The "wow" factor of slide film on a lightbox

Recently, RolandRB posted:

On 20 Aug, 16:12, RolandRB wrote:
On 20 Aug, 14:39, RolandRB wrote:





On 20 Aug, 13:44, "David J. Littleboy" wrote:

"RolandRB" wrote:

This is the Sigma SD10 image:
http://www.pbase.com/rolandrb/image/84194056/original

This is the Sony DSC-R1 image:
http://www.pbase.com/rolandrb/image/84194495

That's pretty badly sharpened to start with, but here ya go. 4MP
of aliasing
artifacts.

Look at slats on the windows on the next floor up on the building
to the right of "Apotheke". In the original Sony image, the R1
resolves those as _equal-width_ slats, but the Sigma camera
renders them as varying in width.

I agree with that. The Sony does that better.

If you start looking around at the fine detail, you'll find lots
more examples of detail that the Sony renders quite reasonably but
which the Sigma either turns them randomness or fails to render at
all. The Sigma loses it pretty badly on the roofs.

I agree with that again but the Sony image looks bad even though it
is holding more details. There are a lot of distracting effects that
spoil the image. The rendering on the walls on the building behind
looks unreal rather than like a true texture. The rooves look
better, that is true, but maybe only because I have oversharpened.
If you look at the corner on the wall of the Sony image of the
building behind to the left of the TV mast then the Sony image
makes it look like there is a drainpipe running down the corner
edge when there is not. The lower roof of the rear building where
the bird is standing looks as though it had a black edge with a
white line on top. The people standing and sitting in front of the
Brötlibar restaurant look vague and unreal. They are too large and
their lack of detail is distracting. The foreground wall on the
right with the public seating in front has an unreal texture. The
"Tel 06" on the blue bin behind this wall does not show the "06"
clearly while the Sigma photo does. The six "BAR" red lettering in
the top windows below the "don't worry - be happy Bar" neon sign
look more like red curtain material in the Sony photo, though
downsized it looks better. The people sat below the "TicketCorner"
notice in the tram shelter look unclear. The people in the posters
to the left of the blue bin look unclear as do the real people in
front of the Rio Bar and the Zum Braunen Mutz. The Sony picture is
too big for the amount of detail it is showing and to me the image
looks bad. Printed out, it might look a bit better than the Sigma
photo, but to look at it on a computer screen then to me the Sigma
photo looks more like a print than the Sony photo does and is
easier on the eyes. It draws my view rather than repelling it.

Perhaps if I could send you the Sony photo jpeg as it came out of
the camera and you did a careful downsizing then the downsized Sony
photo might look better. As you rightly said, the downsized Sony
photo was showing some horrendous artifacts in places, though I
thought overall it looked better.

http://www.pbase.com/davidjl/image/84196441/original

Thanks. Your fixed copy looks quite good to me. The writing is not
quite as distinct but that could be due to the exposure being
higher for the Sony image.

David J. Littleboy
Tokyo, Japan- Hide quoted text -

- Show quoted text -

I was looking at the slats next to the Apotheke building on the floor
above. Today, looking at them, they were not regular. maybe the Sony
was showing false details that human eyes would prefer to see and the
Sigma was showing it more accurately. I will go back there now and
photograph a more detailed image.- Hide quoted text -

- Show quoted text -

Here are the slats as photographed today. The Sigma appears to have
done a better job at reproducing them than the Sony.

http://www.pbase.com/rolandrb/image/84205472

Very interesting, and just the kind of difference I would expect from a
Bayer-pattern vs. Foveon comparison!

Neil

On 20 Aug, 18:36, "Neil Gould" wrote:
Recently, RolandRB posted:





On 20 Aug, 16:12, RolandRB wrote:
On 20 Aug, 14:39, RolandRB wrote:

On 20 Aug, 13:44, "David J. Littleboy" wrote:

"RolandRB" wrote:

This is the Sigma SD10 image:
http://www.pbase.com/rolandrb/image/84194056/original

This is the Sony DSC-R1 image:
http://www.pbase.com/rolandrb/image/84194495

That's pretty badly sharpened to start with, but here ya go. 4MP
of aliasing
artifacts.

Look at slats on the windows on the next floor up on the building
to the right of "Apotheke". In the original Sony image, the R1
resolves those as _equal-width_ slats, but the Sigma camera
renders them as varying in width.

I agree with that. The Sony does that better.

If you start looking around at the fine detail, you'll find lots
more examples of detail that the Sony renders quite reasonably but
which the Sigma either turns them randomness or fails to render at
all. The Sigma loses it pretty badly on the roofs.

I agree with that again but the Sony image looks bad even though it
is holding more details. There are a lot of distracting effects that
spoil the image. The rendering on the walls on the building behind
looks unreal rather than like a true texture. The rooves look
better, that is true, but maybe only because I have oversharpened.
If you look at the corner on the wall of the Sony image of the
building behind to the left of the TV mast then the Sony image
makes it look like there is a drainpipe running down the corner
edge when there is not. The lower roof of the rear building where
the bird is standing looks as though it had a black edge with a
white line on top. The people standing and sitting in front of the
Brötlibar restaurant look vague and unreal. They are too large and
their lack of detail is distracting. The foreground wall on the
right with the public seating in front has an unreal texture. The
"Tel 06" on the blue bin behind this wall does not show the "06"
clearly while the Sigma photo does. The six "BAR" red lettering in
the top windows below the "don't worry - be happy Bar" neon sign
look more like red curtain material in the Sony photo, though
downsized it looks better. The people sat below the "TicketCorner"
notice in the tram shelter look unclear. The people in the posters
to the left of the blue bin look unclear as do the real people in
front of the Rio Bar and the Zum Braunen Mutz. The Sony picture is
too big for the amount of detail it is showing and to me the image
looks bad. Printed out, it might look a bit better than the Sigma
photo, but to look at it on a computer screen then to me the Sigma
photo looks more like a print than the Sony photo does and is
easier on the eyes. It draws my view rather than repelling it.

Perhaps if I could send you the Sony photo jpeg as it came out of
the camera and you did a careful downsizing then the downsized Sony
photo might look better. As you rightly said, the downsized Sony
photo was showing some horrendous artifacts in places, though I
thought overall it looked better.

http://www.pbase.com/davidjl/image/84196441/original

Thanks. Your fixed copy looks quite good to me. The writing is not
quite as distinct but that could be due to the exposure being
higher for the Sony image.

David J. Littleboy
Tokyo, Japan- Hide quoted text -

- Show quoted text -

I was looking at the slats next to the Apotheke building on the floor
above. Today, looking at them, they were not regular. maybe the Sony
was showing false details that human eyes would prefer to see and the
Sigma was showing it more accurately. I will go back there now and
photograph a more detailed image.- Hide quoted text -

- Show quoted text -

Here are the slats as photographed today. The Sigma appears to have
done a better job at reproducing them than the Sony.

http://www.pbase.com/rolandrb/image/84205472

Very interesting, and just the kind of difference I would expect from a
Bayer-pattern vs. Foveon comparison!

Neil- Hide quoted text -

- Show quoted text -

Me too but both methods have their strengths and weaknesses and both
methods are let down by the practicalities of using that method. A
Bayer 10.2 megapixel sensor should beat a Foveon 3.4x3 megapixel
sensor hands down in terms of its black & white resolution. The color
guessing should be a minor problem but would depend any case on the
interpolation software. The problem I see with using the Bayer pattern
is that an anti-aliasing filter is required to stop color Moiré
patterns. The Foveon sensor also has a regular array and so will
suffer Moiré patterns as well but less often and indeed, the Sigma
SD10 does not use an anti-aliasing filter. It is this introduced blur
of the anti-aliasing filter needed by the Bayer sensor that causes a
problem in that to undo the effects of the blur is not being handled
well in the software and perhaps it can't be. And where they try to
sharpen the blur then "unreal" detail come in that is annoying to look
at. But the the Foveon sensor has its faults as well. Colors are not
clearly distinct and two sensors can detect the same light freqeuncy
if the frequency falls between colors. Reds on the Foveon sensor are
not good because of this but then at the same time it gives a better
resolution for reds as red squares are only one quarter of the Bayer
pattern -- same as blue.

My own personal preference is that I want digital images to look good
on a computer monitor and to print the same. The Foveon sensor images
suit this although their colors can be misleading. It's a case of
"swings and roundabouts".

Recently, RolandRB posted:

On 20 Aug, 18:36, "Neil Gould" wrote:

Very interesting, and just the kind of difference I would expect
from a Bayer-pattern vs. Foveon comparison!


Me too but both methods have their strengths and weaknesses and both
methods are let down by the practicalities of using that method. A
Bayer 10.2 megapixel sensor should beat a Foveon 3.4x3 megapixel
sensor hands down in terms of its black & white resolution. The color
guessing should be a minor problem but would depend any case on the
interpolation software. The problem I see with using the Bayer pattern
is that an anti-aliasing filter is required to stop color Moiré
patterns.

That is the common line of thinking, but in practice I don't think it is a
major issue. Leica's DMR lacks an anti-aliasing filter, and as a result
can resolve a scene better than cameras with 50% more pixels. The number
of times that Moiré issues ruin the shot is probably less than those
ruined by the loss of resolution.

[...]
My own personal preference is that I want digital images to look good
on a computer monitor and to print the same. The Foveon sensor images
suit this although their colors can be misleading. It's a case of
"swings and roundabouts".

Or, TANSTAAFL...

Neil

On 20 Aug, 22:15, "Neil Gould" wrote:
Recently, RolandRB posted:

On 20 Aug, 18:36, "Neil Gould" wrote:

Very interesting, and just the kind of difference I would expect
from a Bayer-pattern vs. Foveon comparison!

Me too but both methods have their strengths and weaknesses and both
methods are let down by the practicalities of using that method. A
Bayer 10.2 megapixel sensor should beat a Foveon 3.4x3 megapixel
sensor hands down in terms of its black & white resolution. The color
guessing should be a minor problem but would depend any case on the
interpolation software. The problem I see with using the Bayer pattern
is that an anti-aliasing filter is required to stop color Moiré
patterns.

That is the common line of thinking, but in practice I don't think it is a
major issue. Leica's DMR lacks an anti-aliasing filter, and as a result
can resolve a scene better than cameras with 50% more pixels. The number
of times that Moiré issues ruin the shot is probably less than those
ruined by the loss of resolution.

I agree although there will be situations where it would be a waste of
time to even try taking a shot without an anti-aliasing filter. With
an anti-aliasing filter you can take that shot but like you say, the
downside is the lack of resolution. It is a shame this filter can not
be switched in and out and so only used as the need arises. I don't
know why manufacturers haven't implemented this in some cameras
although I guess moving parts near the sensor would give rise to
abrasion and perhaps specks on the sensor.

[...] My own personal preference is that I want digital images to look good
on a computer monitor and to print the same. The Foveon sensor images
suit this although their colors can be misleading. It's a case of
"swings and roundabouts".

Or, TANSTAAFL...

Neil

"RolandRB" wrote in message oups.com...[snip]


Ever heard the term " sigmatism "? I think you have a perfect case!
AAvK

A very interesting claim. It so happens I don't have any suitable
downsampling software for the task you describe so I was wondering if
you had time to do this and send me the resulting image. I will then
web that with the other two.

This is the Sigma SD10 image:
http://www.pbase.com/rolandrb/image/84194056/original

This is the Sony DSC-R1 image:
http://www.pbase.com/rolandrb/image/84194495



Irfanview has a great one, just use the " Lanczos " algorythm in the drop down to
the lower right of the rezise dialog box, you'll get fine results, download it for free.

AAvK