Kodad Samples - Magnification fools us!

The more pixels of a image that does not fit on your screen while looking at a 100% crop, the more bad that 100% crop (1600x1200) should look like! Ohhh, I got it! Yes!!! Imagine I was looking at a 1600x1200 crop of a 100Mp picture (50 pictures of 2Mp stitched together with panavue), that would look awfull, no detail left over at 100% view, only 2/100 is 2% of sharpness left over per original picture. I think I'll better try to get a 1.2Mp SLR...
Jack.

Geir Ove said:

zettlers said:

Geir Ove said:

Hello,

I first jumped on the booo bandwagon with the rest of the "mob". I
downloaded the Fruit Image (from DPReview) and looked at it at 1:1.
That is 4536 * 3024 pixels.

But wait, this isn't fair... My screen is 1024*768 and thus 4536 *
3024 represents a pictures 4 times higher and 4 times wider thand
my screen: At 1:1 the picture thus fills 4*4 = 16 screens! Thus,
one are looking at the picture at a tremendous MAGNIFICATION! If
you do the same with a 6 Mpix D60 picture, the magnification is
(3072 / 1024 x 2048/768) ~ 3 times.

If I open the picture in e.g. Paint Shop Pro and look at it when it
fills the screen, the picture is not that bad at all!

AND: The Fruit has, according to Jay (at Kodak) some sort of sugar
glazing poored over it in the upper part of the picture: Thus the
artificial look.

The IR pictures are another story: The Tree picture was very bad,
but this can be due to the photographer of post processing

Note: I am not a potential 14n buyer, but I thought we ought to be
fair to Kodak

Geir Ove

Click to expand...

Click to expand...

Click to expand...

Than I should reconsider another site for my hobby. ))
Jack.

Geir Ove said:

b) If you still don't understand it: Reconsider your hobby.

Geir Ove

Geir Ove said:

Hello,

I first jumped on the booo bandwagon with the rest of the "mob". I
downloaded the Fruit Image (from DPReview) and looked at it at 1:1.
That is 4536 * 3024 pixels.

But wait, this isn't fair... My screen is 1024*768 and thus 4536 *
3024 represents a pictures 4 times higher and 4 times wider thand
my screen: At 1:1 the picture thus fills 4*4 = 16 screens! Thus,
one are looking at the picture at a tremendous MAGNIFICATION! If
you do the same with a 6 Mpix D60 picture, the magnification is
(3072 / 1024 x 2048/768) ~ 3 times.

If I open the picture in e.g. Paint Shop Pro and look at it when it
fills the screen, the picture is not that bad at all!

AND: The Fruit has, according to Jay (at Kodak) some sort of sugar
glazing poored over it in the upper part of the picture: Thus the
artificial look.

The IR pictures are another story: The Tree picture was very bad,
but this can be due to the photographer of post processing

Note: I am not a potential 14n buyer, but I thought we ought to be
fair to Kodak

Geir Ove

Click to expand...

Click to expand...

The more of a picture that runs of your screen because of its much higher captured resolution (maybe many pics taken and stitches together) does not mean the 100% crop should equally look worse.

Jack.

Jeff Barkley said:

Geir Ove said:

Hello

A lot of people simply don't understand what I am trying to say.

I'll try to make it clear: Imagine you have a camera with 10.000 *
10.000 resolution. You take a picture of a part of a footballfield:
Imagine that the football in the picture takes up 2000*2000 pixels
(it's not square, but for simplicity).

Now you display this picture in 1:1 on a screen with e.g. 1000*1000
resolution. (Pretty close to what we have). Half (in each
direction) the football is now taking up ALL the screen, and of
course looks bad (steps because of your screen resolution).

If you now increase the Screen Resolution to 3000*3000, the
football takes up 1/9 of the screen area, and now looks very good.

If you reduce the Screen 500 * 500, the 1/4 of the football (in
each direction) now fills the screen, and you are so "close up" to
it that you can't even tell what it is!

Thus, the Screen Resolution versus the Resolution of the Ppicture
DOES matter when you Study a picture.

The only thing that is fair, is to have a Monitor that matches the
Picture resolution and look at the picture in 1:1 at such a
monitor. For the Kodak 14n this would a Monitor with 3500 * 4000
pixels resolution. Most of us does NOT have such a monitor.

Thus: Comparing pictures from different cameras on the same
resolution Monitor is NOT FAIR to the high res cameras

Geir Ove

Click to expand...

--

Click to expand...

Thanks Daniel! Although we have now to reconsider our hobbies....
Jack.

Daniel Kleczynski said:

From human point of view, the progress in digital camera design is,
basically, to deliver higher and higher resolution. From
"computer's" point of view, this simply means more pixels. More
pixels and keeping the picture sharp, so that the details can be
discerned, but no (strong) artifacts are added.

I always analyze samples 1:1 (1 screen pixel is used to represent 1
pixel of the file). And I can observe the opposite of what you are
saying: when I bought D60, I noticed 1:1 crop looked much better
than 1:1 crop from my earlier digicam, Sony F707. Pics from F707
viewed the same way look way sharper than photos from my earlier
Coolpix 990, and so on... It turns, that comparing pics on screen
doesn't favour lower res cameras. It shows bare truth.

Of course, in the cases I mentioned above, the critical factor was
quality of the lenses and CCDs - D60 with a decent L series lens is
obvioulsy superior to F707, and not only beacuse of pixel count.

But, still, this abnegates your thesis, that this sort of
comparison is not fair to higher res cameras.

When I'm going to buy a new camera, I want it at least to keep the
sharpness of the older (I'm talking about 1:1 viewing), but have
more pixels. This way my photos can be bigger in print, or keep the
size, but have more DPI and look better, as less artifacts are
visible (pixels are smaller on paper). If I would use your kind of
thinking, I would still keep one of my older cameras and up-sample
the pics in Photoshop.

When analyzing the pic from a new camera (apart from tone issues
etc.), I don't care what object it depicts - I only focus on
quality: sharpness, CA, noise, banding, moire, aliasing etc. That's
better visible at 1:1. No one was fooled - that's the proper way
to examine camera's quality (and that's why we often see crops from
pics at DPReview or Image Resource).

Geir Ove said:

Hello

A lot of people simply don't understand what I am trying to say.

I'll try to make it clear: Imagine you have a camera with 10.000 *
10.000 resolution. You take a picture of a part of a footballfield:
Imagine that the football in the picture takes up 2000*2000 pixels
(it's not square, but for simplicity).

Now you display this picture in 1:1 on a screen with e.g. 1000*1000
resolution. (Pretty close to what we have). Half (in each
direction) the football is now taking up ALL the screen, and of
course looks bad (steps because of your screen resolution).

If you now increase the Screen Resolution to 3000*3000, the
football takes up 1/9 of the screen area, and now looks very good.

If you reduce the Screen 500 * 500, the 1/4 of the football (in
each direction) now fills the screen, and you are so "close up" to
it that you can't even tell what it is!

Thus, the Screen Resolution versus the Resolution of the Ppicture
DOES matter when you Study a picture.

The only thing that is fair, is to have a Monitor that matches the
Picture resolution and look at the picture in 1:1 at such a
monitor. For the Kodak 14n this would a Monitor with 3500 * 4000
pixels resolution. Most of us does NOT have such a monitor.

Thus: Comparing pictures from different cameras on the same
resolution Monitor is NOT FAIR to the high res cameras

Geir Ove

Click to expand...

Click to expand...

When I stitch say 20 pictures of 6Mp together it should not influence the sharpness of each of them. The 100% crop of the 120Mp picture should look exactly the same as the 100% crop of one single picture it was made of. There is no enlargement at 100% crop view!

Jack.

I did mean magnified instead of enlarged... ;-) Hope you still understand my point of view 100%... ))
Jack.

Jack said:

When I stitch say 20 pictures of 6Mp together it should not
influence the sharpness of each of them. The 100% crop of the 120Mp
picture should look exactly the same as the 100% crop of one single
picture it was made of. There is no enlargement at 100% crop view!

Jack.

Click to expand...

Hi,
Make sense to me peter.

Just like to add to Flaw 2. We all know what happens when trying to judging how something will print on screen. The printed version always looks better. I use the benchmark of 150ppi as the min. print res. and 240ppi and above for top quality.

On the sharpening front, when performing sharpening you need to taking into account the output devices resultion & size of print etc. Hence filters like Nik have diff. setting for screen & printers and you need to enter the size of the final picture etc.

In the tests with nik I have done when sharpening for the printer the screen images does not change much(don't get sharper). If you sharpen the same image for WEB use, then the image change a lot on screen.

Alex

Pete Cockerell said:

Ludicrous. With apologies to Churchill, never has so much been
written by so few to so little effect.

Flaw 1: The 300dpi in the Kodak images is entirely arbitrary. It's
just a number put in by the manufacturer so that if someone naively
prints an image from, say, Photoshop, without doing any other size
manipulation (with or without resampling), they'll get a not too
appallingly large print (10" x 15" approx). Kodak could have easily
put 72 or 150 or anything else in the image. For example, my Sony
F707 uses 72dpi. Using your technique, I would actually have to
increase the size of the image to get the PS ruler to match up
with your physical one. Where does that leave me?

Flaw 2: Your thesis is that to judge what a picture will look like
when printed, you should resample it so that it looks the same size
as the print when displayed on a device (the monitor) that has at
best 10% of the areal resolution of the print device. Since when
does decimating an image by at least an order of magnitude show its
true quality? All your technique does is get the monitor to display
the size of an image when printed at some arbitrary
pixels-per-inch value.

I'm not sure why this is so hard to grasp, but the whole point of
displaying images 1:1 on the screen is to see how they will look
when printed on devices that have a much higher resolution (e.g.
a 2540 dpi Linotype). To say an image is fine because it looks OK
when resampled down to 10% of its original size borders on madness.

Incidentally, assuming your Windows monitor DPI setting is set
reasonably accurately, the Photoshop command "View ~ Print Size"
will do everything your involved techinque does, albeit with quick
'n' dirty real-time scaling and not proper resampling. But since
your technique doesn't actually have anything to do with previewing
print quality , that doesn't really matter.

Cheers,
Pete

--
http://www.pbase.com/pcockerell
http://www.peter-cockerell.net:8080/

Click to expand...

--
Alex
LWS photographic (UK)

seems to be the standards around here. Well, when nothing technical can be argued, I guess that is what we can expect.

What I do for a living: A Software Engineer. Just finished a patent application in the field of Software Engineering: A pity you wheren't there to help me out.

Geir Ove

Pete Cockerell said:

Geir said:

Your explanation and reasoning is very good !

Click to expand...

Haha, good to see you maintain your sense of humor, Geir.

Can I ask you a question? What exactly is it you do for a living?
Or maybe I should ask, when was the last time you got a new
prescription for your glasses? Or, one more, why do you think Kodak
pulled all of the original appalling images from their website and
replaced them with ones where the noise reduction has been
dramatically decreased? (See the I-R article about this.) I mean
what is it about image detail totally obliterated over an area of
100s of pixels that you find acceptable at any viewing size?
Maybe you and David should form a "Not very fussy about image
quality because I can always just resize the image to 1/10 of its
original size club".

Cheers,
Pete

PS I've been trying to work in a line of the form "I think you
should just Geir Ove it," but I'm not sure if the pun would work
with the way you pronounce your name...

--
http://www.pbase.com/pcockerell
http://www.peter-cockerell.net:8080/

Click to expand...

but the AREA they are "spread" over: Of course it matters. If are looking at a 4500 * 3000 (or whatever) at a screen with approx. 1024 * 768 it would take a screen approx. 16 times the size of your screen to display the full image. If you adjusted the distance to the screen accordiingly (watched it a several meters distance) you would compensate: But most of us are sitting pretty close to the screen....

It is the same as watching a Wall Painting at 30 cm distance: You get lost in the detail and can't see what it really is.

Geir Ove

Jack said:

The pixels on your screen are as big as with any other picture from
whatever digital camera, no mather it's resolution. The amount of
detail per area of say 100x100 pixels is significant better with
pictures of other cameras, so the detail of the posted 14n pictures
(the fruit in particular) does not live up to a captured resolution
of 14Mp.

While still being razor sharp, the n14 should be able to produce
much bigger pictures than for example 6Mp cameras. With that
particular bad fruit picture, I found a corresponding resolution of
roughly 1.4Mp. I doubt the n14 can't do better, the images must
have messed up somewhere.

Jack.

Geir Ove said:

Hello,

I first jumped on the booo bandwagon with the rest of the "mob". I
downloaded the Fruit Image (from DPReview) and looked at it at 1:1.
That is 4536 * 3024 pixels.

But wait, this isn't fair... My screen is 1024*768 and thus 4536 *
3024 represents a pictures 4 times higher and 4 times wider thand
my screen: At 1:1 the picture thus fills 4*4 = 16 screens! Thus,
one are looking at the picture at a tremendous MAGNIFICATION! If
you do the same with a 6 Mpix D60 picture, the magnification is
(3072 / 1024 x 2048/768) ~ 3 times.

If I open the picture in e.g. Paint Shop Pro and look at it when it
fills the screen, the picture is not that bad at all!

AND: The Fruit has, according to Jay (at Kodak) some sort of sugar
glazing poored over it in the upper part of the picture: Thus the
artificial look.

The IR pictures are another story: The Tree picture was very bad,
but this can be due to the photographer of post processing

Note: I am not a potential 14n buyer, but I thought we ought to be
fair to Kodak

Geir Ove

Click to expand...

Click to expand...

No, but the Display SIZE of this picture at 1:1 depends on the resolution and size of your monitor: It is the number of lines (or pixels) per mm. (or inch) that determines how BIG the picture looks at 1:1 scaling on any screen.

Geir Ove

Jack said:

When I stitch say 20 pictures of 6Mp together it should not
influence the sharpness of each of them. The 100% crop of the 120Mp
picture should look exactly the same as the 100% crop of one single
picture it was made of. There is no enlargement at 100% crop view!

Jack.

Click to expand...

Hello,

Try this: Take any picture from e.g. a D60. Zoom it up to 1:1 size in PaintShop Pro or PhotoShop.

Put your monitor in 1600*1200 resolution. Then change the resolution down to 640*480: The same picture now looks pretty bad, because the lines per inch (or mm.) has been increased: A zooming / magnifying effect.

The picture must now be looked at a much LRGER distance to give the same perceived quality to our eyes: But we are normally sitting quite close to our computer.

The "mismatch " gets worse the higher res. the picture is compared to the res. settings of your monitor (assuming 1:1 scaling of the picture all the time)

Geir Ove

Geir Ove said:

seems to be the standards around here. Well, when nothing technical
can be argued, I guess that is what we can expect.

Click to expand...

Wrong, my arguments have been very well thought out, cogent, and beyond contradiction. It's only when you and your cohort (singular so far, thankfully) seem incapable of the simplest level of reasoning that it drives me to sarcasm.

Geir Ove said:

What I do for a living: A Software Engineer. Just finished a patent
application in the field of Software Engineering: A pity you
wheren't there to help me out.

Click to expand...

Haha, I'm sure you're very good at what you do. It clearly isn't in the field of imaging, as I'm sure you would have mentioned if it were.

Hmm, you didn't answer my third or fourth questions. Funny that. I guess they're a bit harder for you to answer with any credibility than "what's your job.". Why don't you just save yourself further embarrassment and admit that you were caught with no clothes, trying to defend the indefensible. The sad thing is, subsequents posts by Kodak and on the Japanese site have show that your defence of the flawed images was largely unnecessary, yet still you stick to your strange reasoning. Sad, really.

Cheers,
Pete

--
http://www.pbase.com/pcockerell
http://www.peter-cockerell.net:8080/

OK, I've tried this argument before, but let's try again: if you read what you write below, what you are saying is that a 14Mpixel image is never necessary, because its finest details (or glaring flaws) are going to be lost in the display/printing process anyway. Isn't it a shame no-one told Kodak before they went to all the trouble of investing in that big ol' CMOS sensor (which is made by a 3rd party, I know).

Using the argument that you and David put forward, the only requirement is that the image look good on the screen when viewed at a normal viewing distance. We should have stopped worrying about pixel resolution at around 2.0 pixels then, taking a generous screen resolution of 1600 x 1200.

But wait! If I print my Kodak image on a device that can resolve at 300 dpi it will be 15" x 10". If I stick that mini-poster on my wall, the noise reduction flaws will be painfully obvious to anyone walking past it. Yet, you say, it doesn't matter because if I rescaled the picture to display at 15" x 10" on my 19", 90dpi monitor, the same flaws would be largely masked by the resampling process? I just don't see how you can rationalize that, but since you're never going to admit it's way beyond rationalization, I guess it's a waste of my and everyone else's time trying to get you to do it.

Cheers,
Pete

Geir Ove said:

but the AREA they are "spread" over: Of course it matters. If are
looking at a 4500 * 3000 (or whatever) at a screen with approx.
1024 * 768 it would take a screen approx. 16 times the size of your
screen to display the full image. If you adjusted the distance to
the screen accordiingly (watched it a several meters distance) you
would compensate: But most of us are sitting pretty close to the
screen....

It is the same as watching a Wall Painting at 30 cm distance: You
get lost in the detail and can't see what it really is.

Click to expand...

(BTW this comment really makes me doubt your eyesight again. Are you sure you didn't mean 300cm?)

--
http://www.pbase.com/pcockerell
http://www.peter-cockerell.net:8080/

I think everyone is getting confused with the explainations and not the facts.

1. A 100 percent (magnification) view of an image is a good way to judge say all 3MP cameras. That's a level playing field.

2. But to compare a 6MP image vs a 14MP image at 100 percent mag side by side is not far...to the 14MP camera. First assuming the same lens is used, the small portion of the 14MP image being viewed is at a much higher magnification. It like me giving you a 2x loop to look at a 6MP print and a 6x loop to look at the 14MP print. Sure you will see more problems with the higher magnification.

Putting all this in a different way. If you began with a 6 and 14 MP file and began producing prints from 4x6 to 40x60. The difference would not be evident until you got to something larger then 8x10. At say 20x30 you would begin detecting flaws in the 14MP file printed poster. Some of those flaws would be due to the lens, others to the aliasing filter etc.

Pete Cockerell said:

Using the argument that you and David put forward, the only
requirement is that the image look good on the screen when viewed
at a normal viewing distance. We should have stopped worrying about
pixel resolution at around 2.0 pixels then, taking a generous
screen resolution of 1600 x 1200.

But wait! If I print my Kodak image on a device that can resolve at
300 dpi it will be 15" x 10". If I stick that mini-poster on my
wall, the noise reduction flaws will be painfully obvious to anyone
walking past it. Yet, you say, it doesn't matter because if I
rescaled the picture to display at 15" x 10" on my 19", 90dpi
monitor, the same flaws would be largely masked by the resampling
process? I just don't see how you can rationalize that, but since
you're never going to admit it's way beyond rationalization, I
guess it's a waste of my and everyone else's time trying to get you
to do it.

Cheers,
Pete

Geir Ove said:

but the AREA they are "spread" over: Of course it matters. If are
looking at a 4500 * 3000 (or whatever) at a screen with approx.
1024 * 768 it would take a screen approx. 16 times the size of your
screen to display the full image. If you adjusted the distance to
the screen accordiingly (watched it a several meters distance) you
would compensate: But most of us are sitting pretty close to the
screen....

It is the same as watching a Wall Painting at 30 cm distance: You
get lost in the detail and can't see what it really is.

Click to expand...

(BTW this comment really makes me doubt your eyesight again. Are
you sure you didn't mean 300cm?)

--
http://www.pbase.com/pcockerell
http://www.peter-cockerell.net:8080/

Click to expand...

--
Ken Eis

Geir, I for one am in your camp. I think that because of your "foreign accent", you have had trouble expressing your view in a way that we can really understand.

Sincerly, I am surprised that all these "professional" photographers don't know the basics of screen display.

To all you pro's that have derided Geir's posts, do this: Take the same photo with a 2MP camera and a 6MP camera. Display it 1:1 on screen. The displayed 6MP camera image is only a portion of the 2MP camera image. In order to do a "valid" comparison on screen, you need to shrink the 6MP image to the same VISUAL size as the 2MP camera. Read on before you comment.

So in our comparison of the Kodak 14N to other cameras, we need to compare the images side by side at the same "visual" resolution in order to be fair. The 14N image needs to be visually shrunk to the same size as the comparison camera, or the comparison camera image needs to be enlarged to be visually the same dimension as the 14N image.

I can see it now... "enlarging my image from my 6MP camera isn't fair because the resolution is less". Yes, that's exactly the point! But that is the reality of side by side comparison. You have to use the same ruler.

That does not take away from the fact that the images that were presented by Kodak as samples a couple of days ago were completely out-of-line from what should have been done. Those photos technically sucked. Photo samples should show a product in the best light possible (pun intended); and those images were completely improper for a major release of this type. As a concerned shareholder in Kodak, I have already written management complaining about this fiasco... and I don't care what the intent of the presenter was... it was WRONG.

Edward

Geir Ove said:

Hello David,

Nice to finally see someone who understands this. I was a bit
shocked by the very number of "professionals" here that where
unable to reason about this, but did not hestitate to make fun of a
subject they apparently did not understand: A bit dissapointing...

Your explanation and reasoning is very good !

Geir Ove

Click to expand...

Hi Geir,

As anyone with experience with graphics displays will tell (and has been telling) you, that is not an objectiveway to judge the quality of a high res camera. Let's look beside the obvious point in contention for now. For one, you're introducing an artificial higher bound for images. When you downsample to 1024x768, a 3mp image will look just as good as any 14mp image no matter how bad the camera is, and both will be as good as any 500000x500000 image, because you're essentially reducing every image to 1024x768.

So here's the problem: What if you don't like the camera's results? Does changing your screen resolution make a bad camera better? It really shouldn't, since what people are interested here is an objective comparison of how cameras perform, pixel for pixel, regardless of monitor resolution. Comparing images 1:1 by sensor count (bayar interpolation notwithstanding) is the only way to acheive an objective view. Your method actually INTRODUCES magnification into the picture. Suddenly you're looking at one image at 100% magnification, another image at 75%, and third one at 10%.

This leads into the next, probably the most important point: absoutely nobody who buys a 14n will do so for full-screen viewing, especially at such a paltry resolution as 10x7. Even people with 1900x1600 screens will not need a 14n for that kind of usage. Photographers who choose cameras of this type will be printing images at resolutions far-exceeding that of any consumer monitor.

You are of course free to feel whatever you wish, and if that's how you judge a camera, more power to you. However, please be open to the possibility that your method of measurement is wholely inadequate for anyone who works with digital images bigger than 2mp. To them, your comments just seem ignorant, and that is why you're getting such bad treatment.

Peace,
Kenn

Geir Ove said:

It is the same as watching a Wall Painting at 30 cm distance: You
get lost in the detail and can't see what it really is.

Geir Ove

Jack said:

The pixels on your screen are as big as with any other picture from
whatever digital camera, no mather it's resolution. The amount of
detail per area of say 100x100 pixels is significant better with
pictures of other cameras, so the detail of the posted 14n pictures
(the fruit in particular) does not live up to a captured resolution
of 14Mp.

While still being razor sharp, the n14 should be able to produce
much bigger pictures than for example 6Mp cameras. With that
particular bad fruit picture, I found a corresponding resolution of
roughly 1.4Mp. I doubt the n14 can't do better, the images must
have messed up somewhere.

Jack.

Geir Ove said:

Hello,

I first jumped on the booo bandwagon with the rest of the "mob". I
downloaded the Fruit Image (from DPReview) and looked at it at 1:1.
That is 4536 * 3024 pixels.

But wait, this isn't fair... My screen is 1024*768 and thus 4536 *
3024 represents a pictures 4 times higher and 4 times wider thand
my screen: At 1:1 the picture thus fills 4*4 = 16 screens! Thus,
one are looking at the picture at a tremendous MAGNIFICATION! If
you do the same with a 6 Mpix D60 picture, the magnification is
(3072 / 1024 x 2048/768) ~ 3 times.

If I open the picture in e.g. Paint Shop Pro and look at it when it
fills the screen, the picture is not that bad at all!

AND: The Fruit has, according to Jay (at Kodak) some sort of sugar
glazing poored over it in the upper part of the picture: Thus the
artificial look.

The IR pictures are another story: The Tree picture was very bad,
but this can be due to the photographer of post processing

Note: I am not a potential 14n buyer, but I thought we ought to be
fair to Kodak

Geir Ove

Click to expand...

Click to expand...

Click to expand...

Hi Pete,

I think it may be time to step back; I'm having a harder and harder time believing that this guy isn't just out trolling. Your points are well articulated, and overall the concept of objective comparison is so obvious that at this point, it's unlikely that someone who does not see it will be swayed by more arguments.

I shouldn't have replied myself, but I know I'll be a lot happier when this thread dies.

Pete Cockerell said:

Ludicrous. With apologies to Churchill, never has so much been
written by so few to so little effect.

Flaw 1: The 300dpi in the Kodak images is entirely arbitrary. It's
just a number put in by the manufacturer so that if someone naively
prints an image from, say, Photoshop, without doing any other size
manipulation (with or without resampling), they'll get a not too
appallingly large print (10" x 15" approx). Kodak could have easily
put 72 or 150 or anything else in the image. For example, my Sony
F707 uses 72dpi. Using your technique, I would actually have to
increase the size of the image to get the PS ruler to match up
with your physical one. Where does that leave me?

Flaw 2: Your thesis is that to judge what a picture will look like
when printed, you should resample it so that it looks the same size
as the print when displayed on a device (the monitor) that has at
best 10% of the areal resolution of the print device. Since when
does decimating an image by at least an order of magnitude show its
true quality? All your technique does is get the monitor to display
the size of an image when printed at some arbitrary
pixels-per-inch value.

I'm not sure why this is so hard to grasp, but the whole point of
displaying images 1:1 on the screen is to see how they will look
when printed on devices that have a much higher resolution (e.g.
a 2540 dpi Linotype). To say an image is fine because it looks OK
when resampled down to 10% of its original size borders on madness.

Incidentally, assuming your Windows monitor DPI setting is set
reasonably accurately, the Photoshop command "View ~ Print Size"
will do everything your involved techinque does, albeit with quick
'n' dirty real-time scaling and not proper resampling. But since
your technique doesn't actually have anything to do with previewing
print quality , that doesn't really matter.

Cheers,
Pete

--
http://www.pbase.com/pcockerell
http://www.peter-cockerell.net:8080/

Click to expand...

I see what you are saying: But I was not discussing WHICH METHOD people should use to look at their pictures! I was explaining what happens if you scale a particular picture to 1:1 size on a particular screen.

I will post the formula for this, and one can see how it works. Another poster tried to prove me wrong by using a stiched panorama as an example: But a stiched panorama works differently, as the formula and explanation will show when I finally get ready to post it.

At any rate: I expected people here to behave more civilized, but I guess I expected to much....

Geir Ove

Kenn Hwang said:

As anyone with experience with graphics displays will tell (and has
been telling) you, that is not an objectiveway to judge the quality
of a high res camera. Let's look beside the obvious point in
contention for now. For one, you're introducing an artificial
higher bound for images. When you downsample to 1024x768, a 3mp
image will look just as good as any 14mp image no matter how bad
the camera is, and both will be as good as any 500000x500000 image,
because you're essentially reducing every image to 1024x768.

So here's the problem: What if you don't like the camera's results?
Does changing your screen resolution make a bad camera better? It
really shouldn't, since what people are interested here is an
objective comparison of how cameras perform, pixel for pixel,
regardless of monitor resolution. Comparing images 1:1 by sensor
count (bayar interpolation notwithstanding) is the only way to
acheive an objective view. Your method actually INTRODUCES
magnification into the picture. Suddenly you're looking at one
image at 100% magnification, another image at 75%, and third one at
10%.

This leads into the next, probably the most important point:
absoutely nobody who buys a 14n will do so for full-screen viewing,
especially at such a paltry resolution as 10x7. Even people with
1900x1600 screens will not need a 14n for that kind of usage.
Photographers who choose cameras of this type will be printing
images at resolutions far-exceeding that of any consumer monitor.

You are of course free to feel whatever you wish, and if that's how
you judge a camera, more power to you. However, please be open to
the possibility that your method of measurement is wholely
inadequate for anyone who works with digital images bigger than
2mp. To them, your comments just seem ignorant, and that is why
you're getting such bad treatment.

Peace,
Kenn

Geir Ove said:

It is the same as watching a Wall Painting at 30 cm distance: You
get lost in the detail and can't see what it really is.

Geir Ove

Jack said:

The pixels on your screen are as big as with any other picture from
whatever digital camera, no mather it's resolution. The amount of
detail per area of say 100x100 pixels is significant better with
pictures of other cameras, so the detail of the posted 14n pictures
(the fruit in particular) does not live up to a captured resolution
of 14Mp.

While still being razor sharp, the n14 should be able to produce
much bigger pictures than for example 6Mp cameras. With that
particular bad fruit picture, I found a corresponding resolution of
roughly 1.4Mp. I doubt the n14 can't do better, the images must
have messed up somewhere.

Jack.

Geir Ove said:

Hello,

I first jumped on the booo bandwagon with the rest of the "mob". I
downloaded the Fruit Image (from DPReview) and looked at it at 1:1.
That is 4536 * 3024 pixels.

But wait, this isn't fair... My screen is 1024*768 and thus 4536 *
3024 represents a pictures 4 times higher and 4 times wider thand
my screen: At 1:1 the picture thus fills 4*4 = 16 screens! Thus,
one are looking at the picture at a tremendous MAGNIFICATION! If
you do the same with a 6 Mpix D60 picture, the magnification is
(3072 / 1024 x 2048/768) ~ 3 times.

If I open the picture in e.g. Paint Shop Pro and look at it when it
fills the screen, the picture is not that bad at all!

AND: The Fruit has, according to Jay (at Kodak) some sort of sugar
glazing poored over it in the upper part of the picture: Thus the
artificial look.

The IR pictures are another story: The Tree picture was very bad,
but this can be due to the photographer of post processing

Note: I am not a potential 14n buyer, but I thought we ought to be
fair to Kodak

Geir Ove

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Kenn,

I, like everyone else can see that the original 14n images sucked as far as image quality is concerned but that does not invalidate the beyond simple conclusions that I've brought up concerning viewing images at the proper scale to compare them. This is something I've been wanting to post for a while, the 14n images being released is only a catalyst for the writing of the post. The trolls are the people defending mostly the 1ds for some inexplicable reason believing that just because it costs 8k it must have the best IQ as well...the Japanese 14n samples show quite convincingly that isn't the case. I have been looking at post histories I know who's who, so don't accuse me of trolling thank you. I don't get the objections so far made without looking at or trying to repeat the results,(it's why I put the results out there ..sure they may be wrong..but so far I've seen no mathematical proof showing this just people calling me names because they disagree with what I've presented!) it makes perfect sense given my results that despite having different "arial resolution" (I think that was what Peter stated) the percieved sharpness between the resized image on screen will be similar (almost identical barring measurement error) to a print layed against the screen. Sure the screen represents it's onscreen image with pixels and the monitor has a completely different dynamic response from print, and at quite a different dot density per pixel from the printer used to make the print, but please tell me how that is any different from comparing a dye sub print from an inkjet print of the same image printed at the same physical size ? Do not both output devices have different "dot densities" ?? (The dye sub essentially continuous because it uses the heat and sheet wax method to create the image, the inkjet varies based on the size of the inkdroplets..in other words if you were to print 1 pixel from the image on each device, at magnification it would look like a continuous rectangular point on the page for the dye sub, for an inkjet the rectange would be composed of ink dots of various sizes depending on the print dpi chosen for the print settings) If variation in dot density between print devices is okay, and still allows you to compare them when they have the same physical dimensions on the print ..how is it any different when comparing directly to the screen which can be seen as only a different output device in that the "dot density" (in the monitors case created with screen pixels) is different?? I think people are overcomplicating dpi and resolution to the extreme...and it's not me, anyone is free to reproduce the steps I've used ..I even stated so at the end of it, you'll find it is quite consistent in it's results..now you can think that somehow things just turned out that way and that I fudged the steps to make what I wanted to happen come out (which is really foolish..I don't have that time to waste) or you can try it for yourself and see if it works. I suggest the latter before assuming the former.

Regards,

Kenn Hwang said:

I think it may be time to step back; I'm having a harder and harder
time believing that this guy isn't just out trolling. Your points
are well articulated, and overall the concept of objective
comparison is so obvious that at this point, it's unlikely that
someone who does not see it will be swayed by more arguments.

I shouldn't have replied myself, but I know I'll be a lot happier
when this thread dies.

Pete Cockerell said:

Ludicrous. With apologies to Churchill, never has so much been
written by so few to so little effect.

Flaw 1: The 300dpi in the Kodak images is entirely arbitrary. It's
just a number put in by the manufacturer so that if someone naively
prints an image from, say, Photoshop, without doing any other size
manipulation (with or without resampling), they'll get a not too
appallingly large print (10" x 15" approx). Kodak could have easily
put 72 or 150 or anything else in the image. For example, my Sony
F707 uses 72dpi. Using your technique, I would actually have to
increase the size of the image to get the PS ruler to match up
with your physical one. Where does that leave me?

Flaw 2: Your thesis is that to judge what a picture will look like
when printed, you should resample it so that it looks the same size
as the print when displayed on a device (the monitor) that has at
best 10% of the areal resolution of the print device. Since when
does decimating an image by at least an order of magnitude show its
true quality? All your technique does is get the monitor to display
the size of an image when printed at some arbitrary
pixels-per-inch value.

I'm not sure why this is so hard to grasp, but the whole point of
displaying images 1:1 on the screen is to see how they will look
when printed on devices that have a much higher resolution (e.g.
a 2540 dpi Linotype). To say an image is fine because it looks OK
when resampled down to 10% of its original size borders on madness.

Incidentally, assuming your Windows monitor DPI setting is set
reasonably accurately, the Photoshop command "View ~ Print Size"
will do everything your involved techinque does, albeit with quick
'n' dirty real-time scaling and not proper resampling. But since
your technique doesn't actually have anything to do with previewing
print quality , that doesn't really matter.

Cheers,
Pete

--
http://www.pbase.com/pcockerell
http://www.peter-cockerell.net:8080/

Click to expand...

Click to expand...

--

Whe you look at an image at 100%, the resolution of your monitor has no importance. You are looking at each pixel and each will fill the pixel projection of your monitor. It is like a set of grids, you can have a billion pixel or only 10 and still they will be displayed the same on you screen of any given resolution. The only difference is that you will be scrolling around for the 1 billion pixel image where your 10 pixel image will take a small square in you screen. Just relax and forget that you have been enlightened. It happens once in a while but then reality sets in.

Chunin Martinez

--
http://www.cimphoto.com