«

No difference in sharpness or anything else discernible at high
magnification (about 16x20in).
One's 10MP, the other 6MP and 5-year old technology.
Used a Nikkor 24mm f2.8 AF vs a Canon 24mm f2.8 EF.
Both cameras on tripods, ISO 200 in daylight.

Comments?

On Jan 22, 1:09*pm, Bob G <mrbobja...@yahoo.com> wrote:
> No difference in sharpness or anything else discernible at high
> magnification (about 16x20in).
> One's 10MP, the other 6MP and 5-year old technology.
> Used a Nikkor 24mm f2.8 AF vs a Canon 24mm f2.8 EF.
> Both cameras on tripods, ISO 200 in daylight.
>
> Comments?

Perhaps you could post some pictures for us to compare.

Bob G wrote:
> No difference in sharpness or anything else discernible at high
> magnification (about 16x20in).
> One's 10MP, the other 6MP and 5-year old technology.
> Used a Nikkor 24mm f2.8 AF vs a Canon 24mm f2.8 EF.
> Both cameras on tripods, ISO 200 in daylight.
>
> Comments?

Yes. Using the same professional quality lens with two camera bodies
(6mp and 12mp) taking exposures (raw) of the same scene at about the
same time using a tripod, bracketing to get exposure spot on between
cameras, carefully adjusting the raw files using the same raw converter,
then printing at 18x12", I can see the difference between 6mp and 12mp.
(even if most non-photographers probably wouldn't unless prompted to
what to look for).
In that case, the most obvious difference isn't detail resolved (though
some small improvement is surely visible if you look close), but that in
fine edge detail with 6mp, aliasing can be visible, and that flaw tends
to draw the attention of the eye (mine anyway). In that case I used a
D70 and D300. The difference may be more obvious than in your test, as
the D70 has a very weak AA filter resulting in visible aliasing (if you
have reasonably good lenses/technique), the 300d having (from what I've
seen) a good compromise of AA filter.

Your result doesn't surprise me.

Bob G wrote:
> No difference in sharpness or anything else discernible at high
> magnification (about 16x20in).
> One's 10MP, the other 6MP and 5-year old technology.
> Used a Nikkor 24mm f2.8 AF vs a Canon 24mm f2.8 EF.
> Both cameras on tripods, ISO 200 in daylight.
>
> Comments?

You're right, there's little difference between 6MP and 10 or 12 MP
cameras as far as image quality goes. The big difference is speed, of
starting up, frames/sec, number of shots to fill the buffer, of writing
to the card, etc.

I still use a 300D (as a retiree I am a bit impecunious), but I compete
with my 300D at camera club against owners of D300, 5D, 40D, and others.

Being a bit of a techo, I utilize the best software for post-processing;
shoot RAW, process through DxO Optics Pro, adjust the image as necessary
in Photoshop, and print via Qimage and a Canon 9950 printer. Little
things like never resampling in Photoshop, and using Qimage to do a
once-only resample direct to the printer's native resolution retains the
sharpest artifact-free image possible. The result is the humble 300D
can give the bigger cameras a good fright in terms of image quality.

It's not generally well known that most images suffer a double
resampling - once in Photoshop to get the 'magic' 300 ppi, and again in
the printer driver, resampling up to the printer's native resolution,
600 ppi for Canon printers, and 720 ppi for Epson. The double
resampling alone is guaranteed to take the edge off any image.

Added to that is the resampling quality of the printer drivers is
unknown, and Photoshop resampling is good but not the best. Quimage uses
the best resampling techniques available.

Colin D.

Colin.D wrote:

> It's not generally well known that most images suffer a double
> resampling - once in Photoshop to get the 'magic' 300 ppi, and again in
> the printer driver, resampling up to the printer's native resolution,
> 600 ppi for Canon printers, and 720 ppi for Epson. The double
> resampling alone is guaranteed to take the edge off any image.
>
> Added to that is the resampling quality of the printer drivers is
> unknown, and Photoshop resampling is good but not the best. Quimage uses
> the best resampling techniques available.

This has been so often repeated that it's taken for gospel. I'd love
someone to give just one cite from Canon or Epson in the last two years
that indicates there's a magic number of ppi for input. Of course, there
are minimums as well as maximums for ppi's that'll produce decent to
excellent prints.

And if Qimage resampling was superior three years ago, PS has changed
and improved sampling with each iteration of the CS series, so that
proof of that claim is also lacking. (Save for the Qimage site and parrots)

--
John McWilliams

John McWilliams wrote:
> Colin.D wrote:
>
>> It's not generally well known that most images suffer a double
>> resampling - once in Photoshop to get the 'magic' 300 ppi, and again
>> in the printer driver, resampling up to the printer's native
>> resolution, 600 ppi for Canon printers, and 720 ppi for Epson. The
>> double resampling alone is guaranteed to take the edge off any image.
>>
>> Added to that is the resampling quality of the printer drivers is
>> unknown, and Photoshop resampling is good but not the best. Quimage
>> uses the best resampling techniques available.
>
> This has been so often repeated that it's taken for gospel. I'd love
> someone to give just one cite from Canon or Epson in the last two years
> that indicates there's a magic number of ppi for input.
I doubt that you'll find that - but native dpi resolutions are divisible
by 300dpi and 360 respectively, and if you play with the printers by
feeding them test images resampled to those sizes, there's a very slight
but perceptible (just - to the naked eye) improvement. So as
sharpening should be the last step before printing, resampling to
printer native resolution then sharpening might make sense for a known
print size.
That said, I don't bother these days - but I did when using a 6mp dslr
(D70) where for large prints resampling to larger size and checking
carefully with selective sharpening applied seemed to be well worth the
effort to avoid exacerbating aliasing artifacts, but that's almost a
different subject entirely.
> Of course, there
> are minimums as well as maximums for ppi's that'll produce decent to
> excellent prints.
>
> And if Qimage resampling was superior three years ago, PS has changed
> and improved sampling with each iteration of the CS series, so that
> proof of that claim is also lacking. (Save for the Qimage site and parrots)
>

Me wrote:
> John McWilliams wrote:
>> Colin.D wrote:
>>
>>> It's not generally well known that most images suffer a double
>>> resampling - once in Photoshop to get the 'magic' 300 ppi, and again
>>> in the printer driver, resampling up to the printer's native
>>> resolution, 600 ppi for Canon printers, and 720 ppi for Epson. The
>>> double resampling alone is guaranteed to take the edge off any image.
>>>
>>> Added to that is the resampling quality of the printer drivers is
>>> unknown, and Photoshop resampling is good but not the best. Quimage
>>> uses the best resampling techniques available.
>>
>> This has been so often repeated that it's taken for gospel. I'd love
>> someone to give just one cite from Canon or Epson in the last two
>> years that indicates there's a magic number of ppi for input.
> I doubt that you'll find that - but native dpi resolutions are divisible
> by 300dpi and 360 respectively, and if you play with the printers by
> feeding them test images resampled to those sizes, there's a very slight
> but perceptible (just - to the naked eye) improvement. So as
> sharpening should be the last step before printing, resampling to
> printer native resolution then sharpening might make sense for a known
> print size.

Right. I resample to 300dpi & sharpen there choosing an appropriate
radius at that size. It seems to me printing loses lots of detail so
anything the printer wants to do from there doesn't matter.


> That said, I don't bother these days - but I did when using a 6mp dslr
> (D70) where for large prints resampling to larger size and checking
> carefully with selective sharpening applied seemed to be well worth the
> effort to avoid exacerbating aliasing artifacts, but that's almost a
> different subject entirely.
>> Of course, there are minimums as well as maximums for ppi's that'll
>> produce decent to excellent prints.
>>
>> And if Qimage resampling was superior three years ago, PS has changed
>> and improved sampling with each iteration of the CS series, so that
>> proof of that claim is also lacking. (Save for the Qimage site and
>> parrots)
>>
>


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

all google groups messages filtered due to spam

Paul Furman wrote:
> Me wrote:
>> John McWilliams wrote:
>>> Colin.D wrote:
>>>
>>>> It's not generally well known that most images suffer a double
>>>> resampling - once in Photoshop to get the 'magic' 300 ppi, and again
>>>> in the printer driver, resampling up to the printer's native
>>>> resolution, 600 ppi for Canon printers, and 720 ppi for Epson. The
>>>> double resampling alone is guaranteed to take the edge off any image.
>>>>
>>>> Added to that is the resampling quality of the printer drivers is
>>>> unknown, and Photoshop resampling is good but not the best. Quimage
>>>> uses the best resampling techniques available.
>>>
>>> This has been so often repeated that it's taken for gospel. I'd love
>>> someone to give just one cite from Canon or Epson in the last two
>>> years that indicates there's a magic number of ppi for input.

>> I doubt that you'll find that - but native dpi resolutions are
>> divisible by 300dpi and 360 respectively, and if you play with the

Still, a shibboleth. And it's ppi in the image, not dpi.

>> printers by feeding them test images resampled to those sizes, there's
>> a very slight but perceptible (just - to the naked eye) improvement.
>> So as sharpening should be the last step before printing, resampling
>> to printer native resolution then sharpening might make sense for a
>> known print size.
>
> Right. I resample to 300dpi & sharpen there choosing an appropriate
> radius at that size. It seems to me printing loses lots of detail so
> anything the printer wants to do from there doesn't matter.


You're resampling to 300 ppi, and then applying sharpening, which is the
correct
order, but to test parallel, the correct amount of sharpening needs to
be applied to the, say 213 ppi image, to compare it to the upsampled
image at 300 ppi and also correctly sharpened. And do it with a hq
modern printer!

--
John McWilliams

John McWilliams wrote:
> Paul Furman wrote:
>> Me wrote:
>>> John McWilliams wrote:
>>>> Colin.D wrote:
>>>>
>>>>> It's not generally well known that most images suffer a double
>>>>> resampling - once in Photoshop to get the 'magic' 300 ppi, and
>>>>> again in the printer driver, resampling up to the printer's native
>>>>> resolution, 600 ppi for Canon printers, and 720 ppi for Epson. The
>>>>> double resampling alone is guaranteed to take the edge off any image.
>>>>>
>>>>> Added to that is the resampling quality of the printer drivers is
>>>>> unknown, and Photoshop resampling is good but not the best. Quimage
>>>>> uses the best resampling techniques available.
>>>>
>>>> This has been so often repeated that it's taken for gospel. I'd love
>>>> someone to give just one cite from Canon or Epson in the last two
>>>> years that indicates there's a magic number of ppi for input.
>
>>> I doubt that you'll find that - but native dpi resolutions are
>>> divisible by 300dpi and 360 respectively, and if you play with the
>
> Still, a shibboleth. And it's ppi in the image, not dpi.

And it's dpi in the printer, not in the image. Did you have a point to
make?
>
>>> printers by feeding them test images resampled to those sizes,
>>> there's a very slight but perceptible (just - to the naked eye)
>>> improvement. So as sharpening should be the last step before
>>> printing, resampling to printer native resolution then sharpening
>>> might make sense for a known print size.
>>
>> Right. I resample to 300dpi & sharpen there choosing an appropriate
>> radius at that size. It seems to me printing loses lots of detail so
>> anything the printer wants to do from there doesn't matter.
>
>
> You're resampling to 300 ppi, and then applying sharpening, which is the
> correct
> order, but to test parallel, the correct amount of sharpening needs to
> be applied to the, say 213 ppi image, to compare it to the upsampled
> image at 300 ppi and also correctly sharpened. And do it with a hq
> modern printer!
>
Done. My tests with images and resolution targets did show a small
advantage to resizing image ppi to a divisor or printer native dpi
resolution. (Epson 360dpi) I also commented that it's a small
advantage, so small that I don't generally bother.

"Colin.D" <(E-Mail Removed)> wrote in message
news:glaps3$sm7$(E-Mail Removed)

[...]

> It's not generally well known that most images suffer a double
> resampling - once in Photoshop to get the 'magic' 300 ppi, and again

When I resize to 300ppi in PS I deselect the "Resample image" checkbox.