Johnny Johnson wrote:
> At 04:59 PM 7/5/2006, Moose wrote:
>
>
>> And as I've posted before, it turns out that the 5.1mp in an APS-C sized
>> central portion of the 5D sensor easily and obviously resolve more
>> detail in the subject than the 6.3mp of the 300D sensor
>> <http://www.moosemystic.net/Gallery/tech/Canon%205D%20vs%20300D/C300v5D.htm>.
>>
> Hi Moose,
>
> I hate to join Walt in the "I don't always agree with Moose" club but...
Wow! Two people disagree with me in less than 24 hours. I must be doing
something right! :-)
> I've wondered about that test since you first posted it. While I do
> see that the 5D image appears sharper than the one from the 300D I
> don't see any way to tell which image actually has the higher
> resolution.
Oddly enough, I'm more interested in which result looks sharper to me
than what numbers turn up, since that's how I will be evaluating actual
photographs made with them. That was just a quick test, and might have
been a preliminary to something more formal. Except the difference in
sharpness was so striking that I haven't bothered to go further so far .
> To do that with any certainty I believe that you would
> need to shoot some type of resolution chart. I did that last
> September by shooting a USAF resolution chart with both a 10D (same
> size sensor as the 300D) and a 5D. All exposures were made at the
> same subject to target distance and using the same 50mm lens. My
> tests showed that the native resolution (lines per mm on the sensor)
> was considerably higher for the 10D as would be expected based on the
> pixel pitch of the two sensors.
The 10D and 300D sensors are the same, I believe. The pixel pitch isn't
all that different, 122 px/mm for the 5D and 135 for the 300D, about 11%
denser. Is that "considerably higher"?
> I don't remember the exact numbers
> but the difference was large enough that it was obvious to me that if
> I ran out of reach with my longest telephoto that I was better off
> switching to the 10D body than cropping a 5D image.
>
Obviously, our results vary, and I don't have any numbers for mine. All
I know is that I took the full frame of the 300D and calculated, based
on the pixel pitch, the size image measured in pixels on the 5D that
would cover the same physical sensor area as the area of the 300D
sensor. As expected from the above 11% density difference, the 5D image
had about 10-11% fewer pixels. I then resized one sample by upsampling
the 5D image and resized the other by downsampling the 300D image, so
that any effect of the resampling would show up.
By putting them on top of each other, I could switch back and forth
without moving my eyes. And the result seems obvious to me. The edges of
the letters and the detail of the serifs are much clearer on the 5D sample.
> Out of curiosity I just went to the DP Review site and looked at his
> resolution testing of the 300D and 5D. Those tests give a 10% edge
> to the 300D in the horizontal direction and a 15% advantage to the
> 300D in the vertical direction (If my number crunching is
> correct).
Hmmm... The 300D test shows H= 1600, V= 1450 LPH. The 5d shows H= 2300,
V= 2000. I come up with +44% H and +38% from those numbers. Sure you
were reading the 300D test? Even the 20/30D sensor doesn't give 15%, the
5D is over 20% above them.
HOWSOMEVER, the dpreview resolution numbers are not absolute numbers,
but normalized numbers so sensors of different sizes may be fairly
compared to each other. That's what the "LPH" designation means, Line
pairs Per Height. So a 1 mm high sensor with an absolute resolution of
2000 lp/mm and one with a height of 10 mm with an absolute resolution of
200 lp/mm would both have an LPH of 2000. So those numbers can't be used
directly for what we are talking about, the resolution of part of one
sensor compared to all of another.
Using height only, the 300D resolves 1600 lp over 15.1 mm, or 106 lp/mm.
The 5D resolves 2300 lp across 23.9 mm, or 96 lp/mm. Oops, there's that
10-11% again. So according to the dpreview resolution testing, the 300D
should resolve about 10% more detail than an APS-C sized portion of the
5D sensor.
That is just what I had calculated before buying the 5D, and it's close
enough for me. So to get the same rez on the same subject area with the
5D takes a little more size, so what, it's in the same ballpark. I guess
now I may do some real world testing at 300 mm of natural subjects,
since tele's where I want it to work with this. Hmmm, I wonder if I
already have some samples from shots I did for another reason?
> I'm just wondering if you might rethink your conclusions
> if you tried shooting a resolution chart instead of "same size" type face.
>
Based on the numbers I come up with above, both from theory and from the
dpreview tests, no, I don't see any reason to do my reproduction of the
resolution chart tests. As to why my simplistic tests show what they do,
who knows? I don't see how it can be the methodology. I just replaced
one camera with the other with the same lens on the same QR on the same
tripod, without changing or moving anything else. Theory says one
should be sharper and the simple test shows the opposite. Maybe I have a
particularly lame 300D and/or a particularly great 5D, I don't know.
Maybe the sharpening on the 5D is different than the 300D, but I don't
see how I could sharpen the 300D image to equal the 5D. I seem to recall
that I tried that briefly.
In any case, results are better than I expected, so I'm happy so far.
Moose
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|