R-D1 x M8 highlights & blue channel sensitivity: can somebody explain this?

[crusius]

I have been testing my old R-D1 and my new M8 side by side, and today I ran into something that I can't fully explain: a dramatic difference in blue sensitivity. Attached are two shots from a sunflower in my backyard. They have been over exposed on purpose (that's what I wanted to test). Same lens, same filter (IR/UV), same exposure, ISO 200 in the R-D1 and 160 in the M8 (which is really 200).

Now: the R-D1 file over-exposes to saturated yellow (R+G). But the M8 saturates all three RGB channels. By poking at the yellow petals it seems that the M8 has about thrice the amount of blue as the R-D1: the saturated R-D1 areas are at about 30% blue.

That's just plain weird! The petals are the same! If it were, say, 10% difference, 20%... but three hundred percent?

What's going on here?

(For the record, as I exposed less and less the trend remained the same: R-D1 saturates R+G, M8 saturates all channels.)

Edit: This, by the way, makes a huge difference in this picture. The R-D1 "highlight recovers" to yellow, while the M8 "highlight recovers" to something closer to gray. The M8 completely screws up this one.

 

[crusius]

Adding one more piece of information: this attachment is one M8 picture from the same flower, but now correctly exposed. Check, however, the petal the arrow points to. It has a white band. And it is not because it is overexposed: all channels are below 100%. It is because the M8 thinks it is "white". There is simply no decent way, in LR, to make that region go to yellow, which is the correct color.

That just sucks.

Again, help would be appreciated: what's going on, and how widespread is this?

Edit: forget about this one. From the sample above the R-D1 also thinks it is "white"...

 

[crusius]

Bump!

Nobody out for this one?

 

[Sam N]

Interesting. One thought is that if the uv/ir filter is still on the lens for the R-D1 shot, the R-D1 is getting more UV/IR filtering since it also has a filter on the sensor.

If you overexpose something enough all the channels will blow out... could it be that the R-D1's ISO200 isn't really ISO200 like the M8s 160 is? If you take 2 properly exposed shots with the same exposure settings, do the values come out the same?

 

[crusius]

Interesting. One thought is that if the uv/ir filter is still on the lens for the R-D1 shot, the R-D1 is getting more UV/IR filtering since it also has a filter on the sensor.

If you overexpose something enough all the channels will blow out... could it be that the R-D1's ISO200 isn't really ISO200 like the M8s 160 is? If you take 2 properly exposed shots with the same exposure settings, do the values come out the same?

Yes, the UV/IR filter is on for the R-D1 too.

As far as overexposing enough so that all the channels clip, that's not the problem: those are the two "most overexposed" pictures. After those, I took progressively less overexposed pictures until there was no overexposure at all.

Result? The M8 always overexposed to white, no matter how small the overexposure was. The R-D1 always overexposed to yellow. The difference is really quite dramatic.

You can actually see on those two pictures, if you open them in PS or something like that, that the overexposed areas are basically the same. In other words, they are "equally overexposed." It is just that the M8, on those areas, completely blows out the blue channel for some reason...

 

[Sonnar Brian]

The M8 has a Kodak sensor in it. Kodak spent a lot of time, effort, and money to enhance the blue response of their Silicon-based CCD's. Typical Silicon based CCD's lose a lot of sensitivity in the Blue region. My 1st Generation Digital cameras have poor blue response. The CCD used in the RD-1 is same generation as the Nikon D-100. I'm wondering if it's just the spectral response of the CCD.


typical response of a CCD without improved Blue sensitivity:
http://www.astrosurf.com/audine/pdf/kaf1600.pdf

and the improved spectral response of the KAF1602e with improved blue.

http://www.kodak.com/global/plugins...ISS/datasheet/fullframe/KAF-1602ELongSpec.pdf

 

[crusius]

The M8 has a Kodak sensor in it. Kodak spent a lot of time, effort, and money to enhance the blue response of their Silicon-based CCD's. Typical Silicon based CCD's lose a lot of sensitivity in the Blue region. My 1st Generation Digital cameras have poor blue response. The CCD used in the RD-1 is same generation as the Nikon D-100. I'm wondering if it's just the spectral response of the CCD.


typical response of a CCD without improved Blue sensitivity:
http://www.astrosurf.com/audine/pdf/kaf1600.pdf

and the improved spectral response of the KAF1602e with improved blue.

http://www.kodak.com/global/plugins...ISS/datasheet/fullframe/KAF-1602ELongSpec.pdf

Thanks for the documents, I didn't know about them.

That seems to explain what I'm seeing. Interestingly, from those plots, the orange sensitivity has been boosted dramatically too.

Oh well, know your tools and treat them accordingly, I guess.

 

[Sonnar Brian]

I used to write a lot of FORTRAN code to perform non-uniformity correction for custom imaging arrays and to convert intensity to energy. Spectral response curves went into the calculations. I got hold of the KAF 1600 spec sheet around 1992. When I saw the Near IR response of the datasheet compared with the response of the DCS200 camera, I called Kodak and had them make one for me without the Infrared cut filter. Extra $4,000.

For some reason, I seem to recall that Kodak added Tin to the array to improve the Blue response. But it could have been a different sensor...