Most wanted feature?

Me Too!!

Wadna is right though but it depends on how you look at it.
If you compare pixel for pixel then surely the *istD have pixels off higher quality but since the K7 has more then double the amount of pixel the total sensor is better on some points.
It's still CCD vs CMOS and they both have their own strengths.

I always liked the picture quality from my Ds more than the K20D .

But if you're viewing at 100% you aren't really comparing like-for-like, given the higher resolution. It's the same reason why some lenses that were regarded as great in the film days seem not as good on APS-C digital: the higher resolution sensor puts more demand on the lens performance. So with a given lens, the istD should look sharper than the K7 at 100%, shouldn't it?

[Honestly asking, all you camera physics types feel free to wade in ]

I'm not viewing at 100%.
Prints of photos taken by the *istD look sharper than those taken with the K-7. They also look sharper on my monitor at normal resolution (where I can see the entire image on the screen) which is one of the disappointing aspects of my K-7. The Samsung sensor is not a great one & it's noise performance is poor (also worse than the *istD which is understandable given the extra pixels crammed on the sensor). Which reminds me I must talk to my camera club mate about noise on his 18Mp Canon 7D.

The K-7 images are softer than the *istD - hence less detail.
Sure I haven't tested them side by side but was my initial & continuing reaction is the K-7 is softer.
I'm only telling you my experience without any axe to grind. I'm always sceptical of reviews as you rarely hear the bad news until the new improved model comes out. I gather the Canon 50D is a case in point.

I have no axe to grind either. My only dSLR ever is the K10D, so I can't make any comparison at all.

It just seems an extraordinary claim that a 6MP camera from 7 years ago can produce better images than a current model with over twice the pixels.

I'm genuinely interested in whatever it may be that contributes to the perception. And without testing side by side, and also considering the workflow that leads to the final image, understanding this is virtually impossible.

An interesting debate.

Here's my experience. I know have both a K20D and a K10D, having bought a K10D body recently to avoid swapping lenses in the field. I can detect no difference in image quality between the two. In fact the K10D doesn't suffer from the hideous forcing of the shutter time dark frame noise reduction which means waiting another 30 seconds after taking a 30 second exposure. Logn exposure shots on the K10D are no worse than the same on the K20D with the forced noise reduction.

Also my 8 year old Sony Cybershot 4MP Compact produced excellent pics which I doubt anyone would be able to distinguish from either the K10 or K20 under usual computer screen 1200px wide jpgs.

I want an excellent high ISO performance. Everything is is secondary.

In fact in 2nd place is faster AF performance.

This doesn't surprise me all that much. It's almost impossible to tell how sharp things are, once x-thousand pixels have been resampled to a screen width.

I'd hope the 100% crops can be distinguished, though.

I went for other: Higher Dynamic Range.

Not that I necessarily know what I'm talking about...

Robert.

One thing I forgot:-
greater dynamic range. I habitually run my K-7 on the max. dynamic range setting which means I have to use a min. iso of 200.

Damn, I knew I'd missed something important from the list

link
Where does the K-7 sensor rank? Below the K-5, K-x, K-r, K10D, K20D, K200D & the Km. That sums it up.

i voted other. i would like to see two new things.

my biggest issue is having to wait for sr to kick in. i would like it to be near instantaneous. what i like to do when shooting at large apertures is to just hold the shutter down and let the camera fire exactly when the focus is completely nailed. this minimises photographer and subject movement. i can't do that with an sr delay.

one really cool feature would be automatic focus adjust done perfectly by the camera while being lens specific. i would love this feature to be added, and it's fairly simple, because most cameras can focus very accurately in liveview. so just mount the camera on a tripod, liveview focus, let the camera flip the mirror down, and the camera can see exactly whether it was back focussing or front focussing and make the adjustment for that lens.

it could make front and back focus an issue of the past.