[pics] SD10 vs. 20d sample

Started Feb 21, 2005 | Discussions
PaulyOly Veteran Member • Posts: 7,723
No need to argue which is correct.......

Helge Hafting wrote:

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.

but this test doesn't show an advantage of no AA filter judging by the sigma shot.

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

-- hide signature --

Narrow depth of field ahead
Use extreme caution

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

PaulyOly Veteran Member • Posts: 7,723
Oh now you want to compare..........

SigmaSD9 wrote:

Jafalt wrote:

Why not resize the Canon to double Sigma size?
That would be more fair

Or better still, since the Sigma is showing a bit more detail, size
them both to 3.43MP. You won't lose resolution.

a downsampled 20d image, make up your mind. I see no more detail.

-- hide signature --

Narrow depth of field ahead
Use extreme caution

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

dr.noise
OP dr.noise Veteran Member • Posts: 3,767
Re: No need to argue which is correct.......

PaulyOly wrote:

but this test doesn't show an advantage of no AA filter judging by
the sigma shot.

You forgot one little thing: Sigma shot is upsampled big time, so actually it DOES show the advantage on per-pixel basis.

 dr.noise's gear list:dr.noise's gear list
Sigma DP2s Sony DSC-RX0 Sigma SD10 Sigma fp
Juri D Regular Member • Posts: 428
Re: No need to argue which is correct.......

dr.noise wrote:

PaulyOly wrote:

but this test doesn't show an advantage of no AA filter judging by
the sigma shot.

You forgot one little thing: Sigma shot is upsampled big time, so
actually it DOES show the advantage on per-pixel basis.

no one is arguing with per-pixel advantage of Sigma, we are trying to find out how much overal resolution Sigma has compared to Canon. So far Canon seems to resolve more quite visibly.

I love your idea about the contest. Maybe that's what we should do.

Juri D Regular Member • Posts: 428
SD10 vs. 20d Resolution Analysis Part Deux

I did a couple more of blow ups (watch). 20D seems to be holding detail better. Look at cloth texture, pointers, watch logo etc.

Of course, one see strong aliasing on Sigma's shot . What surprises me, however, is to see color transitions on Sigma: look at botton edge of watch ring bordering on the belt, the nob. Anyway, can someone do it better, both for Sigma and Canon?

CANON 20D

SIGMA SD10

PaulyOly Veteran Member • Posts: 7,723
They are both upsampled big time......

It's just the 20d shot is upsampled in camera from 2mp to 8mp, so really the 20d is upsampled big time, right? I'm just going by what sigmasd9 states in his posts. I'm assuming he has done the research and this is correct, i don't have the time nor the desire to check.

What's your opinion on the Baileys name being whiter on the 20d image? I noticed it a little on the box in the left of the frame. I also noticed during all this pixel peeping that the 20d shot is overall brighter than the sigma shot, is this the lens difference or somehow the flash? There are a few little color differences here and there, did you notice the second hand on the watch is red in the 20d shot, but almost without color in the sigma shot.

dr.noise wrote:

PaulyOly wrote:

but this test doesn't show an advantage of no AA filter judging by
the sigma shot.

You forgot one little thing: Sigma shot is upsampled big time, so
actually it DOES show the advantage on per-pixel basis.

-- hide signature --

Narrow depth of field ahead
Use extreme caution

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

SigmaSD9 Senior Member • Posts: 2,013
Re: Oh now you want to compare..........

PaulyOly wrote:

SigmaSD9 wrote:

Jafalt wrote:

Why not resize the Canon to double Sigma size?
That would be more fair

Or better still, since the Sigma is showing a bit more detail, size
them both to 3.43MP. You won't lose resolution.

a downsampled 20d image, make up your mind. I see no more detail.

My point exactly. Interpolation never adds any optical value. It can't, it's done after the shutter closes.

Bayers upscale their images because interpolation is absolutely required to combine uneven color channels, there isn't enough red/blue to match up with green so you have to insert digital placeholders. It then becomes an extra processing step to downsample the image back to its optical size, plus you lose marketing capital, so why do it?

Interpolation adds no optical value, but it also hurts nothing, so they get greedy and leave the 20D's 2MP optical output as 8MP recorded pixels, and they save the extra processing which would be required to shed the needless bandwidth.

So again, why not downsample instead of upsample these images? You lose nothing by doing so.

dr.noise
OP dr.noise Veteran Member • Posts: 3,767
Re: SD10 vs. 20d Resolution Analysis Part Deux

Juri D wrote:

What surprises me, however, is to see color transitions on Sigma

These are little CA which were blown up.

 dr.noise's gear list:dr.noise's gear list
Sigma DP2s Sony DSC-RX0 Sigma SD10 Sigma fp
Alex Mylnikov Junior Member • Posts: 48
Re: Yo!!! Hold the horses!!! [IMG]

Yes, you are right. But!? It resembles me the computer war Windows against Linux. I think this is not about Sigma or Canon. I have a Canon, and I love it, but the issue is far from the competition between Canon and Sigma. It is about what kind of “digital film” we will have tomorrow. I am sure that Foveon is a future. I am not talking about Sigma. Sigma is like a Comondor in computer history, or even more Xerox Lab. But, we are as a customers have a right to chose and to form the next stage of digital photography that will arrive sooner, not latter.

AlexM

Pepe-Lepue wrote:

You had to process the images . . . right? Using any RAW converter
is a type of process applied to the data to get the image.
There is no doubt that the Bayer needs to have some USM applied to
optimize the image, but when done at (300, .3. 2) the amount of
halo is hardly visible. In fact, our astro imaging devices (which
ALL color sensors use the Bayer (or some five sensor design) apply
a type of USM similar to the formula I just described since the
halo effect is nominal.

I was actually considering the SD10 (along with the Olympus) not
long ago to replace my Canon gear when I felt that Canon QC wasn’t
up to standards. I proficiently used the SD10 for two days, but
couldn’t make it work for my needs. The Olympus came very close
however. My favorite competitor was the Nikon D70 . . . that camera
ROCKED for the price!!!

While very pleased with the SD10 images, I wasn’t all pleased with
the Sigma system. Also the SD10 shadow detail lacked, but maybe it
was my processing technique. I could squeeze so much more shadow
detail out of the 20D. The SD10 just wasn’t for me. Image quality
alone doesn’t make the whole camera system for me. . . and at that
the 20D and SD10 have their tradeoffs in image quality (the
differences are too nominal to make an issue of). I needed
something comparable to the Canon system as a whole. So far Nikon
is the only unified system that can compete with Canon in offering
such a complete camera system. I ended up buying the D70 as a
second body.

In the end Canon came through for me and fixed my issues and I can
continue shooting with my lens investment.

I don't mean to sound like I'm bashing the SD10. I'm just sharing
my experiences with it in a very honest manner. I've had my bad
issues with Canon as well and have loudly voiced my concerns.

Sigma can’t sell on the Foveon alone. While great technology, it
has to be employed in a much more comprehensive system if it is to
see the light of tomorrow. Most of the pros accept the Bayer system
because it gives them exactly what they need in a comprehensive
system that delivers for them consistently. Most pros don’t
pixel-peep like I do (and most here in these forums.

dr.noise
OP dr.noise Veteran Member • Posts: 3,767
Re: No need to argue which is correct.......

Juri D wrote:

So far Canon seems to resolve more quite visibly.

should be added: ... in some areas of the image.

in other areas, Sigma and Canon are on par, and in some others Sigma resolves better.

So we can say that the resolution advantage depends on particular subject - for example, black and white patterns are easy for Canon. Would the background be blue or, better yet, blue and red, do you think Canon still could have the advantage?

I think Sigma's resolution is between 6 and 8 mp.

 dr.noise's gear list:dr.noise's gear list
Sigma DP2s Sony DSC-RX0 Sigma SD10 Sigma fp
dr.noise
OP dr.noise Veteran Member • Posts: 3,767
No...

PaulyOly wrote:

It's just the 20d shot is upsampled in camera from 2mp to 8mp, so
really the 20d is upsampled big time, right?

No. We're talking about spatial upsampling which mainly depends on luminocity information, which, in turn, mainly delivered by green channel. So the luminocity resolution for 20D is at least 4 mp.
20d is "upsampled" color-wise, not spatial resolution-wise.

There is always a chance for 20D to get two distinct luminocity pixels where SD10 could get only one by all means.

What's your opinion on the Baileys name being whiter on the 20d
image? I noticed it a little on the box in the left of the frame.

I don't remember. Need to check the bottle, but I don't think there is some irrecoverable problem.

I also noticed during all this pixel peeping that the 20d shot is
overall brighter than the sigma shot, is this the lens difference
or somehow the flash?

The flash. I used Elinchrom 500 bounced from ceiling.

I first shot 20D, keeping the histogram just near the right edge, then I shot SD10, using the same flash power, but its image happened to be significantly overexposed. So I decreased the flash power and maybe too much. In RAW convertor I just didn't notice the brightness difference as I developed the shots one by one without comparing.

There are a few little color differences
here and there, did you notice the second hand on the watch is red
in the 20d shot, but almost without color in the sigma shot.

That depends of processing (see SPP vs. ACR) and other factors. Better try it yourself with RAW files.

 dr.noise's gear list:dr.noise's gear list
Sigma DP2s Sony DSC-RX0 Sigma SD10 Sigma fp
Pavel Sokolov
Pavel Sokolov Senior Member • Posts: 1,231
My own test with your RAW files

Canon - original image opened with photoshop cs with default settings.

Sigma - opened with photoshop cs with default settings + upsampled with genuie fractals up to canon size + 50% unsharp mask ( 1.1 pix radius) + some white ballance changes.

-- hide signature --

If you found any errors in my english - please let me know.

http://www.pbase.com/crea7or/
http://crea70r.photosight.ru/

PaulyOly Veteran Member • Posts: 7,723
You sharpened the sigma image........

but not the canon, i sharpened all mine equally.

No soup for you! Next!
--

Narrow depth of field ahead
Use extreme caution

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

Pavel Sokolov
Pavel Sokolov Senior Member • Posts: 1,231
Re: You sharpened the sigma image........

but not the canon, i sharpened all mine equally.

canon sharpened by default in the adobe raw convertor

-- hide signature --

If you found any errors in my english - please let me know.

http://www.pbase.com/crea7or/
http://crea70r.photosight.ru/

Erik Magnuson Forum Pro • Posts: 12,247
Re: You sharpened the sigma image........

Pavel Sokolov wrote:

canon sharpened by default in the adobe raw convertor

So does the Sigma (at least if it follows SPP default conventions.) Anyway, it looks to me like there is a slight sharpening halo around the wooden handle in your Sigma crop.

-- hide signature --

Erik

Pavel Sokolov
Pavel Sokolov Senior Member • Posts: 1,231
Re: You sharpened the sigma image........

So does the Sigma (at least if it follows SPP default conventions.)
Anyway, it looks to me like there is a slight sharpening halo
around the wooden handle in your Sigma crop.

I see that the canon image have more details and does not want to say that sigma is cool. but the difference is not too much. this sharpening artefacts visible after the genuie fractals work at this object(before sharpening ) - I don't know why. sharpening at the other object is same as canon.

btw. my second body will be canon 1d m2, I think

-- hide signature --

If you found any errors in my english - please let me know.

http://www.pbase.com/crea7or/
http://crea70r.photosight.ru/

PaulyOly Veteran Member • Posts: 7,723
In that case.......

Pavel Sokolov wrote:

but not the canon, i sharpened all mine equally.

canon sharpened by default in the adobe raw convertor

here's the canon image resized to the native resolution of the sigma which is superior in sharpness to anything by default. I sharpened the canon image, no need to sharpen the sigma right? I'm going to be sorry to see this post hit 150. I can't wait for the next comparison.

-- hide signature --

Narrow depth of field ahead
Use extreme caution

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

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

Juri D wrote:

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!

It creates a surface print for you, it doesn't delvelop a photograph for you.

Erik Magnuson Forum Pro • Posts: 12,247
Now the halo's on the other foot

err, leg.

PaulyOly wrote:

I sharpened the canon image, no need to sharpen the sigma right?

Now there is a distinct sharpening halo around the statuette in the Canon image, most noticably around the leg area. Well, at least it's a change from arguing over whether we can see a spire on the Tower Bridge. I think we've now conclusively proved...

...nothing.

-- hide signature --

Erik

ritz Contributing Member • Posts: 513
Re: Who beat the sh*t out of who.

ritz wrote:

dr.noise wrote:

20D + 50/1.4:

SD10 + 50/2.8 EX upsampled to 8mp:

Both shots made at F11, 1/200, ISO100.
20D shot saved from C1Pro, low noise reduction, no sharpening.
I saved SD10 shot double-size from SPP, then resized to 8mp in
ACDSee. I also tried saving same size then bicubic upsample in PS,
but got absolutely identical result.

Originals saved as 100% JPG:

; (20D; 8.7 mb)

; (SD10; 8.17 mb)

RAW files:
http://zdg.ru/tmp/200502/IMG_0802.CR2 (20D; 8.4 mb)
http://zdg.ru/tmp/200502/IMG07695.X3F (SD10; 8.04 mb)

Probably the SD10 shot could be upsampled better with less
aliasing. I was thinking about pulling sharpness setting back to
-1.0 before upsampling. You can try raw files yourselves.

About color: I didn't do any precise color management as my goal
was to compare details. 20D was shot on Auto WB and it got white
sheet right. So I assumed there is no need for further processing.
SD10 on Auto WB produced bluish white sheet, so I switched to
Custom WB.

And for what we have: 20D shot is mostly wrong except white sheet.
Striped cloth on the left is pinkish, should be more yellow like on
SD10 shot (spot on!), the toy dog is pinkish too (SD10 spot on),
blues are lighter than needed. Yellow stripe on the magazine cover
is too pale and wrong hue. SD10's one is right, but slightly
oversaturated. The red pepper of SD10 is right but a bit pale,
20D's is well-saturated but too orange. Yellow pepper is good on
both shots. I must note that reference white sheet is white on both
shots.

-- hide signature --

No one will admit, but the true is the Sigma is woden and the 20D sharp

different colour yes. If you use Adobe CS for Sigma and 20D and blow the up, a 5 year kid can see it.
Jens Ritzlau

Keyboard shortcuts:
FForum MMy threads