View Full Version : A question for Brian
This is a question for Brian Sweeney. Brian has talked about the desirability of a black-and-white M9 and the advantages it has over converting the current M9 color images to b&w. I think he is absolutely on target. There have been sites on the web that offered black-and-white conversions for Canon and Nikon. Essentially replacing the anti-aliasing filter and.... At that point, I do not know what they are doing. Presumably, they are changing the firmware/circuitry in the camera to allow the 4 pixels that represent a color point to separate pixels that represent luminosity, thus raising the pixel count and ability to capture fine detail or produce large prints. But, once again, I donít really know. Brian probably does. And, more important, he knows the possibility of actually having a black-and-white M9 produced and its advantages and disadvantages. So, Brian, do you think the folks who are converting other cameras could do it to M9ís. Do you know what they are actually doing and does it take full advantage of converting to b&w?
The Party Line at Leica is that it is too difficult and expensive. I don't know enough to argue either way, but I have specifically asked and been told this.
To convert the sensor to monochrome, the Mosaic filter with the red-green-blue filter elements needs to come off. To be Monochrome-visible, the IR absorbing filter needs to stay in place. I have read of some people using acetone to take the filter and microlens array off. You want to use a microlens array, just do not put any dye into it.
Kodak would be willing to make a run of Monochrome KAF-18500 sensors if Leica wanted them. I talked to them about it, as I did many years ago in having an Infrared version of the Kodak DCS200 made.
As to what is required for firmware on Monochrome- it's mostly bypassing routines in the firmware. You do not need to interpolate, just present the pixels as they are. No need for color balance. Raw is easy- just work with the values stored. Scale them to 8-bit values for display on a computer, or 16-bits if the adapter can handle it. As you do not need to interpolate, you get a slightly bigger image as you can use the pixels at the edge of the sensor.
I suspect Leica's biggest problem is maintaining two versions of firmware, one for monochrome and one for color. Having one version of firmware for both would probably produce code that is too big for the embedded processor. "Conditional Compile" is a programmers technique for coping with this problem. Source code is maintained as one version, "Compile time Switches" are used to produce the executable for the target device. One of the tricks that I used to do in supporting multiple versions of hardware.
Phase I makes a digital monochrome camera, reviewed here:
I talked to Kodak about making Infrared version of the Kodak DCS200m in 1993. They called me back, and let me know the Army wanted the same thing. They did a run of 50 KAF-1600 arrays without the integrated IR cut filter. The firmware in those days was simple: Raw only, read into Photoshop. If Leica were to do a Monochrome camera, the easiest way to quickturn the firmware would be to offer it in DNG only, either 8-bit or 16-bit mode. The Display routines are there, and the "white balance" routines could be bypassed.
It was all so much easier when we wrote our own firmware for our Custom digital imagers.
Should we start holding up 'will program for M9' placards somewhere? I'm more of a C++ sort myself - dealing with game consoles, notheing as low level as firmware - but hey, I'm sure I could figure out something out if an M9 were dangled as a carrot. ;)
I would love to take a shot at it. But the easy way to do it is just store the image as a DNG file and do any required processing in Lightroom. Or FORTRAN... I did my DCS200ir Raw convertor in FORTRAN. Just have the camera display the image in Monochrome, and it does that already.
The CCD Sensor in the M9 is not currently offered with a Monochrome/ Microlens option.
As an example of the type of sensor that would be good for a monochrome camera, this one is offered as Monochrome with Microlens.:
It uses 9micron pixels, and is 36.8mm by 36.8mm.
The offset microlens array in the KAF 18500 could be fabricated using the same process, ie "Not add the color dye". The chip would pick up a full stop as the color dye layer absorbs 1/2 of the light falling through it. No color aliasing, and the array would offer a full 72 Line-Pair per millimeter resolution. That puts it on par with TMAX 400 and Tri-X pan. TMAX 100 offers double the resolution, out to ~150LP/mm.
vBulletin® v3.6.8, Copyright ©2000-2013, Jelsoft Enterprises Ltd.