Some comments:
- It's already been done, and with much more rigor than you are
proposing. The only example I can recall off the top of my head is this
<http://luminous-landscape.com/reviews/cameras/1ds/1ds-field.shtml>.
Unless the 2 shots are taken at the same time of the same subject, there
are so many variables that meaningful comparison is compromised.
- Nobody is really claiming that 6mp DSLRs are as good as or better than
medium format film. The above comparison is done with an 11mp, full
frame sensor. By using MF film, you neatly avoid the real field of
comparison, 5-6mp DSLR vs 35mm film SLR.
- A Can*n G5 is not a DSLR. It is a fixed lens prosumer camera. It has a
sensor size of 5.32x7.18mm. The Can*n 10D and 300D have a sensor size of
15.1x22.7mm, a difference of about 9 times in area. Adjusting for
difference in number of pixels, the G5 has a pixel pitch of about 2.8µm
and the 10D is 7.4µm, taking the ratio down to about 7:1. I suspect that
may have an effect on the kind of comparison you are making. :-) So if
you are going to do it, get a real DSLR image.
- Since we are talking interchangeable lens cameras, lens has to be
considered. There is no way lenses won't be a factor, but to minimize
that factor, really first rate lenses should be used.
- The light sensitivity, "speed", of the sensor makes a difference. For
a fair comparison, film speed and iso setting of the DSLR should be the
same.
- As Scott pointed out, you can't really use JPEGs, or any lossy form,
as the source for creating meaningful comparisons. Sure, you have to go
to JPEG to display them, but only at the end. And any editorial comments
by the tester should be based on uncompromised data before compression.
This whole comparison business is really difficult and complex. A few
months ago, Walt offered to the list a CD of some scans from his new
Minolta 5400dpi scanner. One of the images was a lovely bearded iris. I
found a film shot of a similar iris at a similar size on the film from
my garden and scanned it at 2720dpi, half the resolution, but the best
my little scanner can do. I then upsampled my image using Fred Miranda's
SIPro to 200%, giving almost exactly the same image size in pixels of
the 2 irises. The upshot? My iris was at least as sharp as Walt's. Now
can I claim on that basis that no more than 2700dpi is needed for
scanning 35mm film? No, certainly not, too many variables. Irises in the
real world move around; maybe he has some very subtle movement based
unsharpness that would never be noticed without such picky examination.
The lenses weren't the same. The film wasn't the same. Then later
discussion revealed that he was using ICE and ICE definitely has an
effect on sharpness at the level of the comparison.
I could go on and on (and already have :-) ). The important points:
- Comparison that really proves anything is quite difficult.
- The point is the images. In my story above, it is easy to miss the
fact that both Walt's and my iris images will make beautiful prints of
at least 11x14 and probably 20x24 size. And isn't that the point?
Actually, I like the light and framing of Walt's shot better.
"All you need to take the most beautiful pictures in the world is a
simple Praktika with a 50 mm lens.
Stop wasting your time talking cameras, and go take pictures!"
— Hans Vree*
This was taken with a 2 mp auto only P&S camera
<http://galleries.moosemystic.net/Glads/pages/10-1053.htm>, as was this
and all the others in this gallery
<http://galleries.moosemystic.net/Glads/pages/14-1075_.htm>. What more
do I need for the web? Or even an 8x10 print?
All that said, I like playing with what my technological toys, both film
and digital, hard and soft, can do. If you post a photo of something
common that I might be able to find around here, I'll be happy to try
out a similar shot and send it to you.
Simon Worby wrote:
>I've decided to join the Digital vs. Film discussions...
>
>Please could someone send me a sharp focus, no camera-shake image from a good
>DSLR (preferably E-1, but not necessarily). Anything with good details will
>do...
>
>I intend to do the following:
>
>Assume it is to be printed at 9x6, but not actually do it. Take a 1/2" by 1/2"
>portion of the picture (which I reckon will be 160 pixels square for a 6.1
>mega pixel DSLR). Blow up to 4x (i.e. 320x320 so people can see the
>resolution).
>
How you do this will have a big effect on the results. To be fair to the
DSLR, you should use stair interpolation. It would also be nice to show
the original DSLR image vs the film image downsampled to the same pixel
size.
> Compare to a 1/2" by 1/2" scanned portion from a film camera off 9x6 film.
>
Why so small? If people are to be viewing this on the web, the image
dimensions should be at least 5-600 pixels wide. I can't tell much from
squinty little images.
>I appreciate the film camera will be at a disadvantage because of the
>(digital!) scanning process, but I intend to compensate if I can by scanning
>in at 720dpi
>
You should scan at the scanner's full optical dpi, then downsample,
preferrably using something like Fred Miranda's WPPro. Without going
into detail, this is the way to assure the best quality.
>.........
>I know my methods aren't totally scientific, but I'm trying to be objective in
>my own mind, to help me decide whether digital is the way to go yet...
>
A wordy and opinionated Moose
* from this interesting essay
<http://luminous-landscape.com/essays/digital-to-analog.shtml>. Yes, I
do have a Praktika with 50mm lens, but mostly learned with a Topcon with
55mm lens. I recently used the 50mm lens from the Praktica on a 300D and
it is still a fine lens
<http://moosemystic.net/Gallery/WaltChall/pages/CRW_0145com.htm>.
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|