[Zygonyx]

Much more than that. My KP also provides much more dynamic range and color fidelity at higher ISO values than would be expected at those levels.

That's correct, and would reasonnably favour the 26 Mpix versus the 32Mpix APS-C sensor.

[stevebrot]

What does it do then?

It is an electronic distillation of ancient incantations voiced by mystics in the far mountains of legend and serves to focus the aural energies of errant electrons to more useful tasks.


Steve

[stevebrot]

Sure it isn't. It makes coffee...

Really! For the last decade I have been fervently lobbying for K-cup support on Pentax cameras and it has been available since the K-70? All for want of the simple firmware fix?


Steve

(...time to buy an Olympus...err...well...something...)

[stevebrot]

Dynamic range is a weird one

Yep...depends on definition of dynamic range and at DXO, it means S/N ratio.


Steve

[normhead]

Yep...depends on definition of dynamic range and at DXO, it means S/N ratio.


Steve

Well isn't that an interesting approach?
But then they make noise suppressing software, so of course they would. They use the tech they've developed commercially as a measuring tool. That's inventive to say the least.

But that doesn't change the definition of dynamic range, essentially the difference between the amount of light it takes to alter the sensor during exposure, an exposure unit of 1, compared to the brightest the light can be and still produce different values on the sensor. Once you exceed the top end of your dynamic range, all values will be the max, 255 in an 8 bit system, even though the measured light can go up to as much as 5 EV higher than the upper limit, if you consider that a camera is topping out at 15 EV, and environmental DR can be as high as 20 EV..) Noise really has nothing to do with it. One of my complaints about DxO is their bastardization of conventional terms.

[taks]

(1) Would need some shutter re-engineering to have shutter parked in open mode.

(2) current mirror only has a smallish semi translucent area equal to AF focus point coverage so some light lost to optical viewfinder but not full frame.

(2) To implement, would need a semi translucent mirror a la what Sony did with their SLT Alpha cameras before Sony went down the mirrorless path. For Pentax this would split much more light between the optical viewfinder and the sensor assuming Pentax avails itself of the much wider array of focus points on sensors with pdaf functionality. Given Pentax desire for best experience from optical viewfinder, could they come up with a light split that gave good optical viewfinder experience while still giving enough light to the sensor for PDAF to work in low light given these sensors are designed for mirrorless (hence no light loss before sensor)? But for Pentax, mirror could still swing up for actual exposure so no loss of light for the recorded image, unlike how Sony implemented SLT.

Very interesting. Thank you so much for the detailed explanation.

[stevebrot]

Well isn't that an interesting approach?

I do wish I had a sarcasm emoji.

But that doesn't change the definition of dynamic range, essentially the difference between the amount of light it takes to alter the sensor during exposure, an exposure unit of 1, compared to the brightest the light can be and still produce different values on the sensor. Once you exceed the top end of your dynamic range, all values will be the max, 255 in an 8 bit system, even thought the measured light can go up to as much as 5 EV higher.) Noise really has nothing to do with it. One of my complaints about DxO is their bastardization of conventional terms.

Thanks for the excellent explanation of what dynamic range really is. (No sarcasm needed ) I too am not fond of DxO's appropriating existing technical terms to suit their methodology.


Steve

(FWIW, I am not fond of the approach over at Photons to Photos either...)

[Serkevan]

Really! For the last decade I have been fervently lobbying for K-cup support on Pentax cameras and it has been available since the K-70? All for want of the simple firmware fix?


Steve

(...time to buy an Olympus...err...well...something...)

Ah no, have sympathy for all of us in Europe. Nespresso support, at least via firmware!

[beholder3]

Well isn't that an interesting approach?
But then they make noise suppressing software, so of course they would. They use the tech they've developed commercially as a measuring tool. That's inventive to say the least.

But that doesn't change the definition of dynamic range, essentially the difference between the amount of light it takes to alter the sensor during exposure, an exposure unit of 1, compared to the brightest the light can be and still produce different values on the sensor. Once you exceed the top end of your dynamic range, all values will be the max, 255 in an 8 bit system, even though the measured light can go up to as much as 5 EV higher than the upper limit, if you consider that a camera is topping out at 15 EV, and environmental DR can be as high as 20 EV..) Noise really has nothing to do with it. One of my complaints about DxO is their bastardization of conventional terms.

I have to object.

Dynimc range pretty much is maximum contrast between values.
For audio data it is the span between the lowest signal (super silent) and highest signal (extra loud) which you can capture.
And the lower end is limited by noise. Think of putting a seashell to your ears and hearing your own blood pulse there. That is effectively noise on the lower end. Your ears maximum dynamic range is limited on the lower end by the point where you can no longer distinguish a real external sound (like a bee humming in 5m distance) from your self induced noise.

The upper end is when you hear something so loud it physicslly damages your ears.

With photos it is the same.

Often with photos people mix up input maximum dynamic range, that is what a sensor can detect, and data storage, like bits. And third the output dynamic range like prints or screens.
On the data side please keep in mind gamma curves which allow a few bits to contain big contrasts. So on 8 bit the broghtest white can be much more bright than just 256x the darkest value.

Photo sensor max DR is pretty much the difference between full well capacity (a sensor pixel is full of photons and can not count any more. it doesnt get whiter) and the point where you can discern image info in super darkness, the latter obviously is limited by noise.

[normhead]

Dynimc range pretty much is maximum contrast between values

Ya, you mean like in a 8 bit system the exponential difference between the lowest possible value and the highest possible values?
Contrast is the ratio between the lowest and highest value, on an 8 bit system the lowest recorded value is 1 (all zero does is tell you you didn't actually record anything), and 256.
I think that's what I said.

The upper end is when you hear something so loud it physicslly damages your ears.

With photos it is the same.

The upper end has nothing to do with ear damage, only with output volume. (or input volume.) DB is exactly the same kind of measurement EV is, an exponentially increasing value. With photos, in my imaginary 8 bit camera system, any value that would record above 255 is recorded as 255. It could be 5 EV higher using the scale, but the sensor EV DR doesn't cover the full range of the scene, so on my K-3 1EV - 13.9 EV can be recorded, anything over (say a light intensity that should be represented by say 1024), can only be recorded as 255.

On the data side please keep in mind gamma curves which allow a few bits to contain big contrasts. So on 8 bit the broghtest white can be much more bright than just 256x the darkest value.

That explanation creates its own illusion. In nature you can have contrasts up to 22,000:1. The best you can do for contrast on any output device that I know of would be 500:1, using Dynamic Contrast, which is really just another ordinary 150-300:1 TV with a shifting curve. You are actually going to have to compress you images to get them to look right on your very limited output device. If you start mapping two bits onto one to increase contrast, you'll vastly under-expose your shadows and blow out your highlights. What happens with all systems is you compress your DR into blacks and white suited to an 8 bit jpeg. 1-255. From there you could conceivably expand you values to create more contrast to create more DR, but you'd have to have an output device that could handle it. Some of us deliberately under-expose because it creates more realistic looking contrast in the low end., but we have to clip the high end in some scenes to do that. Some of us buy large output amplifiers and then never turn them up over 1or 2 because we get cleaner sound that way, the sound analogy only goes so far. With sound we usually have to limit the capacity of our output device. In digital photography we are always pushing our output device to the max to get the most realistic looking image.

The most contrast you will be getting out of a print will be about 120:1, newspaper or magazine 60:1. Some monitor's and TV s can slide the curve up and down to achieve contrast approaching 500:1 I believe, but not at the same time. They do it by shifting the curves so when they are at their brightest, they can't display their darkest dark at the same time as their brightest bright.


There is no output device that can come anywhere close to the 20 EV you can have outdoors. No input device either. You squeeze your 20 EV external scene's DR into 14 EV on your camera, which you then reduce to 9 EV max (and probably considerably less) on you computer monitor or TV or max. or 7 (and probably considerable less) EV on a print.

And mapping multiple pixels onto one to create contrast will create banding really quick. As will extending the contrast values by remapping pixels values.


It seems to me you're for the most part repeating what I said in your own words. Maybe you find your way of explaining it clearer, but there's couple flaws there that needed to be cleaned up.

[ramseybuckeye]

So now Sony comes out with a full frame 12 mp illuminated sensor. Do you think the K-new could have a sensor from that family? But is this an end to the megapixel wars? Really improved IQ, or more just to do video. I have not read it all yet but it is a big step in a different direction.

[ogl]

So now Sony comes out with a full frame 12 mp illuminated sensor. Do you think the K-new could have a sensor from that family? But is this an end to the megapixel wars? Really improved IQ, or more just to do video. I have not read it all yet but it is a big step in a different direction.

Unreal. A7S is special Sony line of cameras customized for video. And A7SIII is the THIRD camera of S series.

[Erictator]

So now Sony comes out with a full frame 12 mp illuminated sensor. Do you think the K-new could have a sensor from that family? But is this an end to the megapixel wars? Really improved IQ, or more just to do video. I have not read it all yet but it is a big step in a different direction.

Are you talking about the one in the new A7S III ? Isn't that the low noise video focused mirrorless? It looks nice for that purpose, but I don't see it as a stated goal for Pentax in light of their current offerings.

[ramseybuckeye]

Are you talking about the one in the new A7S III ? Isn't that the low noise video focused mirrorless? It looks nice for that purpose, but I don't see it as a stated goal for Pentax in light of their current offerings.

Yep, that's the one, by no means do I think Pentax is using that sensor, this is just to stimulate discussion in a speculative thread. We've all seen what Pentax can get out of a Sony Sensor.

[pres589]

I thought the point of this thread was to guess at the sensor Pentax will be using in the Knew, not sensors that are definitely not going to be used...