[reh321]

This MK2 could have had the 42MP chip if R&D money for social media OOC-jpg accelerator was used in negotiations with Sony.

So, get a Sony already and see if their camera will get the essence of dead fur you want. I've been hearing for months rumors that Sony has added layer(s) to their sensors that reduce noise - maybe doing it directly at the source works better.

---------- Post added 05-09-18 at 05:05 PM ----------

As to all this high and ultra high ISO technobabble, we seem to have forgotten that ISO settings in these ranges are only to be used for photographing UFO's, Bigfoot and Area 51. Since I have never seen an image of any of these things on this site I figured that Pentaxians are just not interested in them.

The picture I posted just above your post was taken at ISO=1600, and the raccoon was just barely discernible. To get a "good" picture I would have needed ISO=6400 at minimum. Fortunately, gimp's "auto WB" button was able to salvage something, but the picture would have been much better, and lower stress for me {I am uncomfortable depending on PP to "fix" things}, with a camera that could give me low-noise pictures at ISO 6400 or higher.

Incidentally, I posted this picture in Rupert's squirrel thread, with that thread in mind when I took the picture
https://www.pentaxforums.com/forums/76-non-pentax-cameras-canon-nikon-etc/26...ml#post4272474
so it isn't something I did just because of this thread - it is the sort of thing I do on a regular basis.

---

Last edited by reh321; 05-09-2018 at 02:14 PM.

[stevebrot]

Do you actually know anything about that, or is what you say speculation?

It's called keeping the thread alive and it works!


Steve

(...unwitting foil...)

[Rico]

@normhead
No way I will waste more money on bad usability. It is good for something but less for something else. I already used over 10k€ in Sony gear during the last 5 years only to find average user experience as a whole. Pentax gives average UX for a lot less money wasted. After all, digital cameras are necessary evil to get some shots not possible with large format.

But if you want low-noise digital camera then get a Sony. Or some 50MP CMOS MF offering. K-1 is not it above base ISO.

This is just not true MJKoski. In fact using the DPR comparison in their review the K-1MKII has much better color and far less noise at all ISO levels not only over the K-1 but also over both the Nikon and Sony BSI sensors.

An easy way to see how much better the K-1MKII is over all three of the other cameras is to look at a spot that only shows a portion of the grey background. Then go up the ISO levels. The K-1MKII holds a better continuous grey tone with less noise than all three other cameras. By ISO 6400 the K-1, Nikon, and Sony are all noisy messes. Do the same on the color charts and black and white targets and you will see the same results the K-1MKII having less noise more continuous clean color.

Two things are interesting the K-1MKII is better by a large margin the higher the ISO. Second the K-1, Nikon and Sony all look very similar with the same relative noise pattern develop across the image.

[Ian Stuart Forsyth]

Then you are aware that "shot noise" is only significant when photon flux is exceedingly low and essentially disappears when that is not the case. At least, that was how the discussion of Poisson distributions went when I took "Prob 'n Stat" back in the stone age.

It really depends on why and how it becomes significant, even when we are shooting at base iso it really is not that hard to find the tell tail signs of shot noise in a photo.

The biggest hurdle is that for the first 3 stops of the tonal range(zone10-8) in a scene is using 75% of the space a sensor can store as light information( signal).

take this photo



If you take a look at 100% you can see shot noise creeping in easily, on the right the image is take using the cameras meter set on middle grey and the right one stop larger exposure

https://photos.smugmug.com/Temp/Temp/i-cZF7J45/0/c819217b/O/_1710418%20crop.jpg

The reason why is how light information is captured, if we expose to the right and place objects found in a scene to appear as white close to clipping ( zone 10) as not to clip that signal is near saturation and we have captured a lot of information.
Now here lies the problem anything that falls in zone 9 only captures half the signal as zone 10. Here this still is not a big deal as our signal is still rather large and shot noise has very little influence.

Now we go to zone 8 it has only gather a signal that is 1/4 of the signal that zone 10

zone 7 gathers only 1/8 the signal

zone 6 gathers only 1/16 the signal that was captured in zone 10
Now we hit zone 5 middle grey has only captured 1/32 or 3% of the signal that was captured at zone 10 , now we start to see the tell tail signs of shot noise. 94% of the capacity of what the sensor can store is used up in the first 4 stops in tonal range of a scene. Naturally this is where we like to place most of the detail of our image is in zone 5 that only uses less than 3% of the saturation capacity of the sensor.

Dip down to zone 4 where we will place a lot of our darker tones and that is only made up of 1/64 of the light that the sensor can store( less than 1.5%) now shot noise is very easy to find.

---

Last edited by Ian Stuart Forsyth; 05-11-2018 at 10:38 PM. Reason: correction in bold

[stevebrot]

Dip down to zone 4 where we will place a lot of our darker tones and that is only made up of 1/64 of the light that the sensor can store( less than 1.5%) now shot noise is very easy to find.

That is why it is important to expose for the shadows, though I am still not convinced that shot noise is as much a factor in blotchy capture at the low end as random misplacement of values by the A/D converter into the few bits available to express zones 1-3. The sensors are simply not that sensitive and probably are programmed to clip at voltages that low.

Another question, where is shot noise in film captures at similar ISO and exposure times? My zone one and two areas on my negatives are pretty smooth.

BTW...I have done linear TIFF on RAW conversion and the vast majority of the sensor sites are reporting low values. One of these days, I will learn how to use MatLab and properly analyze what is actually captured.


Steve

[MJKoski]

K-1MKII has much better color and far less noise

Running a precisely calibrated monitor here and I do not see better colors. Boosted saturation/vibrance if anything. Traditional Pixel-Shift / Foveon provides you accurate colors, not this one-shot magic accelerator. Just get Luminar if you want accelerator-colors for every camera. And if you expose ISO640 on R2/R3 and do some post pushing on that file you have more adjustment capability than with this MK2 set @ ISO640. That is the gain with newer Sony sensors.

[mohb]

The picture I posted just above your post was taken at ISO=1600, and the raccoon was just barely discernible. To get a "good" picture I would have needed ISO=6400 at minimum. Fortunately, gimp's "auto WB" button was able to salvage something, but the picture would have been much better, and lower stress for me {I am uncomfortable depending on PP to "fix" things}, with a camera that could give me low-noise pictures at ISO 6400 or higher.

t I grabbed a shot anyway - almost totally dark

What puzzles me is why photos like this don't actually look as though they were taken in the dark.

[Kunzite]

Running a precisely calibrated monitor here and I do not see better colors. Boosted saturation/vibrance if anything. Traditional Pixel-Shift / Foveon provides you accurate colors, not this one-shot magic accelerator.

Isn't Foveon known for color accuracy issues?

[MJKoski]

Explain how please. Quattro models are not like their older siblings and may have slightly less accurate RGB-result but still very good.

[Kunzite]

I was asking a question. AFAIK there were issues with particularly the reds, due to how color is detected.
If it's no longer the case, and Foveon is superior in this regard - are there tests to prove it?

[MJKoski]

You can use 100% crop as final image with Foveon and properly executed PS as long as you focus it correctly. It still looks good when printed or seen on monitor. Not so with Bayer-files, small details having certain coloration becomes random mess.

[BigMackCam]

You can use 100% crop as final image with Foveon and properly executed PS as long as you focus it correctly. It still looks good when printed or seen on monitor. Not so with Bayer-files, small details having certain coloration becomes random mess.

The detail from the Foven sensor is quite amazing. I'm quite surprised that you don't find the artefacts annoying, though, given that you have very high standards for IQ (and I don't say that as a criticism - merely an observation). I only have DPR's test shots to review, but at 100% reproduction I see many strange little artefacts - cross-hatching and lines - in some of the face shots... They wouldn't bother me personally, but they're clearly visible. Also, I have to agree that the colours are very odd indeed - certainly when compared to all the other major manufacturer interpretations. That said, it's all adjustable in post-production if you've the will to do so (in Sigma's case, though, with their own app or nothing else - sadly)...

[Rico]

Running a precisely calibrated monitor here and I do not see better colors. Boosted saturation/vibrance if anything. Traditional Pixel-Shift / Foveon provides you accurate colors, not this one-shot magic accelerator. Just get Luminar if you want accelerator-colors for every camera. And if you expose ISO640 on R2/R3 and do some post pushing on that file you have more adjustment capability than with this MK2 set @ ISO640. That is the gain with newer Sony sensors.

Running a precisely calibrated monitor here and it is abundantly clear to me how much better the K-1MKII files are than the K-1, Sony and Nikon in the Studio Scene comparison.

[reh321]

What puzzles me is why photos like this don't actually look as though they were taken in the dark.

There is almost always some light out there - that is why "night goggles" work, cats can "see in the dark", cameras find something to record, and in this case, how I knew the raccoon was there. However, even at ISO=1600, the light recorded took up only the lower quarter of the available values {and a majority in the lower eighth}, all of which appear too dark to our eyes.




So in this case, when I pressed the "auto WB" button, gimp in essence multiplied every value by four, having the effect of stretching out the histogram, and giving an image easier for us to absorb.



If I had set the ISO number higher, the camera would have done this via amplifying the values.

[Oakland Rob]

They also could have negotiated K-cup support with Keurig. Just saying...


Steve

Geez dude...just came dangerously close to a spit-take on reading that, with said K-cup beverage just having been imported into my mouth.

But I disagree...it seems to work better in post-processing as a plugin while I wait for Lr, etc to do its thing.