Maximum sharpness on Canon 300D

Archived from groups: rec.photo.digital.slr-systems (More info?)

Hi,

Whats the maximum lens sharpness that the Canon 300D can use? I mean
whats the maximum lpm that the 300D can benefit from? Also, the 20D.

Thanks,

Siddhartha

Archived from groups: rec.photo.digital.slr-systems,rec.photo.digital (More info?)

"Siddhartha Jain" <losttoy2000@yahoo.co.uk> wrote in message
news:1108803834.295404.318740@g14g2000cwa.googlegroups.com...
> Hi,
>
> Whats the maximum lens sharpness that the Canon 300D
> can use? I mean whats the maximum lpm that the 300D can
> benefit from? Also, the 20D.

Theoretically (based on Canon specifications which are unclear about
physical sensor vs image size):
- The 300D cannot exceed 3072px / 22.7mm = 135 lines/mm horizontally,
and 2048px / 15.1mm = 136 lines/mm vertically. That would be equal to
a sample pitch of 7.3 micron on average.
- The 350D (to keep the topic up-to-date) cannot exceed 3456px /
22.2mm = 156 lines/mm horizontally, and 2304px / 14.8mm = 156 lines/mm
vertically. That would be equal to a sample pitch of 6.4 micron.
- The 20D cannot exceed 3504px / 22.5mm = 156 lines/mm horizontally,
and 2336px / 15.0mm = 156 lines/mm vertically. That would be equal to
a sample pitch of 6.4 micron.

That theoretical limit for luminance resolution is deliberately
reduced by an Anti-Aliasing filter, and an attempt is made to
reconstruct that, during Bayer CFA reconstruction. Losses due to
camera shake also play a role.

A more common measure for resolution of so-called discrete sampling
systems is cycles/mm, so to keep things simple you could say that the
300D can resolve up to 68 cy/mm on the sensor, and the 350D and the
20D can resolve up to 78 cy/mm on the sensor.
Again, those are theoretical limits for high contrast features, and
the AA-filter will gradually reduce contrast as that limit is
approached. Low contrast features will be harder to resolve, and with
lenses that produce lower contrast, the visible resolution will suffer
even more. IAW, the 350D/20D requires a somewhat better lens to fully
exploit the better resolution.

Lenses generally are capable of much higher resolution than the
sensor, although contrast may be low. So what you really need to know
is what the combination of lens+AA-filter+sensor brings in terms of
contrast as a function of spatial frequency (=system MTF).

Another important factor is the Raw convertor that's used, and the
amount of sharpening that's applied in post-processing.

So in general, your question can only be answered in general terms.
Good lenses will exceed the capabilities of the sensor. Better lenses
will produce images that look sharper (= have higher contrast
details), especially throughout the image, i.e. also in the corners.
Higher on sensor resolution can be enlarged more, before visible image
resolution suffers. The 350D/20D will better utilize the lens
capabilities (or rather reveal its shortcomings).

Ultimately, it's the (printed) image that must be compared for a real
judgement, but the MTF can predict what the limits are. There are also
other lens parameters than resolution alone, that help to produce
technically good looking images. And then there's the photographer...

Bart

Archived from groups: rec.photo.digital.slr-systems,rec.photo.digital (More info?)

> Lenses generally are capable of much higher resolution than the
> sensor, although contrast may be low. So what you really need to know is
> what the combination of lens+AA-filter+sensor brings in terms of contrast
> as a function of spatial frequency (=system MTF).

Good post, Bart. A chain breaks at weakest link. A system view is very
important when one is searching for the best performance.

Archived from groups: rec.photo.digital.slr-systems,rec.photo.digital (More info?)

> It's not that simple, either. The strength of the weakest link is the
> most important contributor, but the strength of the other links matter.
>
> A lens capable of 75 lp/mm in front of a sensor capable of 50 lp/mm will
> deliver something closer to 30 lp/mm (1 / (1/50 + 1/75) ) than 50 lp/mm

Yes, but the above sensor limit is the weakest link; using your example:
put a 200 lp/mm lens in front and you get 40 lp/mm. Put an infinitely
resolved lens in front of it, and you will get very close to 50 lp/mm (the
weak link). You cannot exceed the weakest link in a chain! The point of my
post was to offer some advice to others so as to help them balance their
purchases. Too many folks seem to think in terms of just one element of a
system.

Archived from groups: rec.photo.digital.slr-systems,rec.photo.digital (More info?)

In article <lLbSd.18757$Tt.3301@fed1read05>,
Steve Gavette <sgavette@no.cox.spam.net.4me> wrote:
>
>"John Francis" <johnf@panix.com> wrote in message
>news:cvb3fv$4v$1@reader2.panix.com...
>> In article <fNKdnQrq59-WmYTfRVn-1g@comcast.com>,
>> Charles Schuler <charleschuler@comcast.net> wrote:
>> >
>> >> Lenses generally are capable of much higher resolution than the
>> >> sensor, although contrast may be low. So what you really need to know
>is
>> >> what the combination of lens+AA-filter+sensor brings in terms of
>contrast
>> >> as a function of spatial frequency (=system MTF).
>> >
>> >Good post, Bart. A chain breaks at weakest link. A system view is very
>> >important when one is searching for the best performance.
>>
>> It's not that simple, either. The strength of the weakest link is the
>> most important contributor, but the strength of the other links matter.
>>
>> A lens capable of 75 lp/mm in front of a sensor capable of 50 lp/mm will
>> deliver something closer to 30 lp/mm (1 / (1/50 + 1/75) ) than 50 lp/mm
>
>I'm new to this and had a question about "resolution capabitity". I have
>seen MTF graphs on several lenses, and few provide much more than 50%
>contrast at 40 l/mm. At 75 l/mm contrast would be very low. So, by "capable
>of 75 l/mm", do you mean any contrast at all? Thanks.

You're getting things backwards. A lens that is capable of resolving
40 lp/mm (that 'p' is important) with a medium-contrast target would
be able to resolve considerably more with a higher-contrast target.
Some tests are carried out with a target that has 1000:1 brightness
ratio between the dark and light parts of the target.
(And, in any case, it was just a number used as an example)

Archived from groups: rec.photo.digital.slr-systems,rec.photo.digital (More info?)

In article <ggeSd.18779$Tt.11892@fed1read05>,
Steve Gavette <sgavette@no.cox.spam.net.4me> wrote:
>
>"John Francis" <johnf@panix.com> wrote in message
>news:cvbj6t$k43$1@reader2.panix.com...
>> In article <lLbSd.18757$Tt.3301@fed1read05>,
>> Steve Gavette <sgavette@no.cox.spam.net.4me> wrote:
>> >
>> >"John Francis" <johnf@panix.com> wrote in message
>> >news:cvb3fv$4v$1@reader2.panix.com...
>> >> In article <fNKdnQrq59-WmYTfRVn-1g@comcast.com>,
>> >> Charles Schuler <charleschuler@comcast.net> wrote:
>> >> >
>> >> >> Lenses generally are capable of much higher resolution than the
>> >> >> sensor, although contrast may be low. So what you really need to
>know
>> >is
>> >> >> what the combination of lens+AA-filter+sensor brings in terms of
>> >contrast
>> >> >> as a function of spatial frequency (=system MTF).
>> >> >
>> >> >Good post, Bart. A chain breaks at weakest link. A system view is
>very
>> >> >important when one is searching for the best performance.
>> >>
>> >> It's not that simple, either. The strength of the weakest link is the
>> >> most important contributor, but the strength of the other links matter.
>> >>
>> >> A lens capable of 75 lp/mm in front of a sensor capable of 50 lp/mm
>will
>> >> deliver something closer to 30 lp/mm (1 / (1/50 + 1/75) ) than 50 lp/mm
>> >
>> >I'm new to this and had a question about "resolution capabitity". I have
>> >seen MTF graphs on several lenses, and few provide much more than 50%
>> >contrast at 40 l/mm. At 75 l/mm contrast would be very low. So, by
>"capable
>> >of 75 l/mm", do you mean any contrast at all? Thanks.
>>
>> You're getting things backwards. A lens that is capable of resolving
>> 40 lp/mm (that 'p' is important) with a medium-contrast target would
>> be able to resolve considerably more with a higher-contrast target.
>> Some tests are carried out with a target that has 1000:1 brightness
>> ratio between the dark and light parts of the target.
>> (And, in any case, it was just a number used as an example)
>
>Sorry, I omitted the "pairs". But what I meant in my question (regardless of
>actual number used, i.e. 75) was in MTF graphs I have seen resolution is
>referenced to a percentage, not an absolute value. I have not seen a graph
>showing maximum line frequency resolved. Would this be the point at which
>you get a solid gray image?

It's usually the point at which the difference between light and dark
falls below some particular threshhold; the case you quote (the point
at which you get a solid grey image, with no discernible differences)
would be the maximal limiting case.

Informally, rather than taking strict MTF measurements, people often
use the simpler test of using a simple alternating black/white target,
at various spatial frequencies, and quoting the highest line density
(at the sensor) which still shows visible banding.
This will be similar to (but slightly higher) than the 10% MTF number.

It's more complicated with digital sensors, as opposed to film, because
you also have the potential problems introduced by aliasing between the
pattern frequency and the grid frequency of the sensor elements.

Archived from groups: rec.photo.digital.slr-systems,rec.photo.digital (More info?)

"Dave Martindale" <davem@cs.ubc.ca> wrote in message
news:cvdc97$m1f$3@mughi.cs.ubc.ca...
> "Steve Gavette" <sgavette@no.cox.spam.net.4me> writes:
>
> >I'm new to this and had a question about "resolution capabitity". I have
> >seen MTF graphs on several lenses, and few provide much more than 50%
> >contrast at 40 l/mm. At 75 l/mm contrast would be very low. So, by
"capable
> >of 75 l/mm", do you mean any contrast at all? Thanks.
>
> There is some objective limit, but it's quite low. "Capable of
> resolving 75 lp/mm" may well mean that the MTF drops to 1% at that
> frequency. That's at the threshold of visibility with a 100:1 contrast
> test target, and below visibility with normal subjects.
>
> Dave

Thanks John and Dave for your replies. I think I've got it.