[pics] SD10 vs. 20d sample

Started Feb 21, 2005 | Discussions
DaSonyGuy Forum Pro • Posts: 12,343
Re: Experiment - (correct me if I'm wrong DaSigmaGuy)

Jafalt wrote:

Lets said that there is a Sigma SD10b camera only producing the
3024*4536 interpoled tiff images on the flash card.

How would then compare this image to a 20D (already out of camera)
Canon-interpolated 8mb image In your test? Resize the Canon to the
double of the interpolated Sigma size?

No, upsize the 20D shot to the same dimensions as the double size SD10 shot out of SPP,...13.7mp(?)

actual Sigma by the software deliveres two format 1) this
Sigma-interpolated 3024*4536 and 2) the non-interpolated ultra
sharp 2268*1512

All 20D images are interpolated so comparing one with an interpolated SD10 shot should be a fairer test, yes?

Regards

DSG

-- hide signature --
 DaSonyGuy's gear list:DaSonyGuy's gear list
Fujifilm FinePix F660EXR Sigma SD14 Sony a7C Huawei P30 Pro +25 more
Jafalt Regular Member • Posts: 161
Experiment - And then you juri D

Lets said that there is a Sigma SD10b camera only producing the 3024*4536 interpoled tiff images on the flash card.

How would then compare this image to a 20D (already out of camera) Canon-interpolated 8mb image In your test? Resize the Canon to the double of the interpolated Sigma size?

actual Sigma by the software deliveres two format 1) this Sigma-interpolated 3024*4536 and 2) the non-interpolated ultra sharp 2268*1512

The new comming polaroid foveon do not (as fare as I have heard) set the ultrasharp as the primary format but let the interpolated format be default. That is also why they claime it is 4.5 pixel (and not 1.5 ultrasharp), Canon would do the same.

Jac

SigmaSD9 Senior Member • Posts: 2,013
Adopting existing terminolgy

Jafalt wrote:

The new comming polaroid foveon do not (as fare as I have heard)
set the ultrasharp as the primary format but let the interpolated
format be default. That is also why they claime it is 4.5 pixel
(and not 1.5 ultrasharp), Canon would do the same.

It's great to see people understanding this issue. I suggest we use the terms "optical" and "interpolated" resolutions in the same way as the scanner industry, whose buyers all intuitively undersood this issue virtually instantly back in the 90s when this same issue came up.

Optical Resolution = Device has a dedicated RGB-triple for every recorded pixel
Interpolated Resolution = Any number of recorded pixels more than optical.

So for example...
SD9 = 3.4MP-optical; 10.3MP-interpolated
1DMKII = 2.1MP-optical; 8.4MP-interpolated

The interpolated number could, of course, be anything, so I suggest we stick with Foveon's philosphy of using the common Bayer standard of listing 1 recorded pixel per 1 sensor (all except Fuji who often advertise MPs at 2 recorded pixels per 1 sensor).

Juri D Regular Member • Posts: 428
Re: Experiment - And then you juri D

Jafalt wrote:

Lets said that there is a Sigma SD10b camera only producing the
3024*4536 interpoled tiff images on the flash card.

How would then compare this image to a 20D (already out of camera)
Canon-interpolated 8mb image In your test? Resize the Canon to the
double of the interpolated Sigma size?

actual Sigma by the software deliveres two format 1) this
Sigma-interpolated 3024*4536 and 2) the non-interpolated ultra
sharp 2268*1512

Interpolating 20D to 13.5MP seems natural here. By the way I find interpolation useful on two counts: (1) some flaws cannot be visible at 100%, and interpolation blows them up, and (2) because usually print is the final destination for an image, it is useful to see images how the printer sees them: namely, the printer interpolates whatever comes its way to its native resolution [see discussions on interpolation and , especially, Qimage on Printer Forum].

SigmaSD9 Senior Member • Posts: 2,013
Re: Experiment - And then you juri D

Juri D wrote:

Interpolating 20D to 13.5MP seems natural here. By the way I find
interpolation useful on two counts: (1) some flaws cannot be
visible at 100%, and interpolation blows them up, and (2) because
usually print is the final destination for an image, it is useful
to see images how the printer sees them: namely, the printer
interpolates whatever comes its way to its native resolution [see
discussions on interpolation and , especially, Qimage on Printer
Forum].

Some printers do, some don't. Laser generated silver halide photographic emulsions won't care either way, and when/if those printers do interpolate to smooth extreme print sizes, it is on the order of 4000+ dpi. I'm not sure why we'd spend money and time worrying about image quality only to use a home printer.

Juri D Regular Member • Posts: 428
Re: Experiment - And then you juri D

SigmaSD9 wrote:

Some printers do, some don't. Laser generated silver halide
photographic emulsions won't care either way, and when/if those
printers do interpolate to smooth extreme print sizes, it is on the
order of 4000+ dpi. I'm not sure why we'd spend money and time
worrying about image quality only to use a home printer.

Go to Printer Forum and you may be surprised; there many myths @ printers to be debumked (in particular, about 5600dpi resolution on inkjets), about relative quality of various printing processes. A $200 home printer, properly handled, can give you spectacular quality.

Juri D Regular Member • Posts: 428
dr.noise, it may find it interesting...

dr.noise wrote:

Look at the star circle (europe symbol) at the tesla box - next to
Energy word. Sigma sees the stars as simply blurred dots. There
isn't enough resolution. But so does Canon. Same blurred dots
instead of stars.

your star cirle is discernible on 20D shot [I had to resize it of course, this time with Lanczos]

AMomcil Senior Member • Posts: 1,018
Close..

Juri D wrote:

It's really simple. Let's say your printer's native resolution is
600 dpi (as in HP inkjets). You are printing 10x15. This means that
the printer will upsize whatever you send to it to 6000x9000
pixels. It does mean different upsizing factors for SD10 (4x) and
20D (2.5x). That's what I am simulating.

You are on a good track, but forgot about colour separation. Printer driver first have to do a colour separation, e.g. CMYK, because printer can print only different intensity of colours from its cartidges. Stated resolution (e.g. 600 dpi) is describing the resolution of those primary colour dots, not fully coloured pixels. Therefore is dpi of Ink jets a bit missleading.
Better would be to use lpi (lines per inch), which is for professional printers.

But, you're right, image will be rescaled upwards to printers native resoution.

-- hide signature --

Greetings, Aleksandar

 AMomcil's gear list:AMomcil's gear list
Sony RX100
ghoerdt Senior Member • Posts: 1,250
Re: Close..

But, you're right, image will be rescaled upwards to printers
native resoution.

Not exactly if speaking of inkjet-printers.

The printerdriver separates the input to CMYK (or more colours, depending on printer) and produces rasterized data from that (a frequency modulated pattern normally). It then sends the pattern data to the printer. This has nothing to do with upscaling to "native resolution" wich btw refers to the positioning accuracy of the printheads. Printed resolution depends on the size of ink drops and the raster algorythms used and is MUCH lower than advertized.

Greetings
Günter

SigmaSD9 Senior Member • Posts: 2,013
Re: Experiment - And then you juri D

Juri D wrote:

SigmaSD9 wrote:

Some printers do, some don't. Laser generated silver halide
photographic emulsions won't care either way, and when/if those
printers do interpolate to smooth extreme print sizes, it is on the
order of 4000+ dpi. I'm not sure why we'd spend money and time
worrying about image quality only to use a home printer.

Go to Printer Forum and you may be surprised; there many myths @
printers to be debumked (in particular, about 5600dpi resolution on
inkjets), about relative quality of various printing processes. A
$200 home printer, properly handled, can give you spectacular
quality.

It can give you spectacular quality compared to older home printers, but it's never going to expose then develop a true photograph for you.

AMomcil Senior Member • Posts: 1,018
Repeating

ghoerdt wrote:

Not exactly if speaking of inkjet-printers.

The printerdriver separates the input to CMYK (or more colours,
depending on printer) and produces rasterized data from that (a
frequency modulated pattern normally). It then sends the pattern
data to the printer. This has nothing to do with upscaling to
"native resolution" wich btw refers to the positioning accuracy of
the printheads. Printed resolution depends on the size of ink drops
and the raster algorythms used and is MUCH lower than advertized.

Günter, you just repeated what I said, just in other words. Printers are not printing pixels we see on the monitor, but smaller particle of different size and colors that looks like it from the distance.
The point is that image size, pixelwise, is changed.

-- hide signature --

Greetings, Aleksandar

 AMomcil's gear list:AMomcil's gear list
Sony RX100
PaulyOly Veteran Member • Posts: 7,723
Re: Experiment - (correct me if I'm wrong DaSigmaGuy)

DaSigmaGuy wrote:

Jafalt wrote:

Lets said that there is a Sigma SD10b camera only producing the
3024*4536 interpoled tiff images on the flash card.

How would then compare this image to a 20D (already out of camera)
Canon-interpolated 8mb image In your test? Resize the Canon to the
double of the interpolated Sigma size?

No, upsize the 20D shot to the same dimensions as the double size
SD10 shot out of SPP,...13.7mp(?)

actual Sigma by the software deliveres two format 1) this
Sigma-interpolated 3024*4536 and 2) the non-interpolated ultra
sharp 2268*1512

All 20D images are interpolated so comparing one with an
interpolated SD10 shot should be a fairer test, yes?

down sampling the native interpolated image from the sigma to the 20d native interpolated image should give the sigma an advantage, no? I compared the first two images (3504x2336) from the orginal post and after sharpening them both the same I see no sharpness advantage that some here claim the sigma has and the 20d image has more detail, not only in the backdrop, but also looking at the dogs nose, the sigma image had some jaggies on the watch clasp and the baileys label, the canon image was smooth. As far as this shot goes there should be little difference in sharpness at any print size.

To be fair this test should be done with the same lens, but one would think the sigma lens would have an advantage at f11 since it's a macro and designed to work best at large apertures.

Why not an outdoor test? I'm sure someone can test it with the same lens, either put a sigma on the canon or one of the canon convert lenses on the sigma.

How about an iso400 test? :^)

Regards

DSG

-- hide signature --
-- hide signature --

Narrow depth of field ahead
Use extreme caution

http://www.pbase.com/paulyoly/root

Juri D Regular Member • Posts: 428
you are both correct [partially]...

AMomcil wrote:

The printerdriver separates the input to CMYK (or more colours,
depending on printer) and produces rasterized data from that (a
frequency modulated pattern normally). It then sends the pattern
data to the printer. This has nothing to do with upscaling to
"native resolution" wich btw refers to the positioning accuracy of
the printheads. Printed resolution depends on the size of ink drops
and the raster algorythms used and is MUCH lower than advertized.

It does, of course, everything you said, but it does also something extra. By "native resolution" I mean the positioning accuracy of printhead. It is usually limited at 600-720dpi, even though you think you are printing 5600dpi. A good discussion of that is on Qimage website:

http://www.ddisoftware.com/qimage/quality/

as well as on Printers Forum of DPREVIEW.

Juri D Regular Member • Posts: 428
Re: Experiment - And then you juri D

SigmaSD9 wrote:

It can give you spectacular quality compared to older home
printers, but it's never going to expose then develop a true
photograph for you.

It does it for me every time!

PaulyOly Veteran Member • Posts: 7,723
I sharpened your jpegs........

People here are going to see what they want to see no matter what, but I figured I'd sharpen your jpgs and post it anyway. Unfortunately I can't download the raw photos due to my slow dialup so I had to go with your processing. I also don't do much raw processing anyway.

http://www.pbase.com/paulyoly/image/40101464

I cropped the middle from each image and merged them, then sharpened the combined image using a two layer process.

My conclusion is simple, the image quality is very similar, the sigma shot shows much more sharpening halos, the 20d shot could have taken even more sharpening. There are a few edges in the sigma shot that are jagged that are not jagged in the 20d shot, i'm sure some expert will explain this (sigmasd9?). The sigma seems to have more contrast, i doubt this is the lens since sigmasd9 seems to praise the canon so highly so it's either the sensor or raw processing settings. Something I find strange is the difference in the white sheet, the strands or texture looks larger in the sigma shot, is this from upsampling?

Things you can't see in this test are the iso performance of the canon, the great buffer and 5fps (which is just plain fun, but unecessary for most shots.) I see no advantage for the sigma in this test.

I'd like to see both shot with the sigma lens or canon, but of course I don't expect you to go out and buy a lens. I appreciate the comparison, it was interesting.
--

Narrow depth of field ahead
Use extreme caution

http://www.pbase.com/paulyoly/root

PaulyOly Veteran Member • Posts: 7,723
I also took crops........

from the combined image so that I could keep the file size relatively small and save them at max quality, i then merged the crops, the full size merged image was huge.

PaulyOly wrote:

People here are going to see what they want to see no matter what,
but I figured I'd sharpen your jpgs and post it anyway.
Unfortunately I can't download the raw photos due to my slow dialup
so I had to go with your processing. I also don't do much raw
processing anyway.

http://www.pbase.com/paulyoly/image/40101464

I cropped the middle from each image and merged them, then
sharpened the combined image using a two layer process.

My conclusion is simple, the image quality is very similar, the
sigma shot shows much more sharpening halos, the 20d shot could
have taken even more sharpening. There are a few edges in the
sigma shot that are jagged that are not jagged in the 20d shot, i'm
sure some expert will explain this (sigmasd9?). The sigma seems to
have more contrast, i doubt this is the lens since sigmasd9 seems
to praise the canon so highly so it's either the sensor or raw
processing settings. Something I find strange is the difference in
the white sheet, the strands or texture looks larger in the sigma
shot, is this from upsampling?

Things you can't see in this test are the iso performance of the
canon, the great buffer and 5fps (which is just plain fun, but
unecessary for most shots.) I see no advantage for the sigma in
this test.

I'd like to see both shot with the sigma lens or canon, but of
course I don't expect you to go out and buy a lens. I appreciate
the comparison, it was interesting.
--

Narrow depth of field ahead
Use extreme caution

http://www.pbase.com/paulyoly/root

-- hide signature --

Narrow depth of field ahead
Use extreme caution

http://www.pbase.com/paulyoly/root

PaulyOly Veteran Member • Posts: 7,723
yellow tint question.........

I know there has been some talk about a yellow tint to the sigma image from the lens. I noticed that the baileys name on the bottle is more white in the canon image than the sigma image, which is correct?
--

Narrow depth of field ahead
Use extreme caution

http://www.pbase.com/paulyoly/root

Peter G Veteran Member • Posts: 5,944
Re: I sharpened your jpegs........

PaulyOly wrote:

processing settings. Something I find strange is the difference in
the white sheet, the strands or texture looks larger in the sigma
shot, is this from upsampling?

No it is from Aliasing. Lack of an AA filter. It is the same phenomena that produces 5 lines from 9 on the res chart in the review. As the details get smaller the actual ouput gets bigger. This gives the impression that it is show more detail when in fact it is obscuring the real detail and giving a bigger false pattern.

dr.noise
OP dr.noise Veteran Member • Posts: 3,767
Re: dr.noise, it may find it interesting...

Juri D wrote:

your star cirle is discernible on 20D shot [I had to resize it of
course, this time with Lanczos]

It is almost as much discernible on Sigma shot. I call it blurred dots anyway since the details are going to extinct and only my knowledge of real subject allows to "see" the stars.

There can be slight DOF variation which affects such ultra-fine details. For absolute sensor resolution test it should be center of the frame and flat subject so the DOF will not be an issue.

 dr.noise's gear list:dr.noise's gear list
Sigma DP2s Sony DSC-RX0 Sigma SD10 Sigma fp
Helge Hafting Contributing Member • Posts: 827
Re: I sharpened your jpegs........

Peter G wrote:

PaulyOly wrote:

processing settings. Something I find strange is the difference in
the white sheet, the strands or texture looks larger in the sigma
shot, is this from upsampling?

No it is from Aliasing. Lack of an AA filter. It is the same
phenomena that produces 5 lines from 9 on the res chart in the
review.

Correct, but I wouldn't call it "lacking" for not having an AA filter. Bayer cameras don't have AA filters to blur out details, they have the AA filter for avoiding unsightly color aliasing. You will always get something wrong when taking pictures of details too small for the sensor. The "smudge it out" approach of an AA filter is no more "correct" than the aliasing we get without.

Either way, we get an image without the details the camera couldn't capture, and then the interesting question is "what looks best". For the cloth, the aliasing still looked like cloth and was therefore a good thing. The AA-filtered image simply looks unsharp. Of course one can find cases where blur look better than aliasing too, but those are rare. Most users of AA cameras fight the filter with unsharp masking.

As the details get smaller the actual ouput gets bigger.
This gives the impression that it is show more detail when in fact
it is obscuring the real detail and giving a bigger false pattern.

The false pattern is indeed false, but it does rarely obscure a real pattern you cannot capture anyway. The false blurring of an AA filter is often the bigger problem.

Sometimes people with a background in audio shows off formulas that "prove" why an AA filter of a certain strength is a good thing - or even necessary. But they work from a wrong assumption, the assumption that any false signal is very bad. This assumption holds for audio, where any false detail really kills the music. But the assumption rarely holds for vision, except in a few pathological cases where a chessboard pattern positioned "just so" turns into a set of broad stripes. Patterns a bit less regular (such as gravel, dirt and foliage) tends to alias into slightly coarser but similar patterns that are much more pleasing to the eye than the "mathematically correct" area of almost uniform color. Areas of uniform color is particularly bad when sharpening is applied later - you get a sharp line between the sky and the distant forest, but at such sharpness the eye really expect some approximation of distant trees and branches, instead of solid green.

As for the resolution chart, the sigma is clearly not showing a "true" image when it merges nine lines into seven and then into five. But at least it shows a set of black and white lines, where AA-equipped cameras shows a single column of gray. The sigma photo carries more information - "there is lines" where the AA photo merely says "don't know - too high resolution for me". There was no gray in the original resolution chart, so the gray column is "wrong".

Helge Hafting

Keyboard shortcuts:
FForum MMy threads