[Alfisti]

That sounds about right but i'd do 300ppi (you mean ppi not dpi right?) because most commercial labs demand 300ppi.

This is where I start getting confused (according to my wife a most common occurrence), but I'm assuming the following to get a fair comparison:
Both images shot in RAW at the same speed/aperture/ISO
both images with sharpening turned off
both images into PS3 at "full size" @240dpi
the K20 "full" image is 4672x3104 pixels
reduce the K20 to 65% to get an image that is very close to the "full" K100 image of 3008x2000 pixels.
then, then I take a 100% crop out of each of them of the same size (say 4x4")
Will that work, or am I still missing something,
Brian

Was shooting lemons a Freudian slip?
It is better to make lemonade from lemons than the other way around!

[rfortson]

Here's my experience going from K100D to K10D to K20D:

K100D - Loved the IQ and the controls were fine (never experienced anything different, so I didn't really know what I was missing)

K10D - Wow, bigger but I like the extra controls and I can set up my camera in many more ways. A little more resolution, but the noise at 800 and above just bugged me. Definitely worse (to me) than my K100D.

K20D - Great, same controls and layout as the K10D, so no learning curve. Plus, I have even more options for setting up the camera. And wow! The extra resolution, plus the lack of noise at high ISO is even better than my K100D. The AF performance is noticably better, and my SDM lenses seem to work better.

Bottom line, for 8x10 prints or viewing full screen (not 100%, but full screen), you'll only notice the difference in IQ between the K100D and the K20D when you get into higher ISOs. There will be some subtle differences, but in general it's not going to jump off the page/screen at you. However, at higher ISO, you'll start to see the difference, and if you crop or print large, then the K20D is hands down better.

[Alfisti]

You talking re-sampled at high iso? Pixel for pixel I'd be stunned if the 20D was better.

[soccerjoe5]

Anything posted on the web gets down to 72ppi and gets converted to sRGB, so we won't be able to tell the difference. What I noticed on the K20D is that it gets to keep a lot more detail in the dark compared to my Nikon system. In shots that I do get some gradation and tones on the K20D, I just get plain ol' black in my older gear.

Oh and print them and you'll see the difference

From my experience (I'm not saying I'm SUPER experienced), Pentax photos are just made for printing. The output is just fantastic in print compared to other brands. Those other brands feel like they're tweaked for on-screen viewing but when it comes to printing, Pentax just blows me away.

[Alfisti]

Anything posted on the web gets down to 72ppi and gets converted to sRGB,

Isn't the ppi figure and colour space dependent on the way the file was saved?

it doesn't just automatically convert to those settings once you view it ont he web.

[FHPhotographer]

From my experience (I'm not saying I'm SUPER experienced), Pentax photos are just made for printing. The output is just fantastic in print compared to other brands. Those other brands feel like they're tweaked for on-screen viewing but when it comes to printing, Pentax just blows me away.

Well, I have to agree. I just shot & printed the same image with the same settings (this time they are both f/4 @1/750 ISO200); all the prints were on my home Epson R1800 with Epson Ultra Matte paper and at best photo resolution.

My conclusions (now) are that:

at the K100 native resolution (8.4x12.5" @ 240dpi)...
there is no significant difference in IQ, in either the mico-resolution or "big picture" sense , even with the K100 print at 240dpi and the K20 scaled down to that size @360dpi;

at A3 size (11.7x16.5") to get a borderless print it gets interesting...
to get the K100 up to the full paper size it has to be enlarged by about 20% and dpi goes down to app. 180dp;
the K20 has to be reduced by about 20% and dpi goes up to app. 300dpi (these are approximate numbers, please don't flame me to point it should really be something like 81.794% or 117.962% or some such thing);
and the outcome is, as you would guess, that the K20 print shows significantly better resolution, color, shadow detail etc.

In summary (for printed output) in my humble opinion ... if you print no larger than about 8x12"" @ 240dpi (approximately) the K100 matches the K20 in IQ, so you're just paying for K20 ergonomics and handling ; if you print 11x14" or larger, or, crop heavily (20%+ of the full image), the K20 will give you better prints and then the ergonomics+IQ justifies the cost,
Brian

[kenyee]

In summary (for printed output) in my humble opinion ... if you print no larger than about 8x12"" @ 240dpi (approximately) the K100 matches the K20 in IQ, so you're just paying for K20 ergonomics and handling ; if you print 11x14" or larger, or, crop heavily (20%+ of the full image), the K20 will give you better prints and then the ergonomics+IQ justifies the cost

Not too surprising...the K10D's 10Mpix sensor cost 1 stop from the old 6mpix sensors. The K20D's CMOS sensor got it back. And it's been widely mentioned that most people really only need a 6MP camera, but no one ever wants a camera that has small numbers than everyone else

[Alfisti]

hehe you've discovered what we thought you would

You mean ppi not dpi BTW, there is a difference.

[Marc Sabatella]

There is essentially no situation in which you ever have to think about the ppi setting of the file. All you need to be concerned about is the number of pixels. If you display an image something on a monitor, it will be about 72dpi *regardless* of the resolution setting in the file. That's because a number in a file can't change the number of dots per inch in the monitor itself. Unless you specifically ask your display software to make the picture bigger or smaller (like if you ask it to display it "full screen"), it's 72dpi. And even if you ask it to dispaly full screen (and thus either interpolate new pixels or throw away existing ones during display), the resolution figure stored in the file remains irrelevant.

Similarly, if you have a 300dpi printer, your image *will* print at 300dpi regardless of the resolution setting in the file, assuming you have enough pixels to support that. If you've got 2000x3000 pixels in your file and you print it at any size less than about 7x10 (2000/300 x 3000/300, to be exact), then it *will be* 300dpi, regardless of what number is stored in the resolution field of the image. If your image doesn't have enough pixels to support printing at 300dpi at the given size, then the software will interpolate new pixels, so you'll have less effective resolution.

So anyhow, Brian, while your basic conclusion is correct, I need to quibble with the terminology. Cameras don't have native resolutions that can be expressed in dpi; that's a meaningless statement. It has a certain number of dots (the "d" in dpi), period. The camera has no idea how inches big (the "i" in dpi) the print is going to be, so there is no way to say it has a "native" dpi. Only when you choose a print size in inches can one calculate dpi. Similarly, the trip from Paris to London does not have has a "native" speed in MPH. It has a distance in miles, period. Only when you actually make the trip and time it can you compute a speed in MPH.

Only printers (or other output devices like monitors) have "native" dpi figures - and that's determined by the physical numbr of ink droplets it can actually spray onto an inch of paper, not anything to do with the camera or the image.

But that aside, the basic conclusion is, as I said, correct. If your *printer* has a native resolution of 240dpi, then it doesn't matter how many extra pixels you started with when printing 8x12": what is going to show on the print is going to be 8*240 by 12*240 pixels - which is to say, both the K100D and K20D will be throwing away pixels so there should be no discernible difference. Only when you get to print sizes big enough that you are no longer throwing away pixels will the extra pixels on the difference matter.

On the other hand, if you have a printer with a native resolution of more than 240dpi, than you might see the benefit of the K20D at smaller sizes. You might, however, need a magnifying glass to see it.

[asdf]

I'm not seeing an IQ difference to justify the $700

What about IQ at higher ISO, with noise-reduction off? Do you see less noise with the K20D?

[dave sz]

I'm running some comparisons between my K100D and my new K20.

First, I really like the ergonomics of the K20: great VF, two-dial setup, larger back screen, etc. I feels good, it shoots good, great resolution for cropping, etc, but here's the issue... I'm not seeing an IQ difference to justify the $700

Two examples: both images in RAW, first the K100 without any other editing, then the K20, both with mirror lock up, tripod, IS off with a the Tamron 90 DI @ f/4; or is it the other way around?

Same basic outcome on 8x10" prints with a 43mm mounted ; however, I start seeing some very subtle differences up close in 10x15" prints, but from a normal viewing distance, no objective differences. Am I doing something wrong, missing something, or is this a reasonable outcome?
Comments?


Brian

I dunno, the detail in the lemon "dimples" is more noticeable to my eyes in the k20 sample...

[Mister Guy]

I just mentioned this in another thread oddly enough. Upgrading your camera should be done on the nominal case. Your selection of scene was done in a case where the K100 can perform admirably well.

Do the same test with a fast lens in an candle lit room, hand held, with the ISO cranked up. Do you still get the same results?

[Alfisti]

I dunno, the detail in the lemon "dimples" is more noticeable to my eyes in the k20 sample...

That's just the difference in WB clipping the shadow and yellow channel a little.

[Class A]

Pixel for pixel i'd be stunned if the Sony 6mp sensor is not in front of anything else out there today from 10 to 15 MP on APS-C.

1. Pixel performance is not the deciding factor for the IQ of the whole image. It also matters how many pixels you have. Even if 4 x # pixel means that each pixel has more noise, when you combine 4 pixels in order to make a fair comparison, the noise gets divided by 4 again. Worse noise figures for higher MP sensors are "per pixel" but in terms of the whole image the noise will typically be as good or better.
2. Sensor technology develops and per pixel performance is increased constantly. A 6MP sensor today would outperform a 6MP from 5 years ago.

The one advantage of a larger MP count is you get a larger image to help with cropping or to help with printing large images, otherwise it is highly likely that IQ will actually be worse.

It is not true that IQ will get worse because of the averaging effect of many pixels if one compares identical print sizes with each other.

Pixel for pixel I'd be stunned if the 20D was better.

But pixel for pixel comparisons are meaningless.

Isn't the ppi figure and colour space dependent on the way the file was saved?

It is. Also, as Marc explained most image software / output devices ignore the dpi information in image files.

If you display an image something on a monitor, it will be about 72dpi *regardless* of the resolution setting in the file.

I agree with everything you've written, but whether it is 72dpi or not depends on the monitor (the pixels size of the monitor). Many monitors may approximate 72dpi but I have used 96dpi monitors as well.

Some browsers (Firefox with default settings) actually scale images in order to normalize the dpi of the monitor. I don't know whether Firefox reads the dpi of the image or just aims at 72dpi but it will scale images depending on its settings. I changed my Firefox settings because the scaling makes any judgement of image sharpness impossible. All images (scaled up) look soft. It is typically much better to leave the image size alone (let it depend on the output device resolution) than to scale the image in order to achieve a certain size.