[biz-engineer]

Doubtful about DXO figures, I used the comparometer of image resources to compare a number of cameras at various ISO and my conclusion is: the truth is in the red.
I compare the red zones of the sames image for different formats of different brands, considering that the Achilles heel of digital sensor is the red.

Based on the red:
- K-3 render slightly better than D7200 at ISO from 100 to 400
- D7200 render slightly better than K-3 above ISO400
- 7DII render reds poorly, not as good as K-3 and D7200, but similarly to the APS sized sensors
- XT-1 renders red poorly since I'd expect the x-trans to be better (maybe I'm wrong, I haven't take the time to really look at that sensor)
- DP1 merrill renders red poorly vs what I'd expect
- D750 is does poor red rendering versus other FF of similar pixel count, including Canon 5DIII
- Canon 5DIII renders red surprisingly well give its poor DR rating from DXO, almost as good as D810
- D810 renders red significantly better than D750
- A7r renders red about the same as D810
- A7rII renders red better than any other FF camera
- 645D renders red as well as A7rII
- 645z beat them all at the red comparison

The truth is in the red.

[RobA_Oz]

Interesting. Thanks for doing the research. Perhaps you might think of turning it into an article, here, complete with comparative images, tables etc.

I have a friend who's tied to Nikon gear, because his wife invested in their lenses (fair call, I reckon) and he's always complaining that his D7100 won't reproduce reds well (red roses are a bit of a specialty with him). Perhaps I should advise him to sell all their Nikon gear and put the proceeds into a 645Z with a single decent lens.

[Na Horuk]

Very cool. Lots of people complain about reds in one or another camera. But we need to differentiate between red, as people see it, and red, as the camera captures it. Problem is that red is highly affected by camera profile, saturation, WB, probably even colour space and of course even by monitor calibration. So I'm not sure how to test true red - shoot raw, use neutral WB, calibrate the camera colour profile, and take a photo of an objectively red object at daylight, have a computer algorithm determine which camera recorded the most, closest tone?
But yeah, good idea to look specifically at red

[LoneWolf]

Very cool. Lots of people complain about reds in one or another camera. But we need to differentiate between red, as people see it, and red, as the camera captures it. Problem is that red is highly affected by camera profile, saturation, WB, probably even colour space and of course even by monitor calibration. So I'm not sure how to test true red - shoot raw, use neutral WB, calibrate the camera colour profile, and take a photo of an objectively red object at daylight, have a computer algorithm determine which camera recorded the most, closest tone?
But yeah, good idea to look specifically at red

A lot depends on how accurate you need the the Red. For Product Pictures (Cars, Clothes Etc) I would think you woulld want it spot on. For other some subjects acuracy is probably less important, so long as the overall result is effective.

[normhead]

The biggest factor in rendering reds is the effective size of the pixel. Diffraction is much greater in the red end of the spectrum than it is in the blue, so as the pixel size gets smaller, the rendering of detail in the reds is more and more affected by diffraction. Most researchers use yellow/ green as the sort of neutral grey for digital cameras, so most of the diffraction charts are done for those frequencies. The good news is on APS-c the blue spectrum won't be diffraction limited until over 100 MP. The bad news is, the red spectrum was already diffraction limited at 16 MP. This is one area where larger pixels on a larger sensor really makes a difference. Diffraction should really be measured against pixel size. So, the smaller the pixel, the more effect it will have.

[rawr]

The original post would be improved by a table showing some identical red area crops of the images discussed presented alongside each other ...

[normhead]

Here's a couple for you....
645z vs A7RII the 645 pretty much wipes the floor with the Sony


D750 vs K-3, not much to look at if you just looked at the 645z, this is where the IQ thing gets ridiculous. People point out minor differences in D750 and K-3 images, but compared to the 645z or even the Sony image... they are pretty much the same. Although you will notice, in the lighter red, that larger pixels in the D750 handle the diffraction much better.


So, I did, my share, if anyone else wants to post a couple, the OP shouldn't have to do all the work.

And now for a question I'd really like answered.

D810 or used 645D


It's close but It's still the 645D for my money.

---

Last edited by normhead; 10-28-2015 at 11:00 AM.

[biz-engineer]

Interesting is the 645D, which is currently at the same price as a new FF camera, but still has better red rendering (based on image resource), although DXO would now rate the 645D lower that most of the most recent full frame cameras.

[normhead]

Just while I'm on a roll here and before I pack it in... looking at the worst 100 ISO red, D7200 and the best 645z, I took the full size 100 ISO images from both, reduced them to 2000 pixels wide and then cropped them to 1080 approx. Can you tell the difference at 2000 pixels wide, between the 645z and the D7200, the best and the worst?





You could argue that if you are printing at 200 DPI, for 8x10s, it doesn't matter whether you shoot a D7200 or a 645z,

Or say 3200 across, which would easily give you a clear crisp 16x12... and probably a 20x16.





I've never determined at exactly what size you need to have the higher resolution camera.

[biz-engineer]

Can you tell the difference at 2000 pixels wide, between the 645z and the D7200, the best and the worst?

Yes, I can tell. The first image is the best. But the difference is small. For printing at 8x10 inches, 2Mpixels is enough. I remember having done 8x10 prints from my old Nikon Coolpix 2100 and the quality was very good.

[normhead]

Yes, I can tell. The first image is the best. But the difference is small. For printing at 8x10 inches, 2Mpixels is enough. I remember having done 8x10 prints from my old Nikon Coolpix 2100 and the quality was very good.

The first rule is find out what the differences are, then decide what it's worth to you.... there's no short cuts, anyway, thanks for starting this thread.

[biz-engineer]

Does anyone know if thanks to the 86K RGB exposure sensor of the K-3 it would be possible to expose specifically for the red channel ?

[RGlasel]

I've never determined at exactly what size you need to have the higher resolution camera.

I don't think resolution is the differentiator, it's how true the colours are. The top pictures (with the 645Z, right?) are clearly better to my eyes on both of my monitors. I do think Pentax does an excellent job of recreating colours, to the point that I never make adjustments to individual RGB channels, even with my lowly K-30, and my son is constantly adjusting the colour on photos from his 60D. I've had in depth discussions about colour perception with a couple of paint reps over the years and you and I are naturally handicapped by our sex. I would be interested to know how obvious the difference is to women who aren't menstruating and drinking coffee (job requirements for pigment testers). When you consider who does the final appraisal of wedding photographs, I don't think this is a trivial matter for professional photographers.

[rawr]

Does anyone know if thanks to the 86K RGB exposure sensor of the K-3 it would be possible to expose specifically for the red channel ?

Perhaps by adjusting your white balance colour temperature settings before shooting?
It would make no difference to your RAW though.

[wombat2go]

The spectral quantum efficiency of the typical sensor well tends to peak at about 700nm and fall away toward blue and red.
Red has the lowest QE because the longer wavelength lower energy photons go deeper into the depletion region and may not end up being stored as electrons.
Also red photons may end up in adjacent wells.
This is explained by Holst and Lomheim in "cmos/ccd Sensors and Camera Systems"

To check this effect with the Pentax K-01 , I took photos of a gray card and a red T shirt, and directly converted (before the Bayer de matrix and de-gamma) the raw file to a hex file.

This jpg is a copy of the spreadsheet I made to show and quantify the raw pixel values.
https://app.box.com/s/7z4i3w4iey5h7r2eqmblgs28nduimcf5

Note that on the gray card LAB 50,0,0 , exposed normally so the camera histogram indicated about 50% brightness, the red wells are only 5.7% of full capacity.
In decibels, the red wells are typically at minus 24 dB ( or minus 4 EV stops) of well capacity, and less than half of the values in the green wells.
To prove these levels, I took an overexposed photo of the bright blue sky, to show that all wells do reach full capacity (approaching FFFF on the 12 bit K-01)

It is apparent that the histogram displayed on the camera lcd is for the resultant jpg, after application of the Bayer de matrix coefficients, and the gamma etc.
The camera histogram does not give direct information about the utilization of the well dynamic range.

This low red level has a worse (by about 20dB compared to full well) signal to noise ratio,
and worse (by 16 times), quantization noise by the adc because the adc on red channel is effectively only with a resolution of 8 bits (1 part in 256)

All this "truth" might sound more depressing than it is. I don't know if the pundits consider this in the performance tests.
However the Bayer sensor works rather well, I think. I suppose it is due to our forgiving eyes.