Lin,
The kernel of importance is that Foveon never invents anything. Bayer does and thus is subject to the accuracy limits of sampling theorem and should therefore be AA filtered. Mead is not an idiot.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Foveon and Bayer: My Editorial
- Thread starter Clint Thayer
- Start date
The kernel of importance is that Foveon never invents anything. Bayer does and thus is subject to the accuracy limits of sampling theorem and should therefore be AA filtered. Mead is not an idiot.
No, I meant general remarks which led to an atmosphere of a near certain expectation.
I don't think anyone took that pic seriously, it wasn't even presented as a good fake, which it obviously wasn't, but rather as a joke.
DaSonyGuy
Forum Pro
Its not bulky at all compared to other pro level DSLR's. (I mean pro level strictly on the grounds of image quality which IMO is what really matters)
???....Your Pentax lenses would not be useless as they will fit straight onto the SA mount with a tiny simple mod to each lens!
You onto have to remove the Auto aperture "tab/lever" and the plastic bezel and voila!
Hmmm, ever heard the expression: "Beauty is in the eye of the beholder"?
You got that right!
Regards
DSG
--
http://sigmasd10.fotopic.net/c173862.html
richard stone
Senior Member
DSG
I am very curious about using the newest Pentax lenses on the Sigma SD10. I don't own one yet, but plan on purchasing one in the near future. The "limited" lens series sounds like it would be a good match for the Sigma. Can you tell me more about the lens modification or where I could learn more?
Richard Stone
I am very curious about using the newest Pentax lenses on the Sigma SD10. I don't own one yet, but plan on purchasing one in the near future. The "limited" lens series sounds like it would be a good match for the Sigma. Can you tell me more about the lens modification or where I could learn more?
Richard Stone
Spiky
Leading Member
but I'm trying to figure out which name you are referencing. I don't believe anyone here actually knows your real name, so that probably isn't it. Your nick is the name of a camera. Uh, have you copyrighted this? Does Sigma Corp know?
Brett DiMichele
Senior Member
--
I am not a Professional but I did stay at Holiday Inn!
Please take a look at my gallery!
http://www.westol.com/~brettd/sd10/gallery/
http://www.pbase.com/sigmasd9/brett_dimichele
Tom Rowland
Veteran Member
Hi Erik,
I have stated in other posts that I am open to calling this "feature" by another name.
Aliasing has a common meaning to most folks that is related to "bad" things that happen when Bayer sensors produce stuff that degrades an image.
I like "apparent detail" better than aliasing, but I also like "false detail" since it implies there is a difference between the true detail of reality, and the false detail of the image. Since formal logic uses the terms true and false with no suggestion of a value judgement I do not associate any value judgement with that term.
On the otherhand I can see how a value judgement is possible in this situation.
There fore I suggest we call what happen when a Foveon sensor produces an image with detail at the Nyquest limit change bokeh.
The term bokeh is a strange sounding word that has no value judgement associated with it. It is different enough that it could easily be recognized, associated with good.
Of course is there is another meaning for this word we could find some equally strange sounding word. lol
I have stated in other posts that I am open to calling this "feature" by another name.
Aliasing has a common meaning to most folks that is related to "bad" things that happen when Bayer sensors produce stuff that degrades an image.
I like "apparent detail" better than aliasing, but I also like "false detail" since it implies there is a difference between the true detail of reality, and the false detail of the image. Since formal logic uses the terms true and false with no suggestion of a value judgement I do not associate any value judgement with that term.
On the otherhand I can see how a value judgement is possible in this situation.
There fore I suggest we call what happen when a Foveon sensor produces an image with detail at the Nyquest limit change bokeh.
The term bokeh is a strange sounding word that has no value judgement associated with it. It is different enough that it could easily be recognized, associated with good.
Of course is there is another meaning for this word we could find some equally strange sounding word. lol
Aaro
Senior Member
I did not try to explain anything but the fact what a round sensors see. They could have been different in geometry ie triangles for example, but the result is the same.
!!!!!
I truly see no problem what the single sensor sees. If the light hits the sensor it outputs 1 if not it outputs 0 or something in between.
What we do with the result or what we call it is another matter.
Aaro
--
Watching you
The man from Snowriver, Mies Lumijoelta
http://www.pbase.com/aaro
Aaro
Senior Member
Not that far:
http://forums.dpreview.com/forums/read.asp?forum=1027&message=12641674
Aaro
Watching you
The man from Snowriver, Mies Lumijoelta
http://www.pbase.com/aaro
http://forums.dpreview.com/forums/read.asp?forum=1027&message=12641674
Aaro
--
Watching you
The man from Snowriver, Mies Lumijoelta
http://www.pbase.com/aaro
No, I saw posts with that photo as conclusive evidence of the "new" SD11discussed on international forums, which were posted by others (non-dpreview Sigma forum readers) in complete seriousness. It's amazing how the world is "tuned in" via the Internet. Mr. Yamaki, Managing Director Mr. Yamaki, (not President Michihiro Yamaki), Sigma, and I talked about that at the PMA dinner. My point to you has often been the wide dissemination of information (or opinions or such fake photos) from this forum to others around the global. I think I read about it on a German forum; it's an international community and we all bear some responsibility for what we opine and making opinions clear from statements based on facts.
I see, you've gone from individuals saying specific expectations (per your previous post) to "general remarks" leading to "near certain expectation" Well, the expectation from the German Sigma website was that Sigma would make announcements on Feb 14, which they did, releasing PRs on the new lenses. I had a certain expectation there would be an announcement, since Sigma so stated on the German language website, but your taking that further to an anticipated camera is a stretch. Or saying that anyone in the know implied so is a stretch, just to be clear. In my opinion, of course.
Best regards, Sandy
[email protected]
http://www.pbase.com/sandyfleischman
Alonso Perez
Well-known member
Thanks for the lens tip, though I'm not keen on irreversible modifications like that.
By atractive I meant the option. The camera is fine as far as looks go. Still, I do consider it bulky; I'm used to an all-manual Pentax MX. I'm not a pro anyway, though I've done paid architectural photography (one of my treasured Pentax lenses is a shift), so I don't need the pro baggage. Even once I move to digital, I don't expect to shoot even 5% of what an average pro does.
By atractive I meant the option. The camera is fine as far as looks go. Still, I do consider it bulky; I'm used to an all-manual Pentax MX. I'm not a pro anyway, though I've done paid architectural photography (one of my treasured Pentax lenses is a shift), so I don't need the pro baggage. Even once I move to digital, I don't expect to shoot even 5% of what an average pro does.
Sigma can't even fix the inexcusable typos languishing for years on their web site, I doubt they have the ability to track all the Spikys in the world.
You still haven't said where you think the "false pixels" in a Foveon image come from. Everyone knows that Bayers guess the vast majority of the information in a Bayer image, and that when their partial samples are taken at a frequency which is insufficient to support the level of accuracy required for a high level of detail, false detail can be generated.
But I'm still wondering where you think Foveon's wrong guesses come from. Any ideas?
--
I am not a Professional but I did stay at Holiday Inn!
Please take a look at my gallery!
http://www.westol.com/~brettd/sd10/gallery/
http://www.pbase.com/sigmasd9/brett_dimichele
Just Looking
Veteran Member
Yes, I had to read a lot into one word, using the picture. As I said, nice try, but to elucidate the controversy about the aliasing problem requires a more subtle picture.
I think your "cheers" picture was just a little misleading, combined with the 9/5 subject, as it showed several rows of sensors well beyond the point where the 9/5 phenomenon happens, into a region where the actual sensor response will be flat. Your bottom set of sensor samples that appears to show 5 blue reactions is NOT where the 5 lines appear. Your readers, who appreciated your illustration, were in fact misled by it.
Sorry, I should make a better picture rather than spend words on it...
j
I think your "cheers" picture was just a little misleading, combined with the 9/5 subject, as it showed several rows of sensors well beyond the point where the 9/5 phenomenon happens, into a region where the actual sensor response will be flat. Your bottom set of sensor samples that appears to show 5 blue reactions is NOT where the 5 lines appear. Your readers, who appreciated your illustration, were in fact misled by it.
Sorry, I should make a better picture rather than spend words on it...
j
You're mixing two different things.
First, if Sigma is infact tight liped about what they aren't doing, then no one outside of their inner circle knows a thing. Certainly not anyone in this group.
Second, yes, I think they've blown it either way. One doesn't need inside information to draw the conclusion that no communication has taken place for way too long. "Too long" meaning that period of time, without a detectable pulse, over which pro outfits have no choice but to declare the DSLR dead as an viable option.
That's a perfect demo of baseless rumor mongering from the few I was talking about.
Well there's no accounting for the gullible nature of Bayer customers, right? The pic was posted here as a funny, admittedly poor fake.
Too bad Sigma hasn't tuned in. And really, statements like that show just how out of touch this company is. They should've finished marveling over the power of the internet in the '90s.
...made by individuals...
See Laurance's post below, for a perfect example of how these unhelpful false expectations come to pass.
Ah c'mon SSD9, short quiz,
what year did the first point and shoot consumer digital camera come out (I just looked it up...by what maker...for a bonus point) what year did the first megapixel consumer camera come out?
what year did the SD9 come out. Bonus point for month. what year did the SD10 come out. Bonus point for month.
5 extra bonus points for correctly predicting month when x530 is selling online for immediate delivery....?
I took my first digital photo summer 2003 and I'm not that technologically backwards. Current photography preoccupation, which fortunately ties in to my job: I blame it on Lake Tahoe, Telecosm conferences. I hadn't taken many photos since our kids were little, but at Tahoe in 2002, I shot about 5 rolls of 35mm film one day. Next year, 2003 I had a new 2MP Canon, and for 2004 I had a SD10.
Best regards, Sandy
[email protected]
http://www.pbase.com/sandyfleischman
Costas Vlachos
Forum Enthusiast
Hi all,
I've been reading with great interest some of the posts on this thread that mentioned Nyquist's Sampling Theorem and how it relates to Foveon and Bayer sensor systems. I just wanted to post this because I believe there was a fair amount of misinformation posted by some, that may lead readers to the conclusion that Nyquist's Sampling Theorem [1],[2] doesn't apply to Foveon sensors.
First of all, any claim that Nyquist's Sampling Theorem does not apply to Foveon sensors is completely false and comes from the poor understanding of the theorem and its implications. The thing to be aware of is that whenever there is sampling taking place, Nyquist's Sampling Theorem ALWAYS applies. No exceptions! There is no point in trying to disprove or deny this, as it is a mathematically proven fact. It doesn't matter if the sensor system is Bayer-based or Foveon-based. The exact same principle applies to audio sampling and the sampling of any other signal.
There were some posts mentioning about a resolution chart containing 9 lines, which appear as 5 lines in an image captured by a Foveon-based system. This is Nyquist's Sampling Theorem at work! In fact, Nyquist's Theorem can be used to mathematically compute how many lines the image will contain, based on the sensor's resolution and the object's line frequency. The formulae are all there and can be used by those who understand them. If the object contains 9 lines and the captured image shows 5 lines, what we see is an "alias" of a frequency component. Loosely speaking, a component with a frequency of "9 lines per length unit" is now represented by an "alias" (read "fake" or "ghost") component with a lower frequency of "5 lines per length unit". This component DOES NOT EXIST in the actual object being photographed. This is a classic result of undersampling. In fact, a good sampling system should have a low-pass filter in the signal path before the sampling is performed, so that those frequencies that are higher than the Nyquist limit are filtered out, resulting in a grey area with no visible lines. This is a MORE ACCURATE representation of the 9-line object than the false 5-line image coming out of the Foveon sensor. Yes, "mush" or "averaging" is better than 5 lines! It may not sound right but it is. It's better to show NO detail, than to show WRONG detail. Just read any good book on sampling theory.
The benefit of the Foveon system is that it doesn't need to interpolate adjacent pixels to generate the colour information. A Bayer system results in a lower Nyquist frequency limit (samples are "coarser") because of the necessary interpolation. No arguments here, Foveon is a superior system. But to claim that Nyquist's frequency limit and other implications of his sampling theorem somehow magically do not apply to Foveon systems is totally wrong! His bones must be shaking in his grave by some of the claims in this thread...
Sorry for the technical content, but after reading some of the posts I felt I had to add this so that any confusion is resolved and misinformation pointed out. I don't mean to insult anyone, I'm merely trying to set the record straight on this very misunderstood subject.
Regards,
Costas
---
References:
[1] Nyquist, H. (1928). "Certain topics in telegraph transmission theory." Trans. AIEE, Vol. 47, pp. 617-644.
[2] Shannon, C. E. (1949). "Communication in the presence of noise." Proc. Institute of Radio Engineers, Vol. 37, No. 1, pp. 10-21.
I've been reading with great interest some of the posts on this thread that mentioned Nyquist's Sampling Theorem and how it relates to Foveon and Bayer sensor systems. I just wanted to post this because I believe there was a fair amount of misinformation posted by some, that may lead readers to the conclusion that Nyquist's Sampling Theorem [1],[2] doesn't apply to Foveon sensors.
First of all, any claim that Nyquist's Sampling Theorem does not apply to Foveon sensors is completely false and comes from the poor understanding of the theorem and its implications. The thing to be aware of is that whenever there is sampling taking place, Nyquist's Sampling Theorem ALWAYS applies. No exceptions! There is no point in trying to disprove or deny this, as it is a mathematically proven fact. It doesn't matter if the sensor system is Bayer-based or Foveon-based. The exact same principle applies to audio sampling and the sampling of any other signal.
There were some posts mentioning about a resolution chart containing 9 lines, which appear as 5 lines in an image captured by a Foveon-based system. This is Nyquist's Sampling Theorem at work! In fact, Nyquist's Theorem can be used to mathematically compute how many lines the image will contain, based on the sensor's resolution and the object's line frequency. The formulae are all there and can be used by those who understand them. If the object contains 9 lines and the captured image shows 5 lines, what we see is an "alias" of a frequency component. Loosely speaking, a component with a frequency of "9 lines per length unit" is now represented by an "alias" (read "fake" or "ghost") component with a lower frequency of "5 lines per length unit". This component DOES NOT EXIST in the actual object being photographed. This is a classic result of undersampling. In fact, a good sampling system should have a low-pass filter in the signal path before the sampling is performed, so that those frequencies that are higher than the Nyquist limit are filtered out, resulting in a grey area with no visible lines. This is a MORE ACCURATE representation of the 9-line object than the false 5-line image coming out of the Foveon sensor. Yes, "mush" or "averaging" is better than 5 lines! It may not sound right but it is. It's better to show NO detail, than to show WRONG detail. Just read any good book on sampling theory.
The benefit of the Foveon system is that it doesn't need to interpolate adjacent pixels to generate the colour information. A Bayer system results in a lower Nyquist frequency limit (samples are "coarser") because of the necessary interpolation. No arguments here, Foveon is a superior system. But to claim that Nyquist's frequency limit and other implications of his sampling theorem somehow magically do not apply to Foveon systems is totally wrong! His bones must be shaking in his grave by some of the claims in this thread...
Sorry for the technical content, but after reading some of the posts I felt I had to add this so that any confusion is resolved and misinformation pointed out. I don't mean to insult anyone, I'm merely trying to set the record straight on this very misunderstood subject.
Regards,
Costas
---
References:
[1] Nyquist, H. (1928). "Certain topics in telegraph transmission theory." Trans. AIEE, Vol. 47, pp. 617-644.
[2] Shannon, C. E. (1949). "Communication in the presence of noise." Proc. Institute of Radio Engineers, Vol. 37, No. 1, pp. 10-21.
Keyboard shortcuts
- f
- Forum