Sign in with
Sign up | Sign in
Your question

Is there any graph that shows lenses versus sensors?

Last response: in Digital Camera
Share
Anonymous
August 11, 2005 7:36:51 PM

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

That is, is there anything that says at what
point sensor resolution outstrips current lenses
in terms of what they are capable of delivering?
Lenses have theoretical limits to resolving power,
but sensors don't, as far as I know.
-Rich
Anonymous
August 12, 2005 5:02:43 AM

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

"RichA" <none@none.com> wrote in message
news:a3anf1d4l0c621651dk8u7e9hi2r40apk2@4ax.com...
> That is, is there anything that says at what
> point sensor resolution outstrips current lenses
> in terms of what they are capable of delivering?
> Lenses have theoretical limits to resolving power,
> but sensors don't, as far as I know.

The theoretical resolution limit is determined by the sensor array's
sampling density or sensor pitch. A sensor array with a sensor pitch
of e.g. 6.4 micron (like in an EOS-20D) takes samples every 6.4
microns. To reliably resolve a small feature, it takes 2 sensors
(Nyquist limit). That means that this sensor can resolve 78.1 cycles
or line-pairs per millimetre.

The contrast towards that theoretical resolution is somewhat modified
by the Anti-Aliasing filter that is used, but tests will show that in
the case of the 20D the 78.1 cy/mm still holds, albeit with reduced
modulation in order to reduce aliasing artifacts.

Bart
Anonymous
August 12, 2005 5:36:30 AM

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

RichA wrote:
> That is, is there anything that says at what
> point sensor resolution outstrips current lenses
> in terms of what they are capable of delivering?
> Lenses have theoretical limits to resolving power,
> but sensors don't, as far as I know.
> -Rich

For Canon L series lenses, it's probably around 25 megapixels. This is
based on Linepairs Per Milimeter (lp/mm) comparisons between the best
film (slow speed slide film) which reolves around 75 lp/mm and
extrapolating the resolution up from the Canon 1Ds Mark II which
resolves aroung 60 lp/mm.

It's certainly possible that some lenses are capable of resolving well
beyond the 75 lp/mm of the best slide film, so the limit of sensor
resolution could be even higher, but the lenses had no way of really
being tested beyond the limits of the best film.

One table that may be of interest to you can be found at:

"http://www.katharos.org/david/index.php?option=com_cont..."

Lp/mm is one more reason why larger sensors make sense as resolutions
increase. The Olympus E300 for sure, and probably the Nikon D2x, are
already exceeding the resolving power of many of the lenses that work
with those cameras, not because of the resolution, but because of the
lp/mm that are a result of the small sensor size.

This is an interesting subject, and I'll probably add lp/mm to the web
site tables.

Steve
http://digitalslrinfo.com
Related resources
Anonymous
August 12, 2005 6:39:38 AM

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

On Fri, 12 Aug 2005 01:36:30 GMT, SMS <scharf.steven@geemail.com>
wrote:

>RichA wrote:
>> That is, is there anything that says at what
>> point sensor resolution outstrips current lenses
>> in terms of what they are capable of delivering?
>> Lenses have theoretical limits to resolving power,
>> but sensors don't, as far as I know.
>> -Rich
>
>For Canon L series lenses, it's probably around 25 megapixels. This is
>based on Linepairs Per Milimeter (lp/mm) comparisons between the best
>film (slow speed slide film) which reolves around 75 lp/mm and
>extrapolating the resolution up from the Canon 1Ds Mark II which
>resolves aroung 60 lp/mm.
>
>It's certainly possible that some lenses are capable of resolving well
>beyond the 75 lp/mm of the best slide film, so the limit of sensor
>resolution could be even higher, but the lenses had no way of really
>being tested beyond the limits of the best film.
>
>One table that may be of interest to you can be found at:
>
>"http://www.katharos.org/david/index.php?option=com_cont..."
>
>Lp/mm is one more reason why larger sensors make sense as resolutions
>increase. The Olympus E300 for sure, and probably the Nikon D2x, are
>already exceeding the resolving power of many of the lenses that work
>with those cameras, not because of the resolution, but because of the
>lp/mm that are a result of the small sensor size.
>
>This is an interesting subject, and I'll probably add lp/mm to the web
>site tables.
>
>Steve
>http://digitalslrinfo.com

Doesn't this imply that resolution of some of the P&S cameras is
around 200 lpmm? Compared to around 100 lpmm for DSLRs because of the
much smaller sensor sizes of some high pixel count P&S sensors?


"Bittorrents are REFUNDS for all the BAD movie products Hollywood
never gave us refunds for in the past"
Anonymous
August 12, 2005 12:27:31 PM

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

RichA wrote:
[]
> Doesn't this imply that resolution of some of the P&S cameras is
> around 200 lpmm? Compared to around 100 lpmm for DSLRs because of the
> much smaller sensor sizes of some high pixel count P&S sensors?

Yes, in terms of linear resolving power, the lenses for some P&S cameras
will far exceed the resolution of some DSLR lenses.

Shock, horror?

No.

David
Anonymous
August 12, 2005 12:29:48 PM

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

RichA wrote:
> That is, is there anything that says at what
> point sensor resolution outstrips current lenses
> in terms of what they are capable of delivering?
> Lenses have theoretical limits to resolving power,
> but sensors don't, as far as I know.

Sensors do. The limit of the sensor is imposed by Nyquist - an anti-alias
filter is required to ensure that the response of the sensor is zero at
half the element spatial frequency. In practice, some AA filters are
better than others.

David
Anonymous
August 12, 2005 3:08:48 PM

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

RichA wrote:

> Doesn't this imply that resolution of some of the P&S cameras is
> around 200 lpmm? Compared to around 100 lpmm for DSLRs because of the
> much smaller sensor sizes of some high pixel count P&S sensors?

Consumers love megapixels. Even many DSLR buyers look only at
resolution, without looking at pixel size, and those that are smart
enough to look at pixel size are usually doing so only from the
perspective of noise versus ISO, and deciding that they don't need high
ISO capability, they aren't looking at the lpmm. You can look at threads
in this newsgroup regarding the noise issues of some DSLRs.

I think that lpmm is probably one of the reasons that Canon ended up
with three different size sensors for their DSLRs (which probably drives
their marketing group crazy). It's not that engineering didn't
understand the cost benefits of sticking with a smaller size, it's that
it made no sense to have a high resolution digital SLR with a small sensor.

Try explaining lpmm to a customer that understands only megapixels.
Anonymous
August 12, 2005 4:42:26 PM

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

"David J Taylor"
<david-taylor@blueyonder.co.not-this-bit.nor-this-part.uk.invalid>
wrote in message
news:T3ZKe.87229$G8.50189@text.news.blueyonder.co.uk...
SNIP
> Yes, in terms of linear resolving power, the lenses for some P&S
> cameras will far exceed the resolution of some DSLR lenses.
>
> Shock, horror?
>
> No.

Indeed, although some will be shocked. Due to the tiny physical
dimensions of the sensor array the image needs a huge magnification
for output, which will reduce the resolution rapidly (and magnify the
noise and lens aberrations).

I use the same criterion as suggested in the
<http://www.katharos.org/david/index.php?option=com_cont...;
table intro, sensor lp/mm divided by magnificaton (from sensor array
size to output size) should not fall below a certain quality level
(e.g. 8 lp/mm for close viewing), anything less is a compromise.

Bart
Anonymous
August 12, 2005 4:42:27 PM

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

Bart van der Wolf wrote:
> "David J Taylor"
> <david-taylor@blueyonder.co.not-this-bit.nor-this-part.uk.invalid>
> wrote in message
> news:T3ZKe.87229$G8.50189@text.news.blueyonder.co.uk...
> SNIP
>> Yes, in terms of linear resolving power, the lenses for some P&S
>> cameras will far exceed the resolution of some DSLR lenses.
>>
>> Shock, horror?
>>
>> No.
>
> Indeed, although some will be shocked. Due to the tiny physical
> dimensions of the sensor array the image needs a huge magnification
> for output, which will reduce the resolution rapidly (and magnify the
> noise and lens aberrations).
>
> I use the same criterion as suggested in the
> <http://www.katharos.org/david/index.php?option=com_cont...;
> table intro, sensor lp/mm divided by magnificaton (from sensor array
> size to output size) should not fall below a certain quality level
> (e.g. 8 lp/mm for close viewing), anything less is a compromise.
>
> Bart

I think it would be more helpful to take sensor size out of the equation,
and think of line pairs (or cycles) per picture width when trying to
compare DSLRs and P&S, although I also find the idea of representing an
MTF curve with a single value as too great a simplification.

David
Anonymous
August 12, 2005 4:42:29 PM

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

SMS wrote:
[]
> Try explaining lpmm to a customer that understands only megapixels.

Cycles per mm are only a relavant measure if the sensor size is the same.
Cycles per picture width are a more relevant system comparison.

David
Anonymous
August 12, 2005 11:51:28 PM

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

"David J Taylor"
<david-taylor@blueyonder.co.not-this-bit.nor-this-part.uk.invalid>
wrote in message
news:D u%Ke.87354$G8.66013@text.news.blueyonder.co.uk...
SNIP
> I think it would be more helpful to take sensor size out of the
> equation, and think of line pairs (or cycles) per picture width when
> trying to compare DSLRs and P&S, although I also find the idea of
> representing an MTF curve with a single value as too great a
> simplification.

That's almost (I used LP/PH) exactly what I posted some time ago,
<http://www.xs4all.nl/~bvdwolf/main/foto/Imatest/SFR_Gra...;, but
that needs an understanding of how such a curve translates to real
life images. Also, MTF doesn't tell the whole image quality story
although it does describe 'resolution' and contrast better than a
single number ever could.

Bart
!