[boomklever]

I am enjoying my brand new K-5 plus some old M lenses I bought. Did several tests (and one concert shooting), so limited experience yet with the camera. But there's one thing I just noticed and it can be repeated: highlight in a low light scene differs significantly between 160 and 200 iso. Here's what I mean:





The pictures were taken at 1/25, 3.5-160 and 1/30,3.5-200. If you look at the bright area just above the lamp (the stairs section) you can see that this is much lighter at 160 ISO, but the rest of the image is nearly identical. At higher ISO's the images are nearly identical to the ISO 200 picture (except for increasing noise of course).

I repeated this 'test' for another composition, but with similar results: the 'ISO 160' clearly shows overshoot at a bright area.

Is there anybody who has an idea what could cause this strange behaviour?

Btw: my first posting, but this is truly a great forum, I already found lots of very very usefull information here. I love forums, especially the Pentax forum!

[enoeske]

Not sure whats happening here. I don't have a K-5 to test it (man, I want one).

Hypothesis: ISO 160 isn't a native ISO value and is actually ISO 200 pulled back 1/3 of a stop. This results in a loss of dynamic range more so than actual ISO 200. So, the ISO 160 images contains less detail, hence blown out whites, due to lack of DR compared to the ISO 200 image.

[sewebster]

200/160 != 30/25 ?

[skyoftexas]

200/160 != 30/25 ?

Yeah, I'm with him. Why did you use a slower shutter speed at a lower ISO speed? Mightn't that be the issue?

[enoeske]

Yeah, I'm with him. Why did you use a slower shutter speed at a lower ISO speed? Mightn't that be the issue?

As you lower the ISO, you need more light to make up for it, thus a longer shutter time.

ISO went from 200 to 160, which is -1/3 stop, so shutter needs to stay open longer.
Shutter went from 1/30 to 1/25, which is +1/3 stop, so equivalent exposure is aquired.

[Smeggypants]

As you lower the ISO, you need more light to make up for it, thus a longer shutter time.

ISO went from 200 to 160, which is -1/3 stop, so shutter needs to stay open longer.
Shutter went from 1/30 to 1/25, which is +1/3 stop, so equivalent exposure is aquired.

Makes sense to me.

[sewebster]

I'm just saying that those are not 1/3 stops.

[Pablom]

I don't think it's a dynamic range issue, just slight exposure differences

the camera doesn't expose exactly as metered but with some error due to the fixed increments in the values of different settings. this difference should be insignificant which is the case with this example

[enoeske]

I'm just saying that those are not 1/3 stops.

they are

1/15, 1/20, 1/25 1/30
100, 125, 160, 200

[Hound Tooth]

they are

1/15, 1/20, 1/25 1/30
100, 125, 160, 200

Those are "approximate" 1/3 stops. Mathematically, they don't actually correlate to 1/3. They're just convenient numbers that are used to represent 1/3 stops. Sometimes 1/3 stop of ISO doesn't correspond to the same amount as 1/3 stop of shutter speed, hence the slight differences in exposure. This is no big secret or defect, it's the way things have always been in photography.

[wombat2go]

Lamp Flicker ?. You are 50 Hz?
Either with cfl or incandescent , there are peaks of intensity each 10 millisec ( google cfl flicker etc to see a graph)
So at 1/30 your shutter is open for 33 millisec . The intensity-time integral depends on the phase between when your shutter "passed" the sensor and the lamp flicker.
Others will be better than me in commenting on how evenly the shutter exposes the across the sensor at that speed. And can I see by eyeball that the difference on the left side is greatest?

[Smeggypants]

Lamp Flicker ?. You are 50 Hz?
Either with cfl or incandescent , there are peaks of intensity each 10 millisec ( google cfl flicker etc to see a graph)
So at 1/30 your shutter is open for 33 millisec . The intensity-time integral depends on the phase between when your shutter "passed" the sensor and the lamp flicker.
Others will be better than me in commenting on how evenly the shutter exposes the across the sensor at that speed. And can I see by eyeball that the difference on the left side is greatest?

This is a very good point. I didn't think of the AC 'flicker'


Thinking about it I point my K-5 and K20D at the TV sometimes and the expiosure value changes rapidly. This will be according to the refresh rate of the screen - similar effect.


I guess the exposure meter of the camera has very fast rise and fall times

[wombat2go]

Yes, Smeg,
I was thinking about it further. I do not know how the k-5 shutter or metering works below X.
Boomklever, can you repeat this test in Av ?
K-5 is a major advance, and it is going to show up some low light difficulties like Boomklever identifies, I think.


The in-camera number crunching ability is presently limited by cpu heat/size and battery power and not by cost.
Eventually we will have a future k-5 that can output an image like the human eye sees from darkness to 12000 K.
Ahh ...Who is gonna disagree? .....

[Hound Tooth]

I think this thread deserves an award for containing the largest amount of incorrect, flawed and baseless guesses on a single page.

Fake ISO that results in a loss of dynamic range?!?
Lamp flicker at 1/30 shutter speed?!?
Limited CPU because of heat/size?!?

Are you people serious?!?!?!?

There's nothing magical about tiny exposure differences from one shot to the next when you're changing parameters!!! This is a complete non-issue!!!

edit: Obviously this is because of the magical gnomes that live inside the camera and operate the light meter. They require a daily feeding of Kraft Mac & Cheese to operate at peak efficiency. Otherwise these variances become more apparent. Just push a couple of pieces of Mac & Cheese pasta through the lens mount, squished up against the mirror. Then mount the lens and let the camera rest for 2 hours while the gnomes feast.

[Kryosphinx]

It's simply a slight exposure difference. It's pretty clear from the shadows that the first image is more exposed than the second one. It's just that the very subtle difference in the shadows is overshadowed (sorry) by that in the highlights.