesota

Mark Thomas <markt@_don't_spam_marktphoto.com> wrote in news:g1nclu$gh1$1
@registered.motzarella.org:

> G Paleologopoulos wrote:

>> Since when have megapixels been equated to quality???

> Since marketing guru's needed just one number to throw at people...

Well, generally speaking, there is potential for greater image quality
with higher pixel densities. This fact is hidden by glitches in viewing
technology. The fact that we have to downsize images to view them in
their entirety on a monitor or panel is nothing short of a tragedy. If
we had 100MP monitors, and upsampled our images to view them in their
entirety, there would be no perceived need for heavy noise reduction, and
very little perception of noise in the recent high-density sensors.

Our viewing options are horrible. The code used in most programs to zoom
out (less than 1:1 pixel magnification) trashes higher-MP images more
than they do lower-MP images, quite unfairly, but maintaining pixel-level
noise instead of combining pixels in a fairly-weighted manner.
Photoshop's on-screen zooming is nothing short of pure garbage. $79
PaintShopPro puts it to shame, in that regard. Even 100% pixel view,
itself, is a poor way to gauge noise in an image, unless you're comparing
two cameras with the same MP.

> The better large sensors and full-frames, imo, definitely offer better
> usable dynamic range.

But why? Could it be because they are also expensive professional
cameras, and they get better electronics? Despite all the monologues to
the contrary, DR is not directly related to pixel or sensor size. DR is
determined by read noise mainly, and is scaled by the inverse of linear
pixel resolution. Today's better high-MP P&S cameras have more DR than
the worst DSLRs of today and yesterday. Unless, of course, you have a
very conservative definition of DR, like the range from a 10:1 SNR up to
saturation, in which case the photon collection of the sensor (and not
really of the pixels) may play the greatest role in determining that
specific type of DR.

> Digitals are more accurate and repeatable, but you will hear those who
> refer to the plastic look.. Which is more about low noise and lack of
> grain, I think, or perhaps that they have only seen web-sized images.

Who, exactly, are the "plastic" complainers? Are they mostly people used
to film, or do people with little photographic experience, all digital,
also have plastic complaints?

One function of grain, IMO, is that our brains are content that they have
focused on the medium and extracted all it offers. Same happens with
aliased digital images; the aliasing is a clue that some people like to
see, to tell them that focus is successful and complete. Same thing
happens with digital noise, but it usually unsettling, because digital
noise is chromatic noise when conversion tries to get pixel-level
saturation and color, and because the read noise often has patterns in it
which don't dissolve at low viewing magnifications. If digital
photosites were much denser, and display mediums could show each of the
many pixels, then I think that digital would look a lot like velvety
slide film, as long as the pattern noises are conquered (and there's no
reason why they couldn't be).

--
John Sheehy

John P Sheehy

On Jun 3, 10:36 am, fl...@apaflo.com (Floyd L. Davidson) wrote:
> Don Stauffer in Minnesota <stauf...@usfamily.net> wrote:
>
>
>
> >On Jun 2, 11:52 am, Fred McKenzie <fm...@aol.com> wrote:
> >> In article <PKWdnUQuI9TC0t7VnZ2dnUVZ_vedn...@giganews.com>,
> >> "David J. Littleboy" <davi...@gol.com> wrote:
>
> >> > "Stefan Patric" <n...@thisaddress.com> wrote:
>
> >> > >The question is how many megapixels (effective,
> >> > > that is) does it take to be equivalent to film.
>
> >> > It's been answered. Years ago. The answer is 6 to 8. See the Normal Koren
> >> > references.
>
> >> > My point here is exactly one. The claim that 35mm is equivalent to 12MP is
> >> > off the wall.
>
> >> David-
>
> >> Most of what you say agrees with my experience. However, the claim that
> >> film equivalent Megapixels is 6 to 8 seems just as off-the-wall as 12.
> >> I've scanned slides to 8 Megapixels, that were clearly pixel-limited
> >> when enlarged to the point where pixels were evident.
>
> >> Fred
>
> >Indeed. The slower speed transparency films were among the best
> >performers in resolution. Also, 8Mp would not suffice for Plus-X
> >developed in dilute developer.
>
> Any comparison that uses a scanner output (film, slides,
> prints, whatever) as compared to a camera output, is
> flawed. Probably by a factor of 2x.
>
> --
> Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
> Ukpeagvik (Barrow, Alaska) fl...@apaflo.com

Sounds reasonable. I assume you get that by squaring the Kell factor
to get areawise resolution.

Don Stauffer in Minnesota

On Mon, 02 Jun 2008 10:39:32 +0900, David J. Littleboy wrote:

> "Stefan Patric" <> wrote:
>>>
>>> No caveat required. People who have compared the wet projection prints
>>> they've made to prints made from scans find that scanning makes as
>>> good or better prints. Partly because the projection system itself
>>> degrades the image, but mostly because the last generation of scanners
>>> (Nikon 5000, Nikon 9000) were quite good.
>>
>> You missed the point:
>
> No. I was responding to the claim that scanning degrades image quality
> when making prints from film. This claim is somewhere between being
> simply wrong and being so small a difference as to be completely
> irrelevant. It's a straw that fans of 35mm film love to grasp at.

The actual scanning process does degrade the film image, usually an
increasing of contrast and "softening" of detail, particularly very fine
detail. It's a optical process after all. You're copying an original.
And like all copying there are consequences. Post work can mostly
correct, from a practical, real world stand point, those problems, if the
scan is of a high enough resolution.

>>The question is how many megapixels (effective,
>> that is) does it take to be equivalent to film.
>
> It's been answered. Years ago. The answer is 6 to 8. See the Normal
> Koren references.

Not really. All Koren's tests as well as those done by others do their
evaluations and reach their conclusions from SCANNED film. It does not
matter whether scanning produces better prints from film or not. To
reach an actual equivalence answer--megapixels to film--evaluation prints
must be made directly from each. Scanning film and doing post on it
would skew the results. This is not photography, but the scientific
method at work.

> My point here is exactly one. The claim that 35mm is equivalent to 12MP
> is off the wall.

I have a friend, who's retired from 30+ years in the commercial printing
industry--4-color printing press, etc. not photographic--and has been a
freelance photographer all that as well. His learned opinion based on 3
decades of practical experience is that you need a full 35mm frame,
digital camera in the 18 to 20 MP range to equal professionally produced
35mm chromes.

> I'd love to be wrong here: I'd much prefer to use a Conax G, the 21mm
> and 35mm lenses (or even better, the new Zeiss Ikon), and a Nikon 5000
> than the 5D. But 35mm just doesn't cut it, and the 5D does, for the size
> prints I want to make.

This thread has never been about whether digital is better than film,
scanned or not, only about establishing a fair test to find equivalence
one to the other.

> I've always found that 6x6 or 645 is a lot better than 35mm. This was
> true for the 11x14 B&W darkroom work I did years ago and remains true
> for the scan to Epson 2400 12x18s I do today. This isn't really
> arguable, unless one doesn't really understand or care about print
> quality.

Not really at question.

> If I look at 12x18 prints made from the 5D, they're in the 645 class,
> not the 35mm class.

Perhaps my test using my testing method will reach the same conclusion.
Perhaps not. Since no one has actually done the test I've proposed.

> There is a large, qualitative difference between 35mm film and the 5D.
> And it simply doesn't matter what technology you use to produce your
> prints from the 35mm film.

Same as above.

>> To answer this question,
>> one must compare film prints directly, warts and all, with digital
>> prints, preferably produced from the file directly out of the camera,
>> no post permitted.
>
> The "no post permitted" bit is, I think, very wrong. The object is to
> make as good looking prints from one's files as possible, and that's
> what one should compare. Film types do a lot of "post", too. High
> accutance developers do sharpening chemically (check the MTF curves for
> Velvia). And any competent darkroom worker does dodging and burning to
> get the shadows and highlights into sensible ranges on the paper, and
> selects an appropriate contrast grade filter or paper for the negative.

The object is to establish an equitable testing method, and not have a
work flow that will give an unfair advantage to either media.

> Inversely, comparing less than optimal prints doesn't compare what the
> technologies are actually capable of, and so could be misleading.

We are not determining the ultimate capability of each media, only a
simple equivalency test: how many megapixels = film? Or, regarding my
test: Is a 12MP Canon 5D better, equal or inferior to 100 speed 35mm
color print film? If the 5D is so superior as you say, it should be
readily apparent in the prints, even if printed right out the cameras.

FWIW: I never, nor do I ever, accept other people's tests or opinions as
valid without first doing my own tests to verify their results. It makes
for good science and better photography.

So far as I can find, no one has ever exactly done the tests I plan to do
regarding the megapixel to film question. Should be interesting.

Stef

Stefan Patric

Ï "Aaron Kuperman" <> Ýãñáøå óôï ìÞíõìá
news:g1mfth$1p466$...
> How many megapixels does a digital camera need to produce an 8" by
> 10" print equivalent in quality to what one would get from a 35 mm camera
> (or for that matter a larger format) shooting a top of the run film, such
> as a major brand 100-125 ASA color (or black and white) film.
>
> The bottom line, if I replace a film camera with a digital, and want the
> same quality enlargements, am I talking about the current models now
> available (costing perhaps $500 to a bit over $1000), typically with
> roughly 10 MP, or am I talking about something not designed for other than
> professional use such as the $8000 Canon with 21 mp, or perhpas something
> yet to be invented.
The problem is that a digital camera has a much greater *effective*
resolution, because the final file results from processing by the camera
processor (jpeg) or your pc (RAW). OTOH, a film print results just from
enlarging the negative/projecting the slide. Today, and since a long time, I
saw a professional with a film camera, an enormous Hasselblad system. And,
fancy photo printers have usually further improving of the final print, eg
my Canon iP 4300 has the Exif print which uses the exif data to print better
prints.



--
Tzortzakakis Dimitrios
major in electrical engineering
mechanized infantry reservist
hordad AT otenet DOT gr

Tzortzakakis Dimitrios

In article <1212963100.915956@ftpsrv1>, frederick <> wrote:

> Stefan Patric wrote:
> >
> > I have a friend, who's retired from 30+ years in the commercial printing
> > industry--4-color printing press, etc. not photographic--and has been a
> > freelance photographer all that as well. His learned opinion based on 3
> > decades of practical experience is that you need a full 35mm frame,
> > digital camera in the 18 to 20 MP range to equal professionally produced
> > 35mm chromes.
> >
> You don't need 18-20mp to equal "professionally produced 35mm chromes".
> Figures like that were bandied around years ago based on the performance
> of the 2-3mp cameras available at the time, extrapolating that
> performance to what would be needed. Problem with that logic was that
> camera sensors got better as well as getting more pixels.

Frederick-

Suppose a 2mp camera does not produce an equivalent image. Suppose it
is pixelization, not optics, that limit it. Now, how much better does
it need to be?

To double the pixels-per-millimeter, you quadruple the total number of
pixels. You are now at 8mp, and only twice the linear pixels. Is that
good enough? Based on my observations, I don't think so.

Double it again. Now you are at 32mp or four times the linear pixels.
Is that good enough? Perhaps, but you should be able to see how
sensitive the megapixel count is for a change in linear pixels.

Whether the "real" number is 18mp or 24mp or 36mp depends on the film in
question.

Fred

Fred McKenzie