Steve's Digicam's classifies the SD-9 and SD-10 as 10.2MP

Discussion in 'Digital Cameras' started by George Preddy, Oct 27, 2003.

  1. George Preddy

    Guest Guest

    You quite clearly have problems getting passed Sigma's BS too...
     
    Guest, Nov 3, 2003
    #41
    1. Advertisements

  2. More or less correct, and that is optically very dirty compared to simply
    knowing colors due to exactly equal weights of co-located RGB samples.
    If recycling already used red and blue data to combine with the green
    orphans is optically superior to using complete, original, co-located RGB
    data only once for each discrete output pixel, why not have 1 Red, 1 Blue,
    and 5,999,998 Green sensors in a "6MP" Bayer? Makes no difference to
    recycle, right? In fact, it's better to re-use the same red and blue color
    data over and over again, right?
    I once thought that too, I've slowly come to realize that but a tiny few
    realize it. I've only read one pro website (outbackphoto.com) that even
    acknowledges there is a difference between a black and white resolution test
    and a color resolution test when comparing color digital sensors. Crazy.

    When a pro reviewer's job is to learn this stuff, it's surprising to me
    only one site gets it. Foveon even produced and distributed an
    all-primary-combos color resoltuion test chart for them, but only
    outbackphoto.com had the guts to publish any results--guess those Canon
    pre-release perks are worth a lot more than educating a readership properly.
    Go to dpreview.com right now, you'll see they compare a color Foveon to
    color Bayer sensors using B&W resolution targets, charts designed to test
    lens optics from the days the sensor could be controlled (film). Absurd
    professional incompetence. I have trouble believing buffoonery like that is
    purely accidental.
    We can all learn.
    If you have only seen that one sample, I'm surprised you are arguing with
    me, to be honest. What are you basing your opinions upon? I have about
    10,000 Bayer images (most 6MP) afteditching more and about 4500 SD-9 images.
    That is what I'm basing my opinions upon. Not a huge amount lot to be sure,
    but enough to draw a solid conclusion when the difference is so vast. My
    worst (focused) SD-9 image is optically sharper than my best Bayer image,
    and not by a little, by a lot.
    Were you there? Color preference isn't set by the SD-9, it is set by the
    user.
    I don't see it here. What color temp is your monitor set to?
    Why is getting full sized unprocessed images from 6MP DSLRs users like
    extracting impacted wisdom teeth?

    pbase.com is free under 10MB, how about a few? Out of the few full size
    10D images I've seen on pbase (out of about 132,000) most appear highly
    postprocessed/USM'd/artifact prone, and are pretty blurry anyway.
    Its ok, better sharpness than Canon's demo images, so good job. Not Foveon
    sharp, but if you are beating Canon with their own camera I think you are
    doing great.
     
    George Preddy, Nov 3, 2003
    #42
    1. Advertisements

  3. George Preddy

    MarkH Guest

    And how would you arrange those pixels so that every pixel had at least
    two neighbours of red, 2 of green and 2 of blue?
    Colour preference is fine, whites are white. It is the colour anomalies
    that I am talking about.
    You don’t see it? Are you looking at the full sized image? The colour
    temp of my monitor is set just fine, the yellow is VERY strong. Look at
    the full sized image (the 3.43MPix one) and look at each pair of legs
    carefully. If you can’t see the yellow then you should get yourself
    checked for colour blindness.
    A few? The full image that I posted a crop of is 3MB, another on the
    same day is 3.6MB. I have posted a crop that has 1:1 actual pixels,
    what’s wrong with that. In this country broadband is very limited, for a
    fast (7-9Mbs) connection there is a data cap of 600MB per month before
    paying a fortune per gig.
    I just realised that picture was taken at ISO 400 so it’s not the very
    best my camera can do (but that doesn’t affect sharpness). With that
    lens I regularly get that degree of quality (Canon 28-135 IS). I just
    don’t see any compelling reasons to consider buying another camera.

    I know I could get sharper images by setting the sharpness to 0 and
    working the image in Photoshop, but I am happy with my results and can’t
    be bothered wasting 10 minutes manipulating every image.

    I also care about other features like the 3fps for 9 frames and being
    able to take 3 bursts of 9 frames within 25 seconds.

    I also like IS lenses, how many are available for the SD-9?
     
    MarkH, Nov 3, 2003
    #43
  4. No need, you can use any two sensors mathematically. BTW, Foveon has 8
    pixels of every color surrounding every other pixel.
    Color accuracy is higher than Canon. AWB is not as good, I would say, but
    also never permanently set and the tools to set it are a lot more
    professional. I can't imagine not wanting better control given that all
    digitals miss WB, only a matter of to what degree.
    It's not yellow here at all. Probably a monitor differences, try looking at
    it somewhere else and see if the color looks different to you.
    Ouch, how do you cope with digital at all underthose conditions? I'm used
    to a lousy 300K/sec, it seems slow already. :)

    Something you may not have thought of, the SD-9 will save you a fortune at
    those rates. Non-interpolated images would make perfect sense for you. A
    SD-9 10.3MP-non-interpolated PNG is about the same file size as a 10D's
    6MP-interpolated JPEG, and that's with both higher optical resolution and
    lossless quality. One of the nicest hidden luxuries of the SD-9 is 75%
    bandwidth savings with no optical penalty. Even when you aren't dealing
    with the internet, the bandwidth savings can be really refreshing--assuming
    you want pro quality (i.e. RAW) all or most of the time.
    Actually, it can.
    The SD-9 only does 2.5 fps at 6MP-interpolated, but it stores 14 and shoots
    continuously through the flush after that at about 1/sec. That is all RAW
    mode, of course. I never dip to that resolution, though. Its 2fps at
    10.3MP, same as the S2 Pro at 6MP (though again, Fuji lies and says 3fps,
    they always add the frame that happens at the top of the next second, as if
    that is a "rate") .

    I also like the SD-9's faster operating speed over the 10D, the 10D seems
    prosumer slow to me during image review and inspection. Probably all what
    you get used to. But these are two very different cameras, the SD-9 is
    aimed at pros wanting museum quality output with a RAW-only workflow and the
    high end tools to match, the 10D is aimed at pro's doing quicker low
    res/JPEG work and amateur enthusiasts. Though both are very affordable
    these days, the 10D only costing about 50% more.
    Just one, the 80-400mm EX OS. An incredible value for the build quality and
    optics. I'm torn between it and the 50-500mm APO, which has an stellar
    reputation for sharpness across the entire range. The SD-9 can also image
    the sharpness of these great lenses optically, the 10D, unfortunately,
    cannot.
     
    George Preddy, Nov 4, 2003
    #44
  5. George Preddy

    MarkH Guest

    Unfortunately you can’t understand the difference between colour balance
    and colour anomalies, since you can’t understand there is little point in
    explaining again.
    I have 2 monitors, white looks white on both. Now is the time for you to
    get yourself checked for colour blindness, not that uncommon a problem in
    males. I have been tested and have very good eyesight.

    I think if the problem was with my monitor then skin tones would look
    yellow on other images.
    Unfortunately I have to use a slower 128K/sec connection so that my
    provider will let me have a reasonable (10GB) data limit per month.
    Here there is a dramatic difference between Sigma and Canon, Canon make
    several IS zooms and several IS Primes. My 28-135 IS is not only a good
    sharp lens, but it is also very reasonably priced. That lens is my
    normal ‘walking around’ lens.
     
    MarkH, Nov 4, 2003
    #45
  6. George Preddy

    ThomasH Guest

    This group is full of George! Well, I just found on:
    http://www.steves-digicams.com/2003_reviews/sigma_sd10.html

    [...
    Image sizes up to 2263 x 1512 pixels (3.43 Megapixels)
    Images stored as lossless compressed RAW, 3:2 aspect ratio
    ....]

    you just read what suits your theory better! :)

    I like this religious war, its better than Canon-Nikon war for a change!

    Thomas
     
    ThomasH, Nov 4, 2003
    #46
  7. Popular Photography can't understand it either, they rate Sigma higher than
    Canon in color accuracy and noise. Here's another decent side by side with
    an SD-9 and 1Ds, though mostly anecdotal:

    http://www.outbackphoto.com/reviews/equipment/sigma_sd9/sigma_sd9_Diary.html
    They're not yellow here, so something is clearly different. It really
    doesn't matter though, it's adjustable in SPP.
    The SD-9 would've be a godsend for you from a bandwidth perspective.
    Believe it or not a 10.3MP .PNG is only about 4.5MBs, that is,
    non-interpolated. Still, with a sensor count this high you can get into
    some extreme bandwidth issues very quickly, a 13.7MP 48-bit TIF saves out at
    about 100MB per image if you keep the RAW file. Crazy.
    Unfortunately, no lens is a very sharp lens on a 6MP (interpolated) Canon
    DSLR. I say that a bit sadly as the 10D is a great all around camera
    otherwise, but it's a problem that I personally can't overlook. The 1Ds
    would benefit from L glass as it is pro quality sharp, but it is priced a
    bit absurdly, as are most L lenses for that matter.
     
    George Preddy, Nov 5, 2003
    #47
  8. George Preddy

    MarkH Guest

    Congratulations, you found a website with SD-9 pictures that includes NO
    pictures of people, that’s the best way to make the SD-9 look good.
    Have you had your eyes checked yet?

    The colour problem is not fixable, no matter where the colour balance is
    set the problem is still there.

    The eyedropper in Photoshop verifies what my eyes see, the skin tones are
    very yellow.

    If you can’t see the yellow legs, get your eyes checked!

    Also please note, I am not going around looking for bad examples from the
    SD9 to point to and say ‘look the camera is no good’, I am simply
    following the links that you provide and seeing the skin tones are Foveon
    yellow.
    Are you nuts? Do you think that I would consider buying an SD-9 so that
    I could upload pictures for you to look at? I have never said that I
    have bandwidth problems, it is only your request for several full sized
    samples that I am not interested in.
    The 10D has a greater resolution per mm on its sensor. The 10D produces
    what is essentially a crop of a 15.6MPix full sized sensor. The 10D
    therefore needs better glass for the part of the image that is near the
    centre of the image circle. If a lens is a little soft at the corners
    then that wouldn’t be a problem.

    Of course my point was the greater availability of IS lenses, which is
    valid.
     
    MarkH, Nov 5, 2003
    #48
  9. I have tons of pics of people...

    http://www.pbase.com/image/21065680

    You are mistaking an AWB issue (with a camera that never sets WB) for a
    color issue.
    Ok, so its not a bandwidth problem, but a refusal to share full size
    unprocessed images. That's fine, so far no 10D users have been willing to
    post full sized unprocessed images.

    Trust me, I've seen plenty.
    It is 100% valid.

    Pointing out that a soft sensor can't make full use of any pro lens is also
    valid. L glass does not need sharpening/USM. At least not when placed in
    front of film. Same for HF and EX glass, when placed in front of the Pro
    10M.
     
    George Preddy, Nov 5, 2003
    #49
  10. George Preddy

    MarkH Guest

    I’m afraid that you misunderstand, I have never claimed that every picture
    will show yellow skin tones. All I am saying is that is several pictures
    there has been a problem with flesh looking very yellow, this makes me
    think that the camera has problems at times producing the right colours.
    This is a problem that I have never seen on the 10D.
    No, I have no problem with white balance on the picture:
    http://www.pbase.com/image/17597457/original
    But even though whites are white, some flesh tones are Sigma yellow. White
    balance may not be spot on, but it is not the cause of yellow leg syndrome.
    If it was a white balance issue then everything would look yellow, or at
    least all the legs.

    Then why are you getting so upset that people don’t feel inclined to
    provide you with what you want?
     
    MarkH, Nov 5, 2003
    #50
  11. For comparison sake I took a few SD-9 cats, no USM...

    http://www.pbase.com/image/23010761
    http://www.pbase.com/image/23008820
     
    George Preddy, Nov 5, 2003
    #51
  12. That is incorrect info, the max SD-9/10 image size is 4536 x 3024
    interpolated (optionally) .
    It is fun. For the SD-9 users. :)
     
    George Preddy, Nov 5, 2003
    #52
  13. Have you tried another monitor, it could be that my laptop is blue temped
    farther than yours. The best thing to do is request the RAW file and see,
    skin tones/color is fully adjustable.
    Because they are more than willing to argue over image quality, yet, oddly,
    no 10D user is willing to post a full sized unprocessed photo. I know why,
    I've seen plenty, but if they want to be considered competitive in the 10+
    MP professional class of DSLRs (which I define as 35mm film quality and up),
    the way to do it is with images, not talk.
     
    George Preddy, Nov 5, 2003
    #53
  14. But let me guess. Those 13+MB performs worse than a 1.5MB 10D/D100
    (Your figures) on a BW resolution chart? Man, that sensor must really
    suck.
     
    Asbjørn Bjørnstad, Nov 6, 2003
    #54
  15. The amazing thing is the SD-9 outperforms the 6MP Bayers even in B&W. It
    has no has no business winning a no-color-interpolation-required battle.
    Ref:
    http://www.dpreview.com/reviews/sigmasd9/page23.asp
     
    George Preddy, Nov 6, 2003
    #55
  16. George Preddy, Nov 6, 2003
    #56
  17. I'm only on a 8-bit display right now, but I had a look.
    Yup, the 1.5 Mp D100 resolves a lot better than the 3.4Mp SD-9. Look
    at those vertical resolution bars around the 14 mark. You can't really
    tell how many lines there are in the sd-9, but the D100 easily
    separates the lines. And that is with only one third of the sensors
    (According to you.) Amazing.

    Oh, and look at the diagonal lines to the right of the vertical
    resolution bars. The region marked 10 has some weird banding pattern
    perpendicular to the lines.) that isn't visible on any of the other
    regions. Why is that? Maybe it's because the D100 don't resolve that
    fine detail.
     
    Asbjørn Bjørnstad, Nov 6, 2003
    #57
  18. You're too late, dpreview already analyzed and rated the cameras' B&W
    performance for you, the SD-9 beats all of the others. In particular...

    Camera
    Catgory: Absolute, Extinction
    -----
    Sigma SD9
    Horiz LPH: 1550, >2000
    Vert LPH: 1550, >2000
    5° Diagonal LPH: 1000, n/a

    Nikon D100
    Horiz LPH: 1600, * 1850
    Vert LPH: 1300, * 1700
    5° Diagonal LPH: 1000, n/a
    Whats really amazing is that this is using a B&W target where the Bayers
    overachieve relative to their full color optical resolution by 300 to 400%.
    Ref:

    http://www.outbackphoto.com/artofraw/raw_05/crop_10D_0000__0171_RT8.jpg
    http://www.outbackphoto.com/artofraw/raw_05/crop1_sd9_0000_00200.jpg
     
    George Preddy, Nov 6, 2003
    #58
  19. BTW, there are heavy duty sharpening artifacts evident all around the
    wiskers, look at how bad they are around the "main whisker" that sticks out
    into the bottom righthand corner. This is a good example of how destructive
    USM can result well below prosumer image quality.
    Note the total lack of artifacts. The "eyelash" crossing the right eye is
    a good example o the Foveon's single pixel resolution. Bayer sensors need
    3-4 pixels to resolve a feature, due to 4:1 spatial upscaling during the
    color interpolation process.
     
    George Preddy, Nov 6, 2003
    #59
  20. George Preddy

    MarkH Guest

    They look fine to me, I am happy with my 10D pictures and I would be happy
    with pictures like your cat ones.

    At normal magnification there does not seem to be much resolution
    difference between your pictures and mine. It’s all good!

    My only issue with the SD-9 is on images like this one:
    http://www.pbase.com/image/21316315

    Take a really close look at her ear, see the yellow fringing by the hairs.

    Look at her skin tones on her face, fairly neutral, look at the shadow
    under her chin, fairly yellow.

    Since your monitor/eyes have a problem seeing yellow, try the eyedropper
    tool in Photoshop.

    Like the yellow legs on the merry go round picture, I wonder what is going
    on. White balance isn’t the problem because you can set the white balance
    on anything neutral and still there are funny yellow skin tones.

    It seems that the SD-9 can take some pretty nice pictures, but it has some
    troubling issues that make it less suitable for portraiture.
     
    MarkH, Nov 6, 2003
    #60
    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.