Anti-aliasing, Color Separation, etc.

Discussion in 'Digital Photography' started by Paul H., Dec 17, 2003.

  1. Paul H.

    Paul H. Guest

    For the last several weeks on this newsgroup, a largely pointless and nasty
    discussion has been raging about Foveon vs. Bayer, Sigma vs. Everything
    else, etc. In any case, topics such as color separation and aliasing have
    been employed by a precious few people who tried to keep the discussion
    centered about the merits of the argument. For those confused
    non-illuminati who need a non-technical but NON-BRAND SPECIFIC intro to such
    technical matters, take a look at

    Not interested? Well, just remember that if you arm yourself with but a
    _little_ knowledge, you, too, can be dangerous!

    An editorial comment: I have become convinced that in any conflict,
    including the Foveon vs. Bayer controversy, perhaps only 2% of the people
    involved in the fight on either side have any clue regarding the actual
    reasons behind the dispute. The remaining 98% of the combatants simply hide
    behind these few leaders, capering about in monkey-like fashion,
    occasionally throwing dull, ill-aimed spears at the "enemy" while shouting
    "Yay, us!" or "Yeah--what HE said" and following up with uncreative streams
    of profane-sounding grunts.

    It's quite depressing, actually, though it can be entertaining. What's even
    more depressing is that given the right subject, I know I'll be sucked into
    being one of the monkeys myself, only to realize my role after the fact.
    Paul H., Dec 17, 2003
    1. Advertisements

  2. Paul H.

    DJ Guest

    It may be "NON-BRAND SPECIFIC" but it's quite specific about not listing any
    disadvantages for vertical colour filters.

    IMHO the author also displays a lack of understanding of, or choses to ignore
    the truth about, anti-aliasing. ALIASING IS NOT A CONSEQUENCE OF LATERAL COLOUR
    FILTERING. It is a consequence of having a regular array of sensors. Fly wire
    doesn't have colour filters but if you look throuigh two layers of flywire and
    rotate one you will see Moire patterns (aliasing). I expect aliasing becomes
    worse in a latteral sensor with the same average detector pitch, because the
    pitch per colour is lower, but that is NOT the same as saying or implying that a
    vertical sensor does not exhibit aliasing.

    I believe that vertical sensors have the *potential* to produce better cameras
    than latteral sensors. I also understand that current vertical implementations
    have shortcomings.
    DJ, Dec 17, 2003
    1. Advertisements

  3. Paul H.

    eawckyegcy Guest

    But what is the "article" is a submarine-propaganda piece written by a
    Foveon dude?

    "note by Dick Merrill (Foveon)"

    Well, I guess we can trust him to be impartial, right? But what if
    the article makes incomplete statements, like:

    "So 2/3 of the color information in every output
    pixel is derived using complex software algorithms,
    instead of measured."

    [Prove you are not a monkey and explain why the above is an incomplete

    And what if the "non brand specific" article includes test charts that
    are designed to make "brand X" sensors look bad, but "brand Y" good,
    without any comment at all about the realism of such charts?

    Given all this (and more) are you really sure this is such a good
    article to direct the "non-illuminati" at for a completely unbiased
    eawckyegcy, Dec 18, 2003
  4. Paul H.

    Paul H. Guest

    Paul H., Dec 18, 2003
  5. Paul H.

    Paul H. Guest

    I don't own a Foveon-based camera, I don't plan to own a Foveon-based
    camera, all of my cameras are based upon the Bayer-filtered sensor, and my
    next camera, the Canon 300D, will be a Bayer-type camera. Nevertheless, I
    found the information presented on the site to be factual, fairly unbiased
    and I still contend the page can serve as a good introduction to some of the
    topics currently under discussion in this newsgroup. I assumed, of course,
    a certain intelligence on the part of the reader which would lead him or her
    to seek out other sources for futher information, if desired. But more to
    the point, you could have used your own post to direct beginners (the
    non-illuminati) to better, less-biased (in your opinion) sources of
    accessible technical information, but you did not; instead, you chose to
    quibble and insinuate. And that, my friend, is the real problem with these days.

    As to the author's Foveon associations, well, if Adolf Hitler himself once
    said the sun rose in the east, the character of the claimant didn't make the
    observation any less true. Maybe you need to stop seeking hidden agendas in

    Finally, I did not mean this thread to turn into another I-hate/love-Foveon
    travesty of debate, but you, having taken that road, impel me to make a
    comment: While it may be true that the current Foven-style sensor isn't
    very good, I think the Foveon goal of one RGB sensor per detector photosite
    is a fine idea and one that ought to be pursued. There are no "heresies" in
    digital cameras, no God of Bayer bestowing His favor upon true believers and
    punishing the heretics, there is only the physics of what works and what
    doesn't work. For my part, I don't give a rip if the sensor is made of spun
    sugar if gives me good pictures.
    Paul H., Dec 18, 2003
  6. It's a Foveon ad.
    And it's technically misleading. In ways that are, not coincidentally,
    advantageous to Foveon.

    "Unfortunately, blurring of color information to suppress the color
    artifacts reduces image sharpness as well."

    This section careful neglects the point that Foveon sensors require
    antialiasing just as much as Bayer sensors do. Not only that, it sleazes:

    "5. an "anti-Aliasing" filter is not required to suppress color aliasing,
    reducing camera cost and increasing image contrast."

    This is a sleaze. An AA filter is required in all discrete sampling systems
    whose input contains frequences at or above the Nyquist frequency. AA
    filters are not to suppress color artifacts, they're to suppress aliasing, a
    phenomenon that occurs in Foveon sensors as much as it does in Bayer
    You seem to have missed the sleaze.
    Sleazy advertising is not a trifle.

    David J. Littleboy
    Tokyo, Japan
    David J. Littleboy, Dec 19, 2003
    Rodney Myrvaagnes, Dec 19, 2003
  8. Paul H.

    Paul H. Guest

    Simply put, you're a zealot with an undue penchant for hyperbole. I'd hardy
    call the the so-called advertising you mention "sleazy", a word more
    appropriately used to describe the practice of selling child pornography,
    not the marketing of digital cameras. Get real. Secondly, I have no
    personal stake in the matter, but simply wanted to point beginners to a page
    giving out simple definitions for phenomena such as aliasing. I happened to
    point them to one of my many bookmarks, a site which I still contend is
    useful if one sticks to technical matters and ignores opinion. You, on the
    other hand, gave out nothing but venom when you could have instead directed
    people to a site or sites you thought more appropriate and/or brand-neutral.
    Thanks for furthering understanding and yet again hijacking another message
    thread to I-hate-Foveon Land.

    You have an Ahab-like fixation on the Foveon sensor that's nearly
    unbelievable; I can even imagine your bedroom littered with slashed
    matresses and torn-open pillows, the sorry renmants of a search undertaken
    because you thought you heard a Foveon whispering evil things whenever the
    lights were out. Get a grip.
    Paul H., Dec 19, 2003
  9. They misrepresent technical issues to sell cameras. That's sleazy. Selling
    child pornography is a crime.
    The definition given is misleading, and the statement about the need for an
    antialiasing filter with Foveon sensors is incorrect.

    The basic concepts have been described here time and time again. Correctly,
    in language that's reasonably free of jargon. Did you miss those posts?
    You happen to be wrong on that contention. That page is a Foveon ad, not a
    technical article.

    I doubt that you'll be happy with that, but it's quite a good page. The last
    time I went looking, I was looking in the Electrical Engineering department
    (which is where this stuff was taught in the mid-70s) and didn't find it.

    Normal Koren's site has a wealth of good stuff. I've pointed a lot of people
    to it. But I don't think it has anything on aliasing. (I think it simply
    assumes that the cameras of concern are implemented correctly.)
    I've got fewer posts in these threads than just about anyone else here<g>.

    FWIW, if the SD9 and SD10 cameras had anti-aliasing filters, and if Foveon
    was arguing that Bayer cameras required a stronger AA filter and that the
    Foveon sensor could deliver, say 66% of the Nyquist frequency whereas Bayer
    cameras could only deliver 55% without objectionable artifacts, and that
    Foveon therefore delivered 66/55 or 20% better resolution than (or 1.44
    times as many megapixles as) Bayer sensors, I'd be on the Foveon side of the
    fence. But that's not what they're saying. (These numbers are guesses. It's
    probably more like 66% vs. 60%.)

    David J. Littleboy
    Tokyo, Japan
    David J. Littleboy, Dec 19, 2003
  10. Paul H.

    pehache Guest

    OK. You reached the "signal processing, level 1". You have understood
    what is aliasing.

    Now try to reach the "Signal processing ,level 2". To be short:

    In a Bayer sensor without AA filter, the luminance component is
    slightly aliased, and the chrominance components are severely aliased.

    In the Foveon sensor without AA filter, the luminance and chrominance
    components are slightly aliased.

    As a consequence, a Bayer-based sensor should have an aggressive
    AA-filter to completely dealiase the chrominance componants (at the
    expense of an unecessary blurring of the luminance component), and a
    Foveon sensor should have a light AA-filter to dealiase all its

    The choice made by the vendors are compromises: light AA-filters on
    Bayer's where a strong one would be needed, and no AA-filters on
    Foveon's where a light one would be needed. In both cases, some
    residual aliasing is left (mainly chrominance for Bayer, mixed
    luminance/chrominance for Foveon).

    Saying that the Foveon does not require AA-filter is false, or at
    least a strong shortcut. But what is true is that the Foveon requires
    less AA-filtering than Bayer's ("no" versus "light" is "less", as well
    as "light" versus strong is "less").
    pehache, Dec 19, 2003
  11. There's never a Bayer sensor without an AA filter. (Except for Kodak 14n and
    some of the MF digital backs, and the 14n exhibits nasty artifacts.)
    There's also never a Foveon sensor without an AA filter. At least not on my
    shelf there isn't.
    The Foveon AA filter must completely, 100%, totally, attenuate all
    frequencies at and above Nyquist.

    Any practical AA filter that does that will, I suspect, be adequate for a
    Bayer sensor. If not, it will be very close to adequate.
    That _may_ be true. (See my other note.*)

    Seriously investigating what
    the practical effects of that difference would be is an interesting
    question. But that's not what Foveon is doing, so it's irrelevant to the
    cameras we have as options. The better Bayer cameras are quite good at
    minimizing artifacts, so as a practical issue were faced with fairly
    reasonable camera implementations on the one hand and an incorrect
    implementation on the other.

    David J. Littleboy
    Tokyo, Japan

    *: Here's what I said.FWIW, if the SD9 and SD10 cameras had anti-aliasing filters, and if Foveon
    was arguing that Bayer cameras required a stronger AA filter and that the
    Foveon sensor could deliver, say 66% of the Nyquist frequency whereas Bayer
    cameras could only deliver 55% without objectionable artifacts, and that
    Foveon therefore delivered 66/55 or 20% better resolution than (or 1.44
    times as many megapixles as) Bayer sensors, I'd be on the Foveon side of the
    fence. But that's not what they're saying. (These numbers are guesses. It's
    probably more like 66% vs. 60%.)
    David J. Littleboy, Dec 19, 2003
  12. Paul H.

    JPS Guest

    In message <bru9ro$tst$>,
    The SD9 records at least 2 or 3 times the nyquist upon capture, but of
    course, it's all below the nyquist in the output file, and 125% of the
    nyquist becomes 75% of the nyquist, 190% of the nyquist becomes 10% of
    the nyquist, 300% the nyquist becomes the nyquist, etc, etc. There are
    no microlenses and the photosites only cover 30% of the sensor surface.
    JPS, Dec 19, 2003
  13. Paul H.

    JPS Guest

    In message <>,
    That would be very nice if they came in the resolutions that monochrome
    sensors with CFAs came in.

    I have my doubts that we will see practical full-RGB senesors for
    high-res digicams anytime soon. As long as you can get more spatial
    witnesses with CFAs, you will get better luminance resolution with them.
    These sensors we're discussing are *far* from exausting the detail
    delivered by the sharpest lenses available. I can put a Canon 1.4x TC
    *AND* a Tamron 2x SP on my 300mm F4L IS lens, and it still delivers 84%
    the resolution of the lens by itself, as limited by the 10D's sensor.
    That means that the lens has at least 2.37x the resolution that the 10D
    sensor can handle, and probably almost 4x as much as the SD9's csensor
    can, if you could mount it on one. And that lens is a toy, compared to
    the f2.8 version.
    JPS, Dec 19, 2003
  14. David Littleboy is one of two or three regular contibutors of actual
    technical information to this news group. I have learned a lot from
    his posts.

    "Sleazy" is a reasonable word for the stealthy presentation of
    misleading marketing hype in the guise of neutral background.
    The site is so organized that a beginner wouldn't be able to do that,
    as apparently you also are not. I took it at face value until I
    noticed the grossly impossible color-response graph for stacked
    sensors and the absence of a "Disadvantages" list for the stacked

    Rodney Myrvaagnes J36 Gjo/a

    "In this house we _obey_ the laws of thermodynamics." --Homer Simpson
    Rodney Myrvaagnes, Dec 19, 2003
  15. Paul H.

    Paul H. Guest

    I'm perfectly happy with stuff like this (I have degree in physics and years
    ago even coded discrete transforms in assembly language) but I was trying to
    find something understandable by and useful to people who have somewhat less
    than a nodding acquaintance with concepts like the delta function and
    distrubution theory. Most people, believe or not, have had very little
    exposure to either discrete (or continuous) Fourier transforms and yet would
    like to acquire a workman-like/intuitive understanding of concepts such as
    aliasing, as well as other technical matters critical to making a decision
    to purchase a high-quality camera. An intelligent person can, for example,
    understand refraction without being able to state Snell's Law.

    But thanks for the link. :)
    Well, sorry about the Ahab comment, but no one likes being termed a stooge
    for anyone else's point of view, even by implication. I don't necessarily
    disagree with many of the more rational complaints about the Foveon
    approach, but I am tired of, and sensitive to, the intimations that somehow
    Foveon was started solely to produce and market worthless crap, and that any
    engineer who ever associated himself with Foveon (or Sigma) is somehow
    tainted so badly his statements can never be trusted again.
    Paul H., Dec 19, 2003
  16. Paul H.

    Paul H. Guest

    Grossly impossible? Grossly over-simplified, perhaps, but no more so than
    was the related graph for the Bayer-type sensor; I simply thought the graphs
    were intended to be schematic and illustrative rather than truly
    mathematical in nature. I must admit, though, I had overlooked the lack of
    a "disadvantage" list for stacked sensors and that is a real oversight by
    the author. However, after re-reading the page I still don't see the
    extreme bias to which others have referred: Indeed, even the conclusion
    comparing stacked sensors to Bayer-type sensors was fairly bland; at no
    point did the author ever say anything approaching "therefore, stacked
    sensors are far superior." Thus while "bias" may be present, I don't see
    real advocacy and I still think the phenomenon of spatial aliasing was
    illustrated effectively.

    What I don't understand, however, is the extreme dislike of Foveon/Sigma
    exhibited by many on this newsgroup: after all, Foveon sensors represent a
    mere blip in the digital camera marketplace and, to me, at least seem more
    of an interesting technical curiosity than a threat to the future of digital
    cameras. However, what I was trying to do was NOT contribute to the
    controversy, so if I can find some other elementary and far less
    controversial sites, I'll post the url's, then put my shields up.
    Paul H., Dec 19, 2003
  17. Paul H.

    eawckyegcy Guest

    It is a thinly disguised Foveon advertisement. Wasn't it obvious?
    Then you assume intelligence that you yourself have not evidenced.
    I could have solved a nasty differential equation too, "but I did
    not". What is the point of your distraction here? Right: that
    _your_ reference is nothing more than a warmed over Foveon ad and you
    didn't pick up on it. I am now obligated to fix your mistake? Get
    real, dude.
    Maybe you need to find your brain; you have apparently lost it.
    Nasty reading comprehension problem you have: I said absolutely
    nothing at all good or bad about Foveon in my response to you.
    eawckyegcy, Dec 19, 2003
  18. Paul H.

    Paul H. Guest


    And what a nasty attitude _you_ have. Here's a quote from your original
    response to me:
    I hardly think the phrase "submarine-proganda piece" is a neutral comment
    regarding Foveon, but perhaps your "writing comprehension" needs some work.

    However, flame me if it pleases you, but where is your link to a site
    containing elementary technical information? You don't provide one and the
    tenor of your posting strongly suggests you're far more interested in a
    meaningless fight than in discussing technical matters.
    Paul H., Dec 19, 2003
  19. Paul H.

    Larry Lynch Guest>,

    I dont think anybody here "HATES" the Foveon chip.. It
    seems to be an experiment with merit.

    What we "HATE" here are idiot zealot(s) (not speaking of
    you here just the idiot(s)) who repeat marketing hype,
    and outright LIES in the guise of information that might
    lead a newbie down the path to purchase a camera that is
    "NOT READY FOR PRIME TIME" and may NEVER be ready.

    The Foveon is an experiment, and a damn fine one, which,
    with a little work in the right direction MIGHT be a
    leap forward in digital cameras.

    Sadly, so far that leap has NOT taken place, and the
    work doesn't seem to be happening, and the Sigma cameras

    We have seen a plethora of posts from certain people
    over and over again praising this not ready for prime
    time sensor/camera combo, so we get TESTY about it.
    Larry Lynch, Dec 19, 2003

  20. I spent 16 years covering the semiconductor industry for an engineer's
    trade magazine. I have listened to many intelligent technical people
    talk pure hype when they are put in a marketing situation.

    I have also seen them become honest engineers again afterward.

    Sliding past the disadvantages of whatever you are selling is standard
    practice. What elevated that web site into the 'sleaze' category was
    disguising its marketing origin and pretending to be an overview.

    If I had received that web file as an article proposal, my first
    reaction woud have been "We want it, but it needs work."

    It would have gone back to the author with such comments as "You left
    off the disadvantages of the stacked sensor approach." Surely there
    are some.

    Second, I would have suggested that graphs of color sensitivity should
    reflect actual silicon behaviour, rather than a fantasy. Since I would
    have been busy on other things, I would probably have accepted the
    article if he fixed those problems.

    But, I might also have found time to read it more closely, and found
    other problems. The easily-spotted gaps in the first submission woud
    put the BS meter on high alert.

    If the article were accepted, the author identfication would include
    his position in the marketing dept. of the company.

    Rodney Myrvaagnes J36 Gjo/a

    "In this house we _obey_ the laws of thermodynamics." --Homer Simpson
    Rodney Myrvaagnes, Dec 20, 2003
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.