Home

Please support mflenses.com if you need any graphic related work order it from us, click on above banner to order!

SearchSearch MemberlistMemberlist RegisterRegister ProfileProfile Log in to check your private messagesLog in to check your private messages Log inLog in

Monochrome sensor; very interesting outlook
View previous topic :: View next topic  


PostPosted: Tue Jan 26, 2010 7:10 am    Post subject: Monochrome sensor; very interesting outlook Reply with quote

No cameras for sale there yet, but the direction certainly looks promising to me:

http://www.maxmax.com/b&w_conversion.htm

I think I'm not alone in wishing for a B&W optimized, and affordable, digital sensor.


PostPosted: Tue Jan 26, 2010 7:15 am    Post subject: Re: Monochrome sensor; very interesting outlook Reply with quote

eeyore_nl wrote:
I think I'm not alone in wishing for a B&W optimized, and affordable, digital sensor.


Yes, I agree! I expect we are not the only two persons on the planet either Laughing


PostPosted: Tue Jan 26, 2010 8:01 am    Post subject: Reply with quote

Thanks for the link! Very interesting to read - a bw sensor -equipped DSLR would indeed be on my wantlist too :)


PostPosted: Tue Jan 26, 2010 8:03 am    Post subject: Reply with quote

Kodak was the forerunner of everything digital, shame they couldn't survive in the competition.

http://www.luminous-landscape.com/reviews/cameras/kodak-760m.shtml


PostPosted: Tue Jan 26, 2010 8:17 am    Post subject: Reply with quote

kansalliskala wrote:
Kodak was the forerunner of everything digital, shame they couldn't survive in the competition.

http://www.luminous-landscape.com/reviews/cameras/kodak-760m.shtml


Yes, I know, and there is a monochrome digital back as well, but that's not exactly affordable.


PostPosted: Tue Jan 26, 2010 11:32 am    Post subject: Re: Monochrome sensor; very interesting outlook Reply with quote

eeyore_nl wrote:
No cameras for sale there yet, but the direction certainly looks promising to me:

http://www.maxmax.com/b&w_conversion.htm

I think I'm not alone in wishing for a B&W optimized, and affordable, digital sensor.


Too bad they remove the microlenses - this is a big loss of light gathering ability with CMOS sensors. Another issue is the introduction of aliasing into images.

Their comparison was also not quite proper: if a sensor has AA filter, the captured image is supposed to be sharpened in software . Comparing unsharpened images from AAless and AAfull (I love creating new words Smile ) sensors is not right.

I must also mention, that I just quickly browsed the link, so anything I wrote above may be just nonsense.


PostPosted: Tue Jan 26, 2010 12:34 pm    Post subject: Reply with quote

Although this sounds good and I prefer B&W to colour, I do like the flexibility of a colour DSLR. I prefer to use RAW converter to get the filter effects that like.


PostPosted: Tue Jan 26, 2010 1:59 pm    Post subject: Reply with quote

i think itd be great to have a b&w camera if indeed the result is sharper images/better IQ than achievable through converting color pix. i really dislike wanting to shoot b&w, but seeing the scene through the vf in color; and then first seeing the pic in pp in color. believe it or not, many times i shoot b&w in jpeg just so i dont have to look at the scene in color! one of the reasons i have been wanting an ep1 is because with the liveview you see what you want to shoot!


PostPosted: Tue Jan 26, 2010 2:28 pm    Post subject: Re: Monochrome sensor; very interesting outlook Reply with quote

Anu wrote:
Their comparison was also not quite proper


Also, they have the "red", "green", and "blue" channels at different levels, probably as the result of processing the image in the usual manner (either in-camera or through a typical raw workflow)… The processing of the sensor data should be done in all stages as a monochrome image with a camera like this, not split into channels with any knowledge of the sensor's old CFA.

Still, an interesting idea, perhaps the knowledge of these modifications will progress once more and more “hackers” begin to have several old DSLRs sitting unused. =)


PostPosted: Tue Jan 26, 2010 2:35 pm    Post subject: Reply with quote

Well, I would not buy a b&w DSLR, but what I really find interesting is a compact or RF type of cam with such a b&w sensor.
Something like a b&w Lumix GF-1 and the 1.7/40, but with a rangefinder and immediate shutter button reaction. And something like Canon Picture Styles.

A perfect cam for street shots.


PostPosted: Mon Feb 15, 2010 12:39 pm    Post subject: Reply with quote

I like the idea with the conversion of a normal DSLR into a monochrome DSLR. For some foregoing experiments I have some simple digital cameras bougth last year.
Now I learn, that others had the same idea and have first prototypes :-/
But it is some relief for me, that this modification seems complicated.

At work I have the pleasure to work with both color and monochrome industrial cameras up to 24x36 imager size (Kodak), but I have to admit I did´t make comparisons up to now.

I would like to have a special monochrome DSLR, I also have my special camera with excangable filtersystem Smile


PostPosted: Mon Feb 15, 2010 1:41 pm    Post subject: Reply with quote

LucisPictor wrote:

Something like a b&w Lumix GF-1 and the 1.7/40, but with a rangefinder and immediate shutter button reaction.


Yes, I would buy that, if the price was within my budget. No need for picture styles either, just RAW output.


PostPosted: Mon Feb 15, 2010 3:53 pm    Post subject: Reply with quote

I have already my B&W sensor, it is call film, I can develop it in ways that will never be possible with a computer, and I can print it with my 250 Eur enlarger with results that no computer print has matched yet... Wink


PostPosted: Mon Feb 15, 2010 4:55 pm    Post subject: Reply with quote

Well, if anyone is interested in such a solution, I know an industrial company which does sensor conversions like this, also for "normal camera" sensors here in Germany. Not cheap but affordable for the ambitioned amateur.


PostPosted: Mon Feb 15, 2010 5:38 pm    Post subject: Reply with quote

I don´t get it Confused

I find it far better just shooting color. This is the only way to achieve full control making the B&W image you like.

But hey, I might be missing the point.


PostPosted: Mon Feb 15, 2010 5:59 pm    Post subject: Reply with quote

Orio wrote:
I have already my B&W sensor, it is call film, I can develop it in ways that will never be possible with a computer, and I can print it with my 250 Eur enlarger with results that no computer print has matched yet... Wink


+100


PostPosted: Mon Feb 15, 2010 6:00 pm    Post subject: Reply with quote

lahnet wrote:
I don´t get it Confused

I find it far better just shooting color. This is the only way to achieve full control making the B&W image you like.

But hey, I might be missing the point.

+1


PostPosted: Tue Feb 16, 2010 4:44 pm    Post subject: Reply with quote

lahnet wrote:
..
I find it far better just shooting color. This is the only way to achieve full control making the B&W image you like.
..


If you want to use virtual color filters in photoshop, then a color sensor will be best for you.
But with a pure monochrome sensor you get a little more light sensitivity. And you get the posibillity to use IR (ok, with a modified camera without IR-Cut you could get that too).
http://www.kodak.com/global/plugins/acrobat/en/business/ISS/datasheet/interline/KAI-16000LongSpec.pdf

With a monochrome camera you get rid of the step of bayer filtering the image on the sensor and the debayering in the camera hardware or RAW software. These steps could cause problems, lets think about small details with its colors changing. The resolution of the image is better (depends of the debayering algorithm).

Without the bayer pattern a sensor has a behavior much more like monochrome film.


PostPosted: Tue Feb 16, 2010 4:56 pm    Post subject: Reply with quote

yeah, i could definitely see that would be true; the directest is often the bestest. Smile i would be very intd in this type of camera....one of the nice things about my newly acquired ep1 is that with live view you can actually see the scene in b&w (as opposed to trying to visualize a color scene in b&w), so you can make adjustments before snapping.


PostPosted: Tue Feb 16, 2010 7:18 pm    Post subject: Reply with quote

kds315* wrote:
Well, if anyone is interested in such a solution, I know an industrial company which does sensor conversions like this, also for "normal camera" sensors here in Germany. Not cheap but affordable for the ambitioned amateur.


let us know some more about it. who is it? and what are they doing?


PostPosted: Sun Mar 14, 2010 12:43 pm    Post subject: Reply with quote

lahnet wrote:
I don´t get it Confused I find it far better just shooting color. This is the only way to achieve full control making the B&W image you like.


The point is resolution, and thus speed. With a color sensor, each usable pixel is essentially a statistical averaging of the 1R+1B+2G elements. So a 20mpx color sensor really only delivers 5-10mpx points, depending on light color. Increasing ISO requires amplifying the output of each pix.element, and those (relatively) fuzzy points bleed over to each other, introducing noise.

Without the Bayer filter, without the fuzziness, you would actually get a REAL 20mpx, about 2-4x better than in color. And adjacent pix.elements would be have more similar output levels than the fuzzy 4-by's, so bleed-over noise would diminish. What had been an ISO 100, supposedly 20mpx sensor now becomes an ISO 35 (or so) effectively 50-60mpx sensor. And that ISO can be boosted to higher levels with less noise. Ooh, grainless digital Panatomic-X!! Pushable to ASA-9600!!

One advantage of color dSLRs is (probably) filtration -- we can make the B&W image look as though it was shot through red, green, blue, yellow, orange etc filters. I don't know if a monochrome sensor can be jiggered to simulate filtration, or if you'd have to actually put a filter in front of the lens. But pull the hot-filter, and good IR shooting should be possible.


PostPosted: Sun Mar 14, 2010 2:16 pm    Post subject: Reply with quote

good points RR; didnt realize that math. sounds wierdly similar to the foveon story, no?


PostPosted: Sun Mar 14, 2010 7:59 pm    Post subject: Reply with quote

Yeah Reb, some similarity to Foveon, but with simpler construction. And a monochrome sensor wouldn't be subject to the CA reported on Foveon.

I may have been slightly wrong about ISO -- let's think about a (non-Foveon) sensor, two versions, S(c) is standard color with Bayer screen, S(m) is the monochrome variant without that screen. With the chip and support electronics otherwise the same, S(m) would have greater sensitivity than S(c) due to not throwing away 1/2 to 3/4 of the light received. So, all else being equal, if the base level of S(c) is ISO 100, then S(m) is more like ISO 250-300 -- and still has 50-60mpx resolution!

I'm not familiar enough with post-sensor signal processing, how naked sensor output is translated to an ISO standard, to know if it's feasible to drop the processed output (what goes into a RAW file) to a lower level. Would a reduced signal be subject to low-voltage noise? I recall one digicam maker claiming a base ISO of 80, but I don't know if that's true or significant.

I bring up this low-ISO stuff because 1) with film, I often used ASA 8-32 emulsions (grainless!!) and 2) I dislike the hassle of shooting wide ranges of exposures. Sure, I could get welder's glass cut for filters, ND 1000 or thereabouts, so I could shoot 10-20 minute exposures at f/2. Why? To isolate a stationary subject, render all extraneous movement invisible. Why can't camera makers let me go low and slow? (gripe grumble whine)

Oh well. On the other end, as I said before, the monochrome S(m) sensor's output could be amplified more before noticeable noise appears. Again, with all else being equal, we could establish a S(m) base of ISO 200. At ISO 600, S(m) noise would be roughly equivalent to S(c) at ISO 100, its base. The noise from S(c) at ISO 3200 wouldn't appear from S(m) until maybe ISO 20,000!! And S(m) still has 50-60mpx resolution, so for same-size displays (print or screen) that S(m) noise would look like S(c) noise around ISO 1200. Yeah, an 8x-10x performance gain!!

[/me breathes hard, sweats slightly, feels faint with excitement]


PostPosted: Mon Mar 15, 2010 8:21 am    Post subject: Reply with quote

RioRico wrote:
The point is resolution, and thus speed. With a color sensor, each usable pixel is essentially a statistical averaging of the 1R+1B+2G elements. So a 20mpx color sensor really only delivers 5-10mpx points, depending on light color. Increasing ISO requires amplifying the output of each pix.element, and those (relatively) fuzzy points bleed over to each other, introducing noise.


The actualy resolution of a bayer-sensor is highly dependant on the algorithm used to create the image out of the raw-file. Only the absolutely worst possible converters do something as simple as you describe - in real world products all the conversion software I am aware of do quite a bit more clever job. That 5-10mpx claim is just incorrect. Even if we simply throw away blue and red for luminance information, we'll get 10 million greyscale pixels.

I don't reallly understand your fuzzy-point ISO-noise statement.
(I have an old and worn out brain with rather limited usability nowdays Smile ).

Quote:

Without the Bayer filter, without the fuzziness, you would actually get a REAL 20mpx, about 2-4x better than in color.

These numbers are way off.

Quote:

levels with less noise. Ooh, grainless digital Panatomic-X!! Pushable to ASA-9600!!

The big problem with pushing to high ISOs is banding noise, fixed pattern noise. Without such problem one could push even a relatively small bayer-sensor to ISO 100.000 and have rather nice greyscale image.

Trying to get noise free images (without processing) at any ISO is just a dream for any technology as light itself produces noisy image.

Quote:

the lens. But pull the hot-filter, and good IR shooting should be possible.


Canon used to sell a modified 20D without the IR filter. Mayby someday the sensor technology will be such that IR (and UV) wavelenghts are just separated by the sensor, instead of having to block them before the sensor.


PostPosted: Mon Mar 15, 2010 8:51 am    Post subject: Reply with quote

RioRico wrote:

I may have been slightly wrong about ISO -- let's think about a (non-Foveon) sensor, two versions, S(c) is standard color with Bayer screen, S(m) is the monochrome variant without that screen. With the chip and support electronics otherwise the same, S(m) would have greater sensitivity than S(c) due to not throwing away 1/2 to 3/4 of the light received. So, all else being equal, if the base level of S(c) is ISO 100, then S(m) is more like ISO 250-300 -- and still has 50-60mpx resolution!

Show me a bayer-filter that throws away 3/4 of the incoming light and I'll show pigs that can fly Smile

I am assuming, you also remove AA filter from the b&w sensor, and give the 20M photosite sensor 20 million pixel resolution. This also means that you will accept aliasing.

With bayer filtered sensors and a good debayering algorithm, one gets very close to the same resolution in real life images. The monochrome sensor's images will have a a little more bite to the images, but nowhere near the magniture you represent.

Quote:

S(m) noise would look like S(c) noise around ISO 1200. Yeah, an 8x-10x performance gain!!

A monochrome sensor might give about a stop, at most a bit more advantage over a bayer color sensor. Without bayer filter the sensor receives about twice the photons, maybe a bit more. However, while the perceive shot noise goes down, the read noise does not. Relatively speaking the higher ISOs improve more on the greyscale sensor than the lower ISO images (on typical CMOS-sensor based devices) as the shot noise is amplified more than the read noise.