[stevebrot]

The generally accepted K-3 DR is 13+EV. If I were to try and remember an exact number i think it was 13.3.
But I;ll change my mind if you give me a definitive reference. A K-5 is 14 EV and a K-1 is almost 15.
Standard every day films were 7 EV, You are probably way under selling the K-3.

Yes, it depends on how it is measured. The "within frame" DR will never be greater than 14 stops due to bit depth, even when the response limits in terms of range of light may be greater. I know you know this stuff, but for benefit of those who are not familiar, the article below is helpful...

Raw bit depth is about dynamic range, not the number of colors you get to capture: Digital Photography Review

DXOMark uses a variant definition of dynamic range that will yield numbers higher than that supported by capture bit depth, but whether that translates to the real world is a matter of some discussion.


Steve

[BarryE]

Yes, it depends on how it is measured. The "within frame" DR will never be greater than 14 stops due to bit depth, even when the response limits in terms of range of light may be greater. I know you know this stuff, but for benefit of those who are not familiar, the article below is helpful...

Raw bit depth is about dynamic range, not the number of colors you get to capture: Digital Photography Review

DXOMark uses a variant definition of dynamic range that will yield numbers higher than that supported by capture bit depth, but whether that translates to the real world is a matter of some discussion.


Steve

I was about to say that using DXOMark to make a point is dubious, but then you saved yourself towards the end of your sentence ;-)

[normhead]

Yes, it depends on how it is measured. The "within frame" DR will never be greater than 14 stops due to bit depth, even when the response limits in terms of range of light may be greater. For benefit of those who are not familiar, the article below is helpful...

Raw bit depth is about dynamic range, not the number of colors you get to capture: Digital Photography Review

DXOMark uses a variant definition of dynamic range that will yield number higher than capture bit depth, but whether that translates to the real world is a matter of some discussion.


Steve

Dynamic range is how many times the lowest level of light is given as 1, compared to the brightest captured before you blow your highlights. 13 EV is 2 to the 13th times the lowest value captured. There is no saying how that is mapped onto the bits by the camera hardware. Dynamic range can certainly exceed 14 bits with a 14 bit depth sensor. Since the camera both captures the images and decides how to map what it captures, there's really no limit,

If a sensor captures a value of 20 EV, which is possible in nature, and maps it onto a 14 bit file, it's still captures 20 EV,

If you have a two bit sensor, and the lowest value it captures is 2 to the 13th brighter than the lowest value it captures, it's captured 13 EV with two bits. There will be no colour depth, and you'd essentially be limited to black, white, and two shades of grey, but it would still be 13 EV.

I don't know what DPR says and I don't care. They quite frequently don't have clue what they are talking about.

[pentageek]

The generally accepted K-3 DR is 13+EV. If I were to try and remember an exact number i think it was 13.3.But I;ll change my mind if you give me a definitive reference. A K-5 is 14 EV and a K-1 is almost 15.Standard every day films were 7 EV, You are probably way under selling the K-3.

8.9 EV comes from review published by optyczne.pl a.k.a. lenstip.com - unfortunately this particular review is available only in Polish. They measured dynamic range for various signal to noise ratios, and for lowest SNR and ISO 100 results were very similar to numbers you quoted (13.4 EV). 8.9 EV was measured @ISO 100 for high image quality, arbitrary set at SNR = 20dB.

[normhead]

8.9 EV comes from review published by optyczne.pl a.k.a. lenstip.com - unfortunately this particular review is available only in Polish. They measured dynamic range for various signal to noise ratios, and for lowest SNR and ISO 100 results were very similar to numbers you quoted (13.4 EV). 8.9 EV was measured @ISO 100 for high image quality, arbitrary set at SNR = 20dB.

That's just confusing.
Amazing how making things more complicated than they need to be rarely helps you get the right answer.

[Kerrowdown]

Some of my sunsets are dialled back to -3 EV

I've quite often seen me dial back -2 stops... to reduce or eradicate blinkies, routinely on the K1s I have -.7 dialed in as standard.

[stevebrot]

If a sensor captures a value of 20 EV, which is possible in nature, and maps it onto a 14 bit file, it's still capture 20 EV,

If you have a two bit sensor, and the lowest value it captures is 2 to the 13th brighter than the lowest value it captures, it's captured 13 EV with two bits. There will be no colour depth, and you'd essentially be limited to black, white, and two shades of grey, but it would still be 13 EV.

Got it. I was assuming linear sensor response with white clipping (saturation) in the 2-bit system above at BIN 11 (2^3-1) and black clipping (lowest sensitivity) two stops lower at BIN 01 (2^1-1), EV 13 and EV 11 respectively. EV values below EV 11 being detected as black. Your explanation makes sense, though a non-linear response is problematic.


Steve

---

Last edited by stevebrot; 08-16-2018 at 03:26 PM.

[stevebrot]

I was assuming linear sensor response with white clipping (saturation) in the 2-bit system above at BIN 11 (2^3-1) and black clipping (lowest sensitivity) two stops lower at BIN 01 (2^1-1), EV 13 and EV 11 respectively. EV values below EV 11 being detected as black.

Or alternatively, EV 0 at BIN 01 and EV 2 at BIN 11 with values higher than EV 2 clipped despite sensor capabilities as high as EV 13. My understanding is that the current product low noise sensors are managed in just that fashion where sensor output is clipped by bit-depth while still well within linear range on the high end.


Steve

[RAART]

I usually expose for highlights and then just raise the exposure and bring out shadows. Works 80% of the time and the rest is just where I have to improvise. I usually shoot in aperture priority mode or manual. When I am in Aperture Priority I just use spot metering point the camera into highlights (sky, clouds, etc) press AE-L and this will lock the exposure for some time (maybe 20-30 sec) then recompose and press the shutter button. When you get familiar where button is located then is very easy, especially if you use back AF button, just next to it on K-3. Hope this helps.

[Excalibor]

Thanks, everybody, for your answers: I have gained a lot of insights and also have been remembered some things I had kind of forgotten.

For starters, I am starting to learn to use parametric masking in Darktable, and it's already being productive... Nice! :-)

Then, I have been exposed to some very useful advice for using flash and light, in particular, I have been reminded of reflectors! I have a very nice set of reflectors I have never used... OK, it's not particularly practical to carry them with you for candid photography, but on some occasions, like family reunions, where I know in advance my nephew will be there, it's a very good idea!

As for metering, I now see that is matrix metering cannot solve a scene, the spot metering will only work on the subject I spot on, but I'm gonna lose some information... It's interesting that you have to routinely compensate exposure down, I have noticed that my camera tends to underexpose, so I have a permanent +0.3 EV that goes up to +0.7 sometimes...

Oh, well... Thanks a bunch for the help and interesting info and ideas!

[twilhelm]

Thanks, everybody, for your answers: I have gained a lot of insights and also have been remembered some things I had kind of forgotten.

For starters, I am starting to learn to use parametric masking in Darktable, and it's already being productive... Nice! :-)

Then, I have been exposed to some very useful advice for using flash and light, in particular, I have been reminded of reflectors! I have a very nice set of reflectors I have never used... OK, it's not particularly practical to carry them with you for candid photography, but on some occasions, like family reunions, where I know in advance my nephew will be there, it's a very good idea!

As for metering, I now see that is matrix metering cannot solve a scene, the spot metering will only work on the subject I spot on, but I'm gonna lose some information... It's interesting that you have to routinely compensate exposure down, I have noticed that my camera tends to underexpose, so I have a permanent +0.3 EV that goes up to +0.7 sometimes...

Oh, well... Thanks a bunch for the help and interesting info and ideas!

The biggest reason you see most of us expose -.3 to -1 is because blown highlights are gone forever, shadows on the other hand can be brightened in post with surprising results.

[Excalibor]

The biggest reason you see most of us expose -.3 to -1 is because blown highlights are gone forever, shadows on the other hand can be brightened in post with surprising results.

I understand, but I have "always" been taught to expose to the right, because the left is less and less pixels of the sensor, and the noise comes where there should be none... I concede this is easier in mirrorless cameras with permanent histograms, but still desirable Ina DSLR.

[photoptimist]

I understand, but I have "always" been taught to expose to the right, because the left is less and less pixels of the sensor, and the noise comes where there should be none... I concede this is easier in mirrorless cameras with permanent histograms, but still desirable Ina DSLR.

The technique of spot-metering the brightest part of the scene and boosting exposure a couple of stops is an ETTR method because it ensures that the brightest part of the scene is as far to the right as it can go without saturation and loss of detail.

Some of this depends on how you use your camera. Broadly speaking, there are two strategies here:

1) Make a Pretty Picture: make a decent SOOC JPG that does not need post processing. Such images may look well exposed for the subject but they can suffer from blown highlights if the scene has too much contrast (e.g., backlit or shadowed subject under a bright sky).

2) Collect the Best Data: capture as much high-quality data about the scene as possible to preserve as much detail in all tonal zones. Such images may look poorly exposed or flat in the thumbnail but post processing brings out all the details in both highlights and shadows.

[normhead]

I understand, but I have "always" been taught to expose to the right, because the left is less and less pixels of the sensor, and the noise comes where there should be none... I concede this is easier in mirrorless cameras with permanent histograms, but still desirable Ina DSLR.

That will produce consistently flat images. Under exposing and pushing in post processing was the way to increase contrast even in film. It's equally true in digital. As long as you don't underexpose so much you lose detail. I like my histogram graph to start with at the left border of the histogram. Where the right is, I don't really care if the DR of the scene doesn't fill the histogram. That will give you your best contrast. Exposing to the right can produce washed out images if you are leaving he left side of the histogram. But it's matter of preference.

Some people prefer smooth low contrast reproduction with many gradations to higher contrast snappier images.

[BarryE]

Thanks, everybody, for your answers: I have gained a lot of insights and also have been remembered some things I had kind of forgotten.

For starters, I am starting to learn to use parametric masking in Darktable, and it's already being productive... Nice! :-)

Then, I have been exposed to some very useful advice for using flash and light, in particular, I have been reminded of reflectors! I have a very nice set of reflectors I have never used... OK, it's not particularly practical to carry them with you for candid photography, but on some occasions, like family reunions, where I know in advance my nephew will be there, it's a very good idea!

As for metering, I now see that is matrix metering cannot solve a scene, the spot metering will only work on the subject I spot on, but I'm gonna lose some information... It's interesting that you have to routinely compensate exposure down, I have noticed that my camera tends to underexpose, so I have a permanent +0.3 EV that goes up to +0.7 sometimes...

Oh, well... Thanks a bunch for the help and interesting info and ideas!

Perhaps I could have added a ref to a style of reflector I like - Lastolite's Trigrip (Trigrip Reflector Large 120cm Silver/White | LL LR3731). One hand on this and the other holding a remote shutter release. These large reflectors can throw a lot of light back onto a subject and are easy to use one handed.