[RiceHigh]

As you can see in my histogram attached above, the camera would have preserved perfectly all the shadows and highlights without the +0.3EV compensation. Now with it, some highlights got burned.

But if your want the skin tone reproduced as it should look, i.e., the area of interest in this case, the highlight must be sacrificed. Mid-tone reproduction is also very important.

You cannot seriously blame the K-7 for having metered perfectly, can't you?

Don't get me wrong. I am not blaming the K-7 but just observe that there might be too much +EV compensation than it should be. At least that hadn't happen so much with the official K-m samples.

And who decides what exactly "mid-tone" is? Is it you? Every vendor has decided on a preferred arbitrary value here, a long time ago. And it better keeps this value from model to model. If you jump brand, just apply an EV correction which is a constant value between two brands. Where is the problem here?

No one should decide. But reproduction should be faithful. If you cannot view a more closely resembled scene on the monitor than it should be, then there is a problem. (putting aside the limited DR issue first, but only the subject/area of interest is in concern first)

[ftpaddict]

RH, worry not. The K-7 has -/+5 EV compensation. Just leave it on +0.3 at all times; there's plenty left to compensate, if you need.

[RiceHigh]

RH, worry not. The K-7 has -/+5 EV compensation. Just leave it on +0.3 at all times; there's plenty left to compensate, if you need.

Or +0.7EV??

My *ist D and K100D needed to be compensated from -0.5 to +1.5EV for quite some "normal" scenes, although I set them to +0.5 most of the time with DA lenses. So, what should I do? I really don't want the K-7 to be such again!

My 5D and my Canon/Nikon DSLRs at work don't have similar problem, why almost all Pentax DSLRs I have used (including the K20) alone should do more frequent compensations and in larger values? The K-m is a bit better but problem is still there.

[melander]

But if your want the skin tone reproduced as it should look, i.e., the area of interest in this case, the highlight must be sacrificed. Mid-tone reproduction is also very important.

So you admit that you want a portrait program, landscape program etc then. Cause how would the camera know that you would like to sacrifice highlights for better skintone?

Go buy yourself a PS

And wouldnt it also meen that Canikon that you say always behave perfectly, infact always overexposes at 0 EV? While pentax got perfect metering?

---

Last edited by melander; 05-22-2009 at 08:47 AM.

[falconeye]

Don't get me wrong. I am not blaming the K-7 but just observe that there might be too much +EV compensation than it should be. At least that hadn't happen so much with the official K-m samples.

@RH, I understand why you are concerned. But from the two samples, which actually even turn out to be just fine, you cannot deduce a tendancy for underexposure to happen again. But I am grateful that you watch out. Just, please, don't jump to conclusions.

But if your want the skin tone reproduced as it should look, i.e., the area of interest in this case, the highlight must be sacrificed. Mid-tone reproduction is also very important.

@RH, please don't be too philosophical here. We have one sample photo to discuss. It has +0.3EV compensation and a histogram which, at 0EV, would have been perfect w/o any clipping. Again, what exact problem do you see?

I really want you to concede that in this particular case, you may have been a bit quick with your critics. Really.

And in general, let's wait until we have a larger set of samples to play with.

[Isaac314]

i think they look good

[UK-K20-User]

Now they look nice

[falconeye]

Or +0.7EV??

We should still start from the assumption that K-7 matrix mode needs no EV compensation, even for your needs.

As far as center-weighted mode is concerned. You may make the following experiment:

If you have a Pro-level Canon and a Pentax, they both should meter identical in spot mode (for a grey color uniform surface). At least, they should do so.

Now, my K20D meters exactly identical in spot, center-weighted and matrix mode (for a grey color uniform surface). Now, as there is no norm for this, other vendors may meter brighter in center-weighted and matrix modes.

So, you may want to meter a grey color uniform surface in spot and center-weighted modes with your Canon and note the EV difference in exposure readings. On the Pentax, use this value as EV compensation in center-weighted mode and both brands should give you identical results in center-weighted mode. As I said, it is an arbitrary choice, anyway.

In matrix mode, things may be trickier. Actually, to expose a uniform surface as it would be 18% grey in matrix mode is a bit disappointing. Matrix metering should be able to detect that a surface is uniform and therefore, expose brighter. Here, I am with you, RH. But if they changed that for the K-m, than probably for the K-7 as well

[jeffkrol]

In matrix mode, things may be trickier. Actually, to expose a uniform surface as it would be 18% grey in matrix mode is a bit disappointing. Matrix metering should be able to detect that a surface is uniform and therefore, expose brighter.

What color/shade of uniform surface? Black, white, 1, 12, 18%grey, blue, orange???
Daylight, coal mine, ect???
No, a uniform surface will always defeat a monochromatic matrix meter.
Without pattern data the meter is dumb....
and will probably default to the metering standard....

[Digitalis]

Metering systems all have their weaknesses...ever photgraphed a black cat on white snow? or even better, a white cat on white snow...every exposure meter in christendom will underexpose that - all they see is grey.

[rparmar]

Wow, what a lot of discussion over some ok but not exciting images. I have no idea how anyone can tell much from these. I wait until photographers who I know and who have a large body of work I can inspect for comparison make test shots available, along with their own experiential comments.

And that's only because I won't be able to test the camera myself.

However I am extremely skeptical that results will be significantly better than the K20D, because I have a strange feeling we've reached close to the theoretical maximum IQ. Fewer pixels will result in less noise. More noise reduction will remove detail. But other than that...

[jaad75]

Didn't you notice they are NOT firmware 1.0? Exif header was edited, but if you look deeper it's still ver 0.2!

[RiceHigh]

Okay, let's wait and see. I agreed to agree!

Anyway, I shall have a production K-7 to test very soon once it arrives in HK. I hope I shall not need to cancel my order! There are many Pentax glass sitting there at my home waiting for the new Pentax body!

@RH, I understand why you are concerned. But from the two samples, which actually even turn out to be just fine, you cannot deduce a tendancy for underexposure to happen again. But I am grateful that you watch out. Just, please, don't jump to conclusions.


@RH, please don't be too philosophical here. We have one sample photo to discuss. It has +0.3EV compensation and a histogram which, at 0EV, would have been perfect w/o any clipping. Again, what exact problem do you see?

I really want you to concede that in this particular case, you may have been a bit quick with your critics. Really.

And in general, let's wait until we have a larger set of samples to play with.

[RiceHigh]

Didn't you notice they are NOT firmware 1.0? Exif header was edited, but if you look deeper it's still ver 0.2!

So, they (Pentax) intended to cheat??

Too bad.. And, pls teach us how to see the info *deep* inside the EXIF!

[ddhytz]

Or +0.7EV??
....

... I hope I shall not need to cancel my order! ...!

Maybe you want to consider that as those samples all have +0.3EV to +0.7EV exposure compensation!