[codiac2600]

I am unhappy but I must repeat myself:

Do not compare ISO IQ between images scaled to different sizes. Don't. It's useless. Just enjoy the nice dog but draw no conclusions whatsoever. Period.

This is my original statement: https://www.pentaxforums.com/forums/188192-post19.html.

You can draw plenty of conclusions from these. I have 8X12 prints made of some of these and they don't look any different from what you're seeing here and how often does a basic user print larger than that? Showing 100% crops is stupid because you don't stare at a print right at your nose, you don't have a monitor big enough to show a full res shot from the K20D to enjoy it, so why should I show images at 100%???? I'm not proving anything except that everyone was scared to show off ISO 1600 even smaller than this from the K10D because of noise and I'm freely showing off images I wouldn't blow up from any other camera in the class.

[Gooshin]

I am unhappy but I must repeat myself:

Do not compare ISO IQ between images scaled to different sizes. Don't. It's useless. Just enjoy the nice dog but draw no conclusions whatsoever. Period.

This is my original statement: https://www.pentaxforums.com/forums/188192-post19.html.

caveat!

one only cares about what one sees,

if the end product has less visable noise than a competitor, you have achieved your goals.

[nostatic]

I'm just happy to see images instead of yet another "comparison table". I don't hang numbers on my wall...

[WMBP]

Showing 100% crops is stupid because you don't stare at a print right at your nose, you don't have a monitor big enough to show a full res shot from the K20D to enjoy it, so why should I show images at 100%????

Well, falconeye has a point--but it's an engineer's point, not a photographer's point.

There are two different ways to "worry" about noise. The engineer worries about it as a kind of math problem. The engineer is interested in noise as a number representing a certain number of the originally captured pixels. This means looking at everything at 100%, where you can see each original individual pixel--whether good or noisy. I think this is a perfectly legitimate way to think about it, so long you understand that you're doing math, not photography.

But of course, the rest of us aren't engineers. I don't personally give a rat's ass whether the number of noisy pixels per 100 originally captured is higher in the K20D or the K10D. Actually, I don't care if there are pixels involved at all. It might be magic for all I know. What matters to me is how the photos I post on the web look, and how the photos that I print out look. If the K20D's photos start out noisier than the K10D's, but end up looking less noisy due to down sampling, then so what?

It really boils down to what you're interested in. If you're interested in the camera as a technical problem, then falconeye is right. I genuinely do respect this interest. I want to emphasize that I'm not knocking the engineers. We owe it to them that the cameras we use work so wonderfully.

On the other hand, if, like me, you ultimately don't really like cameras very much at all, you are much more interested in photographs, then you think of noise as a different kind of problem.

Will

[falconeye]

Well, falconeye has a point--but it's an engineer's point, not a photographer's point.
[...] If the K20D's photos start out noisier than the K10D's, but end up looking less noisy due to down sampling, then so what?

Thank everybody for letting me alive

Well, I would not have said a word if the thread wouldn't have named "K20D noise handling" in section "Pentax DSLR Discussion". So, it is about a technical performance. codiac2600 has started many similiar threads, with similiar names, and now I have started to react. It is to give everybody here a chance not to be fooled.

And no offense intended. I do not think that this was codiac's intention. He most likely is as impressed about his camera's ISO performance as everybody else. His samples simply cannot underline his enthusiasm because of this nasty "math argument"

Somebody please show me a K10D image scaled to 25% or a K20D image scaled to 20% (to make them comparable) which are correctly exposed, not pushed for exposure or contrast, which exhibit any visibly noise when looked at on a computer screen.

Because this is the "photographer's" version of my argument:


Images from a DSLR, downsampled to fit a normal-sized web page, show no noise whatever be its ISO or make.


So, why show them here other than for artistic reasons?

[codiac2600]

Thank everybody for letting me alive

Well, I would not have said a word if the thread wouldn't have named "K20D noise handling" in section "Pentax DSLR Discussion". So, it is about a technical performance. codiac2600 has started many similiar threads, with similiar names, and now I have started to react. It is to give everybody here a chance not to be fooled.

And no offense intended. I do not think that this was codiac's intention. He most likely is as impressed about his camera's ISO performance as everybody else. His samples simply cannot underline his enthusiasm because of this nasty "math argument"

Somebody please show me a K10D image scaled to 25% or a K20D image scaled to 20% (to make them comparable) which are correctly exposed, not pushed for exposure or contrast, which exhibit any visibly noise when looked at on a computer screen.

Because this is the "photographer's" version of my argument:


Images from a DSLR, downsampled to fit a normal-sized web page, show no noise whatever be its ISO or make.


So, why show them here other than for artistic reasons?

Your math DOES NOT hold water. Now lets see your math take a picture... thats a tough one.

[nostatic]

Well, I would not have said a word if the thread wouldn't have named "K20D noise handling" in section "Pentax DSLR Discussion". So, it is about a technical performance.

I don't think that is the only interpretation. To me, I don't see how "Pentax DSLR Discussion" only = "technical performance" talk. I kinda assumed that people would use their DSLR to take pictures rather than as lab toys.

Oh yeah...they do take pictures. Of newpaper print across the room. Silly me

I understand the need and desire to talk about "the maths." But it seems that people get really out into the weeds and make some rather silly extrapolations based on numbers with no connection to an actual picture being created.

I've got a couple of photos from HK that I'll end up printing at about 18x32 initially, and later at about twice that for an exhibition (although with my Leica D-Lux 3 rather than the K10d). I'll take a picture of those prints reading a newspaper so people can calculate the noise issues

[falconeye]

Your math DOES NOT hold water. Now lets see your math take a picture... thats a tough one.

Oh, I see. Discussions are over when one party gets polemic.

That's a pitty because what I say isn't math, it is about how the size at which images are to be rendered influences a photographer's decisions including ISO considerations.

And for the photo, look up my gallery if you need some. BTW, one has a noise problem because it is cropped, boosted for contrast and from a DS

[mattdm]

I'm not proving anything except that everyone was scared to show off ISO 1600 even smaller than this from the K10D because of noise and I'm freely showing off images I wouldn't blow up from any other camera in the class.

Yep. I see distracting levels of noise in ISO 1600 pictures from my K10D scaled to this size. (The main subject is correctly exposed; the shadow areas are visibly noisy.) I'm happy to see the K20D definitely performs better.

[wowtip]

Being able to (hypothetically) print a K20D 3200 at the same max size as K10D 800 - 1600 ISO shot seems to me like a real win.

Add to that the bonus ability to print low ISO images up to something like 28x20 or crop the **** out of an image when using a good lens.

[falconeye]

Yep. I see distracting levels of noise in ISO 1600 pictures from my K10D scaled to this size.

In my humble self I must apologize then.

I don't have a K10D but thought that it would at least be as good as a *istDS when images are scaled to the same physical size.

I showed in the other thread https://www.pentaxforums.com/forums/189878-post28.html that there aren't no distracting levels of noise from a DS in ISO1600 pictures scaled to this size. I even challenge people to tell the difference between ISO200 and ISO1600

So, I'm glad I skipped the K10D and jumped to K20D

[distudio]

In my humble self I must apologize then.

I don't have a K10D but thought that it would at least be as good as a *istDS when images are scaled to the same physical size.

I showed in the other thread https://www.pentaxforums.com/forums/189878-post28.html that there aren't no distracting levels of noise from a DS in ISO1600 pictures scaled to this size. I even challenge people to tell the difference between ISO200 and ISO1600

So, I'm glad I skipped the K10D and jumped to K20D

ISO 800 noise is good too:



Check the EXIF for a laugh. Kind of makes your point.

[benjikan]

You can draw plenty of conclusions from these. I have 8X12 prints made of some of these and they don't look any different from what you're seeing here and how often does a basic user print larger than that? Showing 100% crops is stupid because you don't stare at a print right at your nose, you don't have a monitor big enough to show a full res shot from the K20D to enjoy it, so why should I show images at 100%???? I'm not proving anything except that everyone was scared to show off ISO 1600 even smaller than this from the K10D because of noise and I'm freely showing off images I wouldn't blow up from any other camera in the class.

I totally agree with you. When an image of mine is published in a large format magazine, I don't shove the image in to the face of the viewer. They view the images from around 25 to 30 cm away. When I exhibiting my images in a gallery, the viewer moves back to a comfortable distance to view the image depending on the size. Chris, in real world applications even the 3200 could be blown up to 40 x 60 inches and look wonderful. It would probably be viewed from at least six feet away.

Pixel peeping isn't very esthetic for the image, to say the least.

Ben

[falconeye]

Pixel peeping isn't very esthetic for the image, to say the least.

Ok, I feel guilty. Must be the German engineer in me, as may be proven here: DigitalFotoNetz.de :: Thema anzeigen - K20D / K10D / K100D Super Rauschverhalten . Probably the most exhaustive analysis on this thread's topic so far

But I agree with Ben anyway. Say that the K20D shoots high-res, noise-free images. But don't try to prove it with web-sized images.