this is the first I'm hearing of this, but if the color casts are the same in every shot with the same WB, then certainly, it could be calibrated out with a smoothed stack of black frames (a simple single-frame black frame subtraction would increase random read noise by 41%). I have noticed something like this in ISO 200K D4 shots, but I don't know if repeated in every frame.
I never even noticed that ACR adjustment before since this is the first time I'm trying ISOless on a serious basis. A +2 adjustment took away the green cast I was seeing in the shadows and with acceptable side effects to other aspects of the image. Thanks John.
I haven't been looking for it in my images but it does show up on the blackframe: http://forums.dpreview.com/forums/read.asp?forum=1021&message=41036302 . I don't see nearly the same effect on the D7000's black frames.
Would this involve something like averaging, say, four frames...more? Would you apply gaussian blur to smooth in addition?
OH NO! I just absolutely f*cking hate banding.. it destroys almost every image that needs a "shadow-lift".. all shots taken in strong sunlight that is.. F*ck! I had almost decided on the D800.. But shadow banding is a NO NO for me after over 4 years with the D3-banding.. its horrible.. And the worst thing is, I got the D5000 as a small backup, and it has absolutely ZERO banding, gives me better image-quality for pro-retouching then my D3.. and thats really a scam..
4 is not a lot. 4 has 1/2 as much random noise as 1. Starting at about 16 of them, where random noise is 1/4, will take away fixed noise with little additional random noise. The softening thing could work better with 4, so long as all the patterns that you want to remove are low-frequency. Sometimes, cameras have fixed differences in offsets of odd and even lines, which blurring a blackframe would not help; you'd have to blur lines or color channels independently.
Manufacturers seem to be more concerned with speed than accuracy, and of course, the converters have not stepped up to take care of that, in cases where they could help. Astrophotographers do this stuff all the time, but the software they use does not give the full, glossy product that commercial converters give with full color space conversions.
No. Pushing 1600 would have been better, provided that you didn't need the extra 4 stops of highlights that 100 would afford. The camera is not perfectly "ISO-less"; it has noise independent of amplification which is fairly low, but visible because of its character and because photosite noise is so low at base ISO.
That's a shame because the D800 banding is very simple; it runs consistently along the horizontal. Simply adding 100 to 200 columns of masked pixels on the left or right edge of the sensor would allow this banding to be mostly subtracted. Someone posted a conversion of a D800 ISO 100 blackframe the other day, and even after the conversion, which is not the optimal point at which to remove banding, I eliminated it using 200 columns of pixels, ignoring the 20 biggest outlier pixels in each horizontal strip. I used the left edge of the image for the sample pixels, and it cleared up the banding on the right edge of the image.
On the contrary; it is the people who have a better grasp of what exposure really is on a digital camera that understand the value of having a system in which analog gain is pointless, and exposure is absolute, rather than relative to a default JPEG.
Well, there are still uses for some kind of meter, even after we have cameras with only one gain (or pure photon counters).
I like how they do their test. Testing both highilght and shadow retention in RAW files at base ISO. Seems the D800 has the edge on both ends.
I'm going to have to start a small collection of black frames, if for no other reason than to try to get a handle on the shape of it. This kind of noise is the only thing in my view that holds the D800 back above ISO6400.
This generation of Exmor has more pixels, and I suspect more active circuitry on chip to support extended features such as video frame subsampling. This seems bound to create more heat problems. As you say, things have to move faster -- more pixels multiplied out at video-frame rates. I think the next milestone is 30 full frame captures per second.
Still, I think "exposure be dammed" is one of the benefits ISOless shooting. It means you don't have to fiddle with ISO in changing light conditions, provided the aperture/shutter you've selected is at/below the clipping level for the highest LV you'll see. That means never clipping a highlight, whereas you run that risk w/Auto ISO if the camera meters a given scene too hot.
The intensity is pretty light and could probably be removed with Dfine without too much trouble. The only three cameras/sensors I've personally used that have zero banding are the D5000, A900, and the S90.
