Aptina Clarity+ What's the trade-offs?

Started Apr 27, 2015 | Discussions
dosdan Contributing Member • Posts: 536
Aptina Clarity+ What's the trade-offs?
1

ON Semiconductor/Aptina has a white paper on its Clarity+ CFA. This uses RCB with "C" being Clear. What trade-offs are being made here for the greater sensitivity?

https://www.aptina.com/products/technology/Clarity+WhitePaper.pdf

Dan.

 dosdan's gear list:dosdan's gear list
Pentax K100D Super Pentax K20D Pentax K-5 Pentax K-01 Pentax K-3 +1 more
Mark Scott Abeln
Mark Scott Abeln Forum Pro • Posts: 13,327
Re: Aptina Clarity+ What's the trade-offs?

dosdan wrote:

ON Semiconductor/Aptina has a white paper on its Clarity+ CFA. This uses RCB with "C" being Clear. What trade-offs are being made here for the greater sensitivity?

https://www.aptina.com/products/technology/Clarity+WhitePaper.pdf

This reminds me of a recent thread, where clear pixels were discussed:

http://www.dpreview.com/forums/post/55362243

 Mark Scott Abeln's gear list:Mark Scott Abeln's gear list
Nikon D200 Nikon D7000 Nikon D750 Nikon AF-S DX Nikkor 35mm F1.8G Nikon AF Nikkor 50mm f/1.8D +2 more
OP dosdan Contributing Member • Posts: 536
Re: Aptina Clarity+ What's the trade-offs?
1

I see Eric mentioned in that thread that Samsung uses RWB in on of their sensors:

http://image-sensors-world.blogspot.com.au/2015/03/samsung-announces-8mp-rwb-isocell-sensor.html

I wonder if Samsung licensed it off Aptina and whether that use NR like Aptina:

Aptina’s Clarity+ CFA pattern (RC/CB) is unique in that we have the same number of B and
R pixels as Bayer, but determine G through a subtraction of R and B from C. Because we
have the same number of C pixels as other 50% C schemes, we can extract the same SNR
improvement. However, because we have the same configuration as Bayer, we uniquely
maintain the high fidelity of luma information and low level of chromatic artifacts to which
the industry is accustomed. Because G is determined through a subtraction, there is a
tendency to amplify noise in the conversion to the standard RGB (sRGB) representation. This is addressed by image processing algorithms that are part of the Clarity+ technology.

Clarity+ uses advanced chroma-luma processing to reduce noise in the red and blue channels, and correlates what’s left with the noise in the clear channel. This processing prevents the noise from being amplified by the color correction matrix (CCM), which converts RCB to RGB. After the CCM, the clear channel is used again in a final mixing and filtering step to further reduce noise and enhance detail. Similar advanced image processing can be applied to Bayer sensors, with approximately 1 dB benefit in noise. However, in Clarity+, the high-fidelity clear channel, with its better SNR, generates an additional 3 dB advantage beyond what is achieved in Bayer.

Dan.

 dosdan's gear list:dosdan's gear list
Pentax K100D Super Pentax K20D Pentax K-5 Pentax K-01 Pentax K-3 +1 more
Eric Fossum
Eric Fossum Senior Member • Posts: 1,370
Re: Aptina Clarity+ What's the trade-offs?
3

dosdan wrote:

I see Eric mentioned in that thread that Samsung uses RWB in on of their sensors:

http://image-sensors-world.blogspot.com.au/2015/03/samsung-announces-8mp-rwb-isocell-sensor.html

I wonder if Samsung licensed it off Aptina and whether that use NR like Aptina:

RGBW is not from Aptina nor Samsung. RGBW goes back (at least) to US 2,755,334 by Banning, filed in 1951, and referenced by Bayer as prior art.

In more recent times it was reintroduced by Toshiba as well as by Gang Luo at the 2007 IISW.  You can find these papers at the IISS website library with free access.

[1] H. Honda, Y. Iidaa, G. Itoha et al., “High Sensitivity Color CMOS Image

Sensor with WRGB Color Filter Array and Color Separation Process Using

Edge Detection,” Proc. 2007 Int. Image Sensor Workshop, pp. 263–266,

Jun. 2007.

[2] G. Luo, “Color Filter Array with Sparse Color Sampling Crosses for

Mobile Phone Image Sensors,” Proc. 2007 Int. Image Sensor Workshop, pp.

162–165, Jun. 2007.

 Eric Fossum's gear list:Eric Fossum's gear list
Sony RX100 II Nikon Coolpix P900
OP dosdan Contributing Member • Posts: 536
Re: Aptina Clarity+ What's the trade-offs?

Eric Fossum wrote:

dosdan wrote:

I see Eric mentioned in that thread that Samsung uses RWB in on of their sensors:

http://image-sensors-world.blogspot.com.au/2015/03/samsung-announces-8mp-rwb-isocell-sensor.html

I wonder if Samsung licensed it off Aptina and whether that use NR like Aptina:

RGBW is not from Aptina nor Samsung. RGBW goes back (at least) to US 2,755,334 by Banning, filed in 1951, and referenced by Bayer as prior art.

Eric, this is RWB aka RCB, not RGBW.  Is RCB/RWB patentable?

Dan.

 dosdan's gear list:dosdan's gear list
Pentax K100D Super Pentax K20D Pentax K-5 Pentax K-01 Pentax K-3 +1 more
Eric Fossum
Eric Fossum Senior Member • Posts: 1,370
Re: Aptina Clarity+ What's the trade-offs?
1

dosdan wrote:

Eric Fossum wrote:

dosdan wrote:

I see Eric mentioned in that thread that Samsung uses RWB in on of their sensors:

http://image-sensors-world.blogspot.com.au/2015/03/samsung-announces-8mp-rwb-isocell-sensor.html

I wonder if Samsung licensed it off Aptina and whether that use NR like Aptina:

RGBW is not from Aptina nor Samsung. RGBW goes back (at least) to US 2,755,334 by Banning, filed in 1951, and referenced by Bayer as prior art.

Eric, this is RWB aka RCB, not RGBW. Is RCB/RWB patentable?

Dan.

Oops, right. I doubt the CFA pattern itself is patentable at this point...but probably Aptina has tried (or ON-Semi is trying).  The algorithm to create the RGB data is probably patentable. I am not sure when Aptina started their effort. I know I looked at RWBW at Samsung starting in 2010 or so and published a paper in 2011.  Actually, I looked only at RWBW briefly and did not get great color performance and did not spent much time on it. I doubt I was first. Seems like Aptina put much more effort into it and got better results. Their white paper is dated 2013 but who knows when they started.

BTW, the question is not really if something is patentable.  That is only the first step and is an imperfect process.  The question is if patent is defensible years later.  Also imperfect but that is where the rubber meets the road.

 Eric Fossum's gear list:Eric Fossum's gear list
Sony RX100 II Nikon Coolpix P900
Detail Man
Detail Man Forum Pro • Posts: 17,026
Re: Aptina Clarity+ What's the trade-offs ?

dosdan wrote:

ON Semiconductor/Aptina has a white paper on its Clarity+ CFA. This uses RCB with "C" being Clear. What trade-offs are being made here for the greater sensitivity?

https://www.aptina.com/products/technology/Clarity+WhitePaper.pdf

In the course of finding what appear to be Aptina's US patents issued related to their "dual conversion gain" technology (reported in this post):

http://www.dpreview.com/forums/post/55596836

... I came across some other Aptina patents appearing to relate to clear filter-array technology at:

http://www.faqs.org/patents/assignee/aptina-imaging-corporation/

Went back to the listings. Found 4 separate US patent applications filed in 2013 (in chrono. order):

http://appft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PG01&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.html&r=1&f=G&l=50&s1=%2220130242148%22.PGNR.&OS=DN/20130242148&RS=DN/20130242148

http://appft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PG01&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.html&r=1&f=G&l=50&s1=%2220140125838%22.PGNR.&OS=DN/20140125838&RS=DN/20140125838

http://appft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PG01&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.html&r=1&f=G&l=50&s1=%2220140063300%22.PGNR.&OS=DN/20140063300&RS=DN/20140063300

http://appft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PG01&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.html&r=1&f=G&l=50&s1=%2220140160326%22.PGNR.&OS=DN/20140160326&RS=DN/20140160326

All of these applications make mention of the (at least potential) use of clear filter-array elements.

DM

rubank Senior Member • Posts: 1,089
Re: Aptina Clarity+ What's the trade-offs?

Why not get rid of the CFA alltogether, and use the same type of software algorithm to extract R and B too? Why wouldn´t that work?

Mark Scott Abeln
Mark Scott Abeln Forum Pro • Posts: 13,327
Re: Aptina Clarity+ What's the trade-offs?
1

rubank wrote:

Why not get rid of the CFA alltogether, and use the same type of software algorithm to extract R and B too? Why wouldn´t that work?

Where would you get color information?

 Mark Scott Abeln's gear list:Mark Scott Abeln's gear list
Nikon D200 Nikon D7000 Nikon D750 Nikon AF-S DX Nikkor 35mm F1.8G Nikon AF Nikkor 50mm f/1.8D +2 more
Detail Man
Detail Man Forum Pro • Posts: 17,026
Re: Aptina Clarity- ...
5

Mark Scott Abeln wrote:

rubank wrote:

Why not get rid of the CFA alltogether, and use the same type of software algorithm to extract R and B too? Why wouldn´t that work?

Where would you get color information?

Many a good argument is ruined by some fool who knows what he is talking about.

- Marshall McLuhan

Detail Man
Detail Man Forum Pro • Posts: 17,026
Re: Aptina Clarity+ What's the trade-offs ?

dosdan wrote:

I see Eric mentioned in that thread that Samsung uses RWB in on of their sensors:

http://image-sensors-world.blogspot.com.au/2015/03/samsung-announces-8mp-rwb-isocell-sensor.html

I wonder if Samsung licensed it off Aptina and whether that use NR like Aptina:

Aptina’s Clarity+ CFA pattern (RC/CB) is unique in that we have the same number of B and
R pixels as Bayer, but determine G through a subtraction of R and B from C. Because we
have the same number of C pixels as other 50% C schemes, we can extract the same SNR
improvement. However, because we have the same configuration as Bayer, we uniquely
maintain the high fidelity of luma information and low level of chromatic artifacts to which
the industry is accustomed. Because G is determined through a subtraction, there is a
tendency to amplify noise in the conversion to the standard RGB (sRGB) representation. This is addressed by image processing algorithms that are part of the Clarity+ technology.

Clarity+ uses advanced chroma-luma processing to reduce noise in the red and blue channels, and correlates what’s left with the noise in the clear channel.

Don't know what to make of the "cryptic" meaning(s) of the above sentence - which appears to describe the 1st (post-photosite) "block" in Figure 2, PDF Page 3 of the Aptina white-paper.

My intuition says to look for "NR massaging" within that initial process "block" as perhaps in some way being adverse to the spatial frequency response (which is an ineluctable trade-off regarding all image-capture/processing systems). The graphics allegedly showing spatial frequency resolution (Figure 6, PDF Page 8) are themselves too sparse in pixel-dimensions to be useful for some real "pixel peeping" - and (both) images seem somehow a bit "softened". I wonder what happens with low (PSN) SNR ?

This processing prevents the noise from being amplified by the color correction matrix (CCM), which converts RCB to RGB. After the CCM, the clear channel is used again in a final mixing and filtering step to further reduce noise and enhance detail. Similar advanced image processing can be applied to Bayer sensors, with approximately 1 dB benefit in noise. However, in Clarity+, the high-fidelity clear channel, with its better SNR, generates an additional 3 dB advantage beyond what is achieved in Bayer.

In the "Benefits and Challenges" section (PDF Page 4) these statements may hint at something:

As with other patterns involving C pixels, there is a very slight trade-off (a few tenths of a dB) in visual noise when compared to Bayer in bright light situations, but this is not noticeable in those situations, because SNR is high (typically 35dB or greater) and does not typically limit quality.

OK - but what happens in cases of low (rather than bright) light, then ? Is SNR somehow better ?

There is the benefit in bright light situations, that Clarity+’s increased sensitivity will result in exposure times that are half that for Bayer and this is useful for capturing action shots with less blur and motion artifacts.

So, the above implies that there does not exist a "benefit" in other than bright light situations ?

Source of quotes: https://www.aptina.com/products/technology/Clarity+WhitePaper.pdf

DM

rubank Senior Member • Posts: 1,089
Re: Aptina Clarity+ What's the trade-offs?

Mark Scott Abeln wrote:

rubank wrote:

Why not get rid of the CFA alltogether, and use the same type of software algorithm to extract R and B too? Why wouldn´t that work?

Where would you get color information?

The same way you get green out of the clear pixels, by subtraction of the other colours.
If the Aptina algorithm  is aware of the spatial position of the "green" pixels I guess it could be aware of the R and B pixels as in the Bayer matrix.

But I guess I´m wrong then, it was just a thought.

Mark Scott Abeln
Mark Scott Abeln Forum Pro • Posts: 13,327
Re: Aptina Clarity+ What's the trade-offs ?

Detail Man wrote:

dosdan wrote:

Because G is determined through a subtraction, there is a tendency to amplify noise in the conversion to the standard RGB (sRGB) representation. This is addressed by image processing algorithms that are part of the Clarity+ technology.

Clarity+ uses advanced chroma-luma processing to reduce noise in the red and blue channels, and correlates what’s left with the noise in the clear channel.

I really don't know "what to make" of the "cryptic" meaning of the above sentence (which clearly appears to describe the first post-sensor "block" appearing in Figure 2, PDF Page 3 of Aptina "WP".

Yes that does seem a bit mysterious.

My intuition says to look for NR massaging within that initial process "block" as perhaps in some way being adverse to the spatial frequency response (which is an ineluctable trade-off regarding all image-capture/processing systems). The graphics allegedly showing spatial frequency resolution (Figure 6, PDF Page 8) are themselves too sparse in pixel-dimensions to be useful for some real "pixel peeping" - and (both) images seem somehow a bit "softened".

The softening might be due to their use of soft algorithms; some processing might make them crisper?

Clearly, the CCM will have large coefficients, and so more noise, but their processing separates chroma and luma. How this works out in detail, I can’t really say.

Unfortunately, luma noise usually looks better than chroma noise, and so any benefits that might be gained from having cleaner luma may be squandered on noisier chroma, or smeared chroma, or overall smeared detail, depending on what algorithms they are using.

Another problem is that demosaicing ought to be done late in the processing pipeline to avoid color fringing artifacts, but clearly they are attempting to find correlations between the channels early on, which is like demosaicing, and so such correlations may be poor.

This processing prevents the noise from being amplified by the color correction matrix (CCM), which converts RCB to RGB. After the CCM, the clear channel is used again in a final mixing and filtering step to further reduce noise and enhance detail. Similar advanced image processing can be applied to Bayer sensors, with approximately 1 dB benefit in noise. However, in Clarity+, the high-fidelity clear channel, with its better SNR, generates an additional 3 dB advantage beyond what is achieved in Bayer.

In the "Benefits and Challenges" section (PDF Page 4) these statements may hint at something:

As with other patterns involving C pixels, there is a very slight trade-off (a few tenths of a dB) in visual noise when compared to Bayer in bright light situations, but this is not noticeable in those situations, because SNR is high (typically 35dB or greater) and does not typically limit quality.

OK - but what happens in cases of low (rather than bright) light, then ? Is SNR ... better ?

There is the benefit in bright light situations, that Clarity+’s increased sensitivity will result in exposure times that are half that for Bayer and this is useful for capturing action shots with less blur and motion artifacts.

The above implies that there does not exist a "benefit" in other than bright light situations ?

This should have somewhat better low-light performance, if they allow chroma to float. But they probably strive to still have at least somewhat clean chroma, which will drag down the performance of the entire system.

The only real advantage I see in this sort of arrangement is in producing low-noise, high ISO, mainly-monochrome images.

 Mark Scott Abeln's gear list:Mark Scott Abeln's gear list
Nikon D200 Nikon D7000 Nikon D750 Nikon AF-S DX Nikkor 35mm F1.8G Nikon AF Nikkor 50mm f/1.8D +2 more
Alien from Mars
Alien from Mars Contributing Member • Posts: 569
Re: Aptina Clarity+ What's the trade-offs?

rubank wrote:

Mark Scott Abeln wrote:

rubank wrote:

Why not get rid of the CFA alltogether, and use the same type of software algorithm to extract R and B too? Why wouldn´t that work?

Where would you get color information?

The same way you get green out of the clear pixels, by subtraction of the other colours.

Without the CFA the "other colors" would not exist so you wouldn't have the starting point for the color subtractions.

 Alien from Mars's gear list:Alien from Mars's gear list
Panasonic Lumix DMC-GX1 Olympus OM-D E-M5 Olympus E-M1 II Panasonic Lumix G 20mm F1.7 ASPH Olympus M.Zuiko Digital 45mm F1.8 +4 more
Mark Scott Abeln
Mark Scott Abeln Forum Pro • Posts: 13,327
Re: Aptina Clarity+ What's the trade-offs?
1

rubank wrote:

Mark Scott Abeln wrote:

rubank wrote:

Why not get rid of the CFA alltogether, and use the same type of software algorithm to extract R and B too? Why wouldn´t that work?

Where would you get color information?

The same way you get green out of the clear pixels, by subtraction of the other colours.
If the Aptina algorithm is aware of the spatial position of the "green" pixels I guess it could be aware of the R and B pixels as in the Bayer matrix.

But I guess I´m wrong then, it was just a thought.

The RWBW sensor works because White = Red + Green + Blue, and so if you want to find green, it is equal to White - Red - Blue.

To get full color information, you need all three primary colors, or three distinct arithmetic combinations of these three primary colors.

 Mark Scott Abeln's gear list:Mark Scott Abeln's gear list
Nikon D200 Nikon D7000 Nikon D750 Nikon AF-S DX Nikkor 35mm F1.8G Nikon AF Nikkor 50mm f/1.8D +2 more
Detail Man
Detail Man Forum Pro • Posts: 17,026
Re: Aptina Clarity+ What's the trade-offs ?

Mark Scott Abeln wrote:

Detail Man wrote:

dosdan wrote:

Because G is determined through a subtraction, there is a tendency to amplify noise in the conversion to the standard RGB (sRGB) representation. This is addressed by image processing algorithms that are part of the Clarity+ technology.

Clarity+ uses advanced chroma-luma processing to reduce noise in the red and blue channels, and correlates what’s left with the noise in the clear channel.

I really don't know "what to make" of the "cryptic" meaning of the above sentence (which clearly appears to describe the first post-sensor "block" appearing in Figure 2, PDF Page 3 of Aptina "WP".

An initial thought was they might be using the term "correlate" in a purposely vague way ...

... but this statement (from the 1st in series of 4 related US patent applications filed in 2013):

IMAGING SYSTEMS WITH CLEAR FILTER PIXELS

... array of image sensor pixels arranged in color filter unit cells each having one red image pixel that generates red image signals, one blue image pixel that generate blue image signals, and two clear image sensor pixels that generate white image signals. The image sensor may be coupled to processing circuitry that performs filtering operations on the red, blue, and white image signals to increase noise correlations in the image signals that reduce noise-amplification when applying a color correction matrix to the-image signals.

... appears to imply that (somehow, by some particular means) the combining of separate RGW channels such that their noise components *correlate* improves the net, composite SNR. Sort of seems bass-ackwards to my mind. Such would represent a -3 dB *degradation* of ("correlated") channel noise . They are claiming an (overall system) *boost* of SNR by +3 dB ...

However, I have not slogged through these patents - which tend to be tedious documents to read.

.

Mark wrote:

... if you want to find green, it is equal to White - Red - Blue.

Ah, so one would want the Red and Blue channel noises to correlate in their combining to maximally subtract from "White channel" noise (?). However, the "White channel" noise subtracted from would need to (itself) be correlated with the Read and Blue channel noises.

Since we are talking about (essentially random, uncorrelated) Photon Shot Noise appearing in all three individual channels, I don't think that I understand how one could manage to go about such.

.

Yes that does seem a bit mysterious.

My intuition says to look for NR massaging within that initial process "block" as perhaps in some way being adverse to the spatial frequency response (which is an ineluctable trade-off regarding all image-capture/processing systems). The graphics allegedly showing spatial frequency resolution (Figure 6, PDF Page 8) are themselves too sparse in pixel-dimensions to be useful for some real "pixel peeping" - and (both) images seem somehow a bit "softened".

The softening might be due to their use of soft algorithms; some processing might make them crisper?

Clearly, the CCM will have large coefficients, and so more noise, but their processing separates chroma and luma. How this works out in detail, I can’t really say.

From Claim Number 1:

... the processing circuitry is configured to increase the noise correlations associated with the red Image signals, the blue image signals, and the white image signals to greater than 70 percent of all noise associated with the red Image signals, the blue image signals, and the white image signals to greater than 70 percent of all noise associated with the red Image signals, the blue image signals, and the white image signals. ...

... the processing circuitry is configured to perform the filtering operations by, for each image sensor pixel of a given color, generating a weighted sum of image signals generated by at least 25 image sensor pixels.

They seem to be saying that a minimum of 5x5 photosite array data would be utilized ...

I wonder how that "squares" with *equaling* the RGB-rendered system output spatial frequency "resolution" of the (as small as) 2x2 photosite array size alternatives ? It seems not improbable that various demosaicing algorithms may well themselves utilize higher than 2x2 photosite array-size image-data in the process of rendereding individual pixel-locations.

Patent applications can twist one's head off. One must read and consider their content carefully.

Unfortunately, luma noise usually looks better than chroma noise, and so any benefits that might be gained from having cleaner luma may be squandered on noisier chroma, or smeared chroma, or overall smeared detail, depending on what algorithms they are using.

No mention of a given "color space" used for luminance/chrominance differentiation. Seems like one might well want/need to come up with something non-standard and correspondingly unique.

Another problem is that demosaicing ought to be done late in the processing pipeline to avoid color fringing artifacts, but clearly they are attempting to find correlations between the channels early on, which is like demosaicing, and so such correlations may be poor.

The whole bit (above) indicating that they *purposely* make noise-sources *correlate* (thus, not cancel at all) seems different enough from what I imagine that demosaicing algorithms' may be trying to accomplish where it comes to RAW RGB channel noise-sources.

This processing prevents the noise from being amplified by the color correction matrix (CCM), which converts RCB to RGB. After the CCM, the clear channel is used again in a final mixing and filtering step to further reduce noise and enhance detail. Similar advanced image processing can be applied to Bayer sensors, with approximately 1 dB benefit in noise. However, in Clarity+, the high-fidelity clear channel, with its better SNR, generates an additional 3 dB advantage beyond what is achieved in Bayer.

In the "Benefits and Challenges" section (PDF Page 4) these statements may hint at something:

As with other patterns involving C pixels, there is a very slight trade-off (a few tenths of a dB) in visual noise when compared to Bayer in bright light situations, but this is not noticeable in those situations, because SNR is high (typically 35dB or greater) and does not typically limit quality.

OK - but what happens in cases of low (rather than bright) light, then ? Is SNR ... better ?

There is the benefit in bright light situations, that Clarity+’s increased sensitivity will result in exposure times that are half that for Bayer and this is useful for capturing action shots with less blur and motion artifacts.

The above implies that there does not exist a "benefit" in other than bright light situations ?

This should have somewhat better low-light performance, if they allow chroma to float.

Not sure specifically why or how. What do you mean by the statement, "... chroma to float" ?

But they probably strive to still have at least somewhat clean chroma, which will drag down the performance of the entire system.

I wonder if "chroma sub-sampling" may be employed - which would seem to not visibly appear when displaying B&W test-targets, and as well not visibly appear in the color-patch test-target.

The only real advantage I see in this sort of arrangement is in producing low-noise, high ISO, mainly-monochrome images.

Thus, favoring "non-challenging" illumination spectrums (lacking Red and Blue fine details) ?

DM

rubank Senior Member • Posts: 1,089
Re: Aptina Clarity+ What's the trade-offs?

Alien from Mars wrote:

rubank wrote:

Mark Scott Abeln wrote:

rubank wrote:

Why not get rid of the CFA alltogether, and use the same type of software algorithm to extract R and B too? Why wouldn´t that work?

Where would you get color information?

The same way you get green out of the clear pixels, by subtraction of the other colours.

Without the CFA the "other colors" would not exist so you wouldn't have the starting point for the color subtractions.

So W-R-B=G, but W-R-G isn´t B?

I´m too stupid to understand that.

rubank Senior Member • Posts: 1,089
Re: Aptina Clarity+ What's the trade-offs?

Mark Scott Abeln wrote:

rubank wrote:

Mark Scott Abeln wrote:

rubank wrote:

Why not get rid of the CFA alltogether, and use the same type of software algorithm to extract R and B too? Why wouldn´t that work?

Where would you get color information?

The same way you get green out of the clear pixels, by subtraction of the other colours.
If the Aptina algorithm is aware of the spatial position of the "green" pixels I guess it could be aware of the R and B pixels as in the Bayer matrix.

But I guess I´m wrong then, it was just a thought.

The RWBW sensor works because White = Red + Green + Blue, and so if you want to find green, it is equal to White - Red - Blue.

Yes, but why isn´t W-R-G=B or W-G-B=R?

To get full color information, you need all three primary colors, or three distinct arithmetic combinations of these three primary colors.

As above, as W-R-B=G what is the difference?

(Do I really have to sober up to get this?)

Mark Scott Abeln
Mark Scott Abeln Forum Pro • Posts: 13,327
Re: Aptina Clarity+ What's the trade-offs?

rubank wrote:

So W-R-B=G, but W-R-G isn´t B?

I´m too stupid to understand that.

Sure, you can design a filter array with white, red, and green pixels.

 Mark Scott Abeln's gear list:Mark Scott Abeln's gear list
Nikon D200 Nikon D7000 Nikon D750 Nikon AF-S DX Nikkor 35mm F1.8G Nikon AF Nikkor 50mm f/1.8D +2 more
rubank Senior Member • Posts: 1,089
Re: Aptina Clarity+ What's the trade-offs?

Mark Scott Abeln wrote:

rubank wrote:

So W-R-B=G, but W-R-G isn´t B?

I´m too stupid to understand that.

Sure, you can design a filter array with white, red, and green pixels.

Very funny.

Keyboard shortcuts:
FForum MMy threads