Sigma SD9 Photos Reviewed

Discussion in 'Digital Cameras' started by Frank Weston, Dec 27, 2003.

  1. Frank Weston

    Frank Weston Guest

    First let me state that my name is not George. I never have been George,
    and I never will be George.

    I made a decision earlier this year on a Canon 10D. I made this decision on
    the basis of the lenses available, and features of the camera as compared to
    similar Nikon and Sigma DSLRs. I chose on the basis of features that were
    important to me.

    For the last couple of days, I 've spent time reviewing hundreds, if not
    thousands of images posted on-line. I looked at a lot of Canon 10D, Nikon
    D100 and Sigma SD9 images. I judged these images with my eyes, viewing a
    top quality, well calibrated monitor. The argument over resolution and
    pixel count made no difference in this judgment. Here are some conclusions:

    Sigma SD9 images in general showed better resolution and depth of color, and
    appeared more vibrant to me than either Canon 10D or Nikon D100 images.
    This could be the camera, or it could be that the level of skill of the
    photographers who chose the SD9 was on average higher than those who chose
    otherwise.

    Sigma SD9 images, in many cases, appeared to have a yellowish cast,
    particularly in the highlights.

    Images shot with Sigma lenses didn't seem to suffer in comparison to those
    shot with more expensive Nikon and Canon glass. The main difference seemed
    to be the skill of the photographer.

    All, in all, this informal review I've conducted has gotten me thinking
    seriously about the new SD10, and about saving some money on lenses. I'd be
    very curious to see if any other readers on this list would draw similar
    conclusions after reviewing a good sample of Sigma vs. other images. If you
    feel you can give an unbiased look at photos and report back here, I'd love
    to hear what you have to say. Search the web, pick the photos you want, but
    look at a lot of them and tell me what you think. And, could we try to keep
    our biases out of this?

    Finally, George, I would appreciate it if you would stay the hell out of
    this discussion.
     
    Frank Weston, Dec 27, 2003
    #1
    1. Advertisements

  2. Prepare for foul language!
    Prepare for foul language!
    This is due to Auto processing and AWB, and probably old firmware. It's not
    inherent to RAW image.
    Good luck finding full sized images from the other DSLRs.
    That's no way to thank me for turning you onto these fantastic 10.3MPs
    machines. :^)

    The SD10 is a great camera, the best you can currently get, but at about
    double the cost of the 9 it could be hard to justify (though still
    significantly cheaper than the blurry, only 6MP 10D).

    The two compare this way: The 9 is ever so slightly sharper at ISO 100 and
    under some conditions it displays a little bit more depth (intangible I
    admit). Most will tell you there is no difference, I'm convinced there is,
    although it will fall into the minutia category for 90% of users, and both
    are dramatically sharper than any DSLR alternative. The SD10 has much lower
    noise at higher expicit ISOs (there are workarounds for this with the 9 up
    to ISO 800, but they shouldn't be needed). But again, both have a much
    higher resolution:noise ratio than any alternative. That said, if you buy
    faster EX lenses with the price difference, that'll be worth more than an
    ISO stop or two. The single source battery system of the 10 is more
    convenient, and more important than it may seem, plus it'll save you maybe
    $100 over the long haul. AWB of the 10 is spot on virtually every time, the
    9 may need WB adjusting in harsh lighting. Custom WB seems about the same
    and is easy to set either before or after any shot (prevailing conditions
    permitting, for the latter).

    And that's about it. The 9 may not be available when the SDX is released in
    Feb, so think fast.

    On Sigma lenses, you'll want to buy the super high value Sigma staples for
    any D/SLR. "EX" denotes the lens is part of their professional line. The
    ones with no competition on value are (that isn't to say all are cheaper,
    but most are much cheaper than similar performers)...

    *15-30mm/f3.5-4.5 EX ($550)
    *24-70mm/f2.8 EX ($340)
    24-70mm/3.5-5.6 HF ($85, exc corner to corner sharpness, but slower not EX)
    *70-200mm/f2.8 EX ($650)
    70-300mm/f4-5.6 ($150 [APO version $180], exc sharpness, but slower not EX)
    120-300mm/f2.8EX
    50-500mm/f4 EX
    300-800mm/5.6 EX
    Possibly, but still an unknown: 80-400/f4.5-5.6 OS EX

    Note the two non-EX zooms sell in a 2 lens kit for about $150 (free 24-70).

    14mm/f2.8
    24mm/f1.8
    28mm/f1.8
    50mm/f2.8
    105mm/f2.8
    300mm/f2.8
    800mm/f2.8

    *1.4X EX TC
    2.0X EX TC

    I *'d an excellent professional lens plan, this would apply to whatever DSLR
    you choose. All have 9 blades except the 8 blade 15-30, but that is such a
    wonderful lens it earns a kitchen pass. The only "flaw" is that the 70-200
    only has a 77mm front element and the others are 82mm, which may cause
    filter pain. If you can't deal with the EX cost, go with the 24-70/70-300
    lens kit for superb value, but the slower speed of that kit would warrant
    more strongly considering the SD10 so it's a catch-22. The 24-70 HF also
    has inverted bokeh, which can be no fun. Also worth considering, no two
    f-stops are created equal, Sigma's very fine EX glass is extrememly
    transmissive and this results in a third to half stop brighter image than
    most alternatives when used at the same aperture setting.

    Good luck with whatever you purchase.
     
    George Preddy, Dec 27, 2003
    #2
    1. Advertisements

  3. A couple of things that it's worth pointing out, to ensure that the test
    was really showing you what you think you saw:

    Were you looking at the images at the same final size? If you were
    examining the full-frame images downsized to fit your monitor, that's
    automatically true. But if you're looking at the images at 100% scale,
    you're using different magnifications if the images have different pixel
    counts. To be fair, you need to reduce all of the images to 3.4 MP, or
    enlarge all of them to 6 MP, or pick some other common size for all.

    Were they sharpened appropriately? The Sigma raw converter includes
    rather strong sharpening by default (you have to set sharpness to "-1"
    to turn this off) so most Sigma images you see will be strongly
    sharpened even if the text says they are straight out of the raw
    converter. The Canon and Nikon images have little or no sharpening in
    camera, deliberately, so the photographer can judge the amount of
    sharpening needed after other processing. But these images need
    sharpening to look their best - "straight from the camera" images will
    look soft compared to the sharpened Sigma images. The images on review
    sites like dpreview are *not* sharpened appropriately, so you'll have to
    sharpen them yourself to see the output of the cameras the way they were
    intended to be used.

    These are things you need to do before you can start comparing sharpness
    of the cameras themselves.

    Dave
     
    Dave Martindale, Dec 27, 2003
    #3
  4. Frank Weston

    Tony Spadaro Guest

    Comparing images on the web at whatever res the and whatever quality, is a
    chancy business at best. Look at magazine pictures. YOu can pick up a Sigma
    cheap right now as they are failing on the market. Try one out and you'll
    either find it's good enough for you or not - and you can send it back.
     
    Tony Spadaro, Dec 27, 2003
    #4
  5. Frank Weston

    Frank Weston Guest

    Dave,

    I'm comparing end results, attained by whatever means the photographer
    thought best. I really don't care what the means were....I'm interested in
    the image quality as presented, not the process required to attain it. I
    looked at a lot of images, and I'm sure all had been manipulated in some way
    to fit or look better, but that makes no difference. Most photographers
    are going to try to present their work in the best way possible. So, in
    essence, by looking at a lot of images, all from the same viewpoint of my
    monitor, I'm comparing the best to the best for one given set of
    circumstances. That seems a pretty fair comparison, far better than the
    incomplete, restricted sample comparisons done by most photo review sites.
    And, by looking at a lot of work by a lot of photographers, any bias on the
    part of one particular reviewer is pretty much eliminated.

    Please look at some photos...lots of them and come back and tell me what you
    think. I'm really curious to get a feel for the general perception of an
    unbiased group of viewers...if there is such a thing.
     
    Frank Weston, Dec 27, 2003
    #5
  6. Frank Weston

    Frank Weston Guest

    George. You really are an idiot, and you're doing neither yourself nor your
    cause any good by replying to this thread.
     
    Frank Weston, Dec 27, 2003
    #6
  7. I couldn't care less what you buy, but its a good discusion for others who
    actaully want good information.
     
    George Preddy, Dec 27, 2003
    #7
  8. Frank Weston

    J Guest

    I use a special caliration DVD called 'Video Essentials' to calibrate my
    home theater screen - which is NTSC based. How are we supposed to
    calibrate a computer monitor? Specifically, what can I use as
    reference colors? I can get my hands on pure white and pure black, but
    what about the rest? I am obviously a newbie to calibrating monitors,
    so bear with me if this is obvious to everyone!

    Frank Weston wrote:
    I looked at a lot of Canon 10D, Nikon
     
    J, Dec 27, 2003
    #8
  9. Frank Weston

    Jeff Shoaf Guest

    George, what's your source of authority? How do we know the info you
    provide is good information? You have given us nothing but quotes from
    sources with a vested interest in seeing Fovean and Sigma succeed, along
    with your opinions and a batch of images that most observers agree have
    imaging problems.

    Why should anyone reading this newsgroup give your opinion any weight?
    You tend to write as if you consider yourself an authority on digital
    imaging, but you haven't given us any reason to consider you an
    authority.

    You're always asking for links to other folks images. Where are the links
    to your images that have been sold, published, or mentioned by a
    recognized authority on digital imaging, photography, art, etc.?
     
    Jeff Shoaf, Dec 27, 2003
    #9
  10. How do you know this? This is an honest question. I'd like to hear
    your rationale for this or read the source of this information.
     
    Edward Seabass, Dec 27, 2003
    #10
  11. That seems reasonable to me since I do the same thing. Also, shots of
    resolution charts and color charts are important but do not tell the whole
    story. A large sample of the real work of lots of photographers is a
    valuable resource, but time consuming. The test shots are justified on this
    basis and that's sensible.
    Absolute non-bias is not possible. However, some are better at it than
    others. There are lots of great photos on the web and they are the progeny
    of a wide range of equipment and users. If you are interested, some of my
    stuff is at
    http://home.comcast.net/~charlesschuler/wsb/html/view.cgi-photos.html-.html

    I, too, find some of the SD9 photos to be quite nice! However, my current
    camera is a 300D.
     
    Charles Schuler, Dec 27, 2003
    #11
  12. Frank Weston wrote:

    [snip]
    [snip]

    I'm sure you've looked at the dpreview.com review. I've looked at
    the sample pics here:

    http://www.dpreview.com/reviews/sigmasd9/page20.asp

    and these are my opinions:

    The first SD9 enlargement clearly shows the shingles on the roof.
    The D60 doesn't.

    In the second enlargement, I can more easily detect the panes in the
    windows in the SD9 image.

    The third and fourth are pretty close, but again, the stones in the
    building and the panes in the windows are more easily discernible in
    the SD9 images.

    I can't really read much in either of the fifth enlargement except
    for "CHARTERS LTD."

    The sixth enlargement is kind of a kicker, I think, since the SD9
    image shows the "pickets" (or "tines," maybe) in the red fringe on
    the riverboat. The D60 image is just a blur.

    The images are pretty close in the seventh enlargement, but I think
    the SD9 is slightly better since it looks to me like there's a bulb
    in more of those sockets.

    The eighth and ninth enlargements are pretty close, but the SD9
    images are slightly sharper.

    The D60 is clearly better in the tenth enlargement, since the gold
    color is blown out in the SD9 image.

    The eleventh and twelfth enlargements are close, I'd say.

    In the images I've looked at on the web, including George's, the
    detail of some of the SD9 images is remarkable -- the resolution of
    eyelashes, for example. The SD10 supposedly fixes the color problems
    some SD9 images have.

    Confession: I shoot a 3 meg Epson PhotoPC 3100Z, which I like well
    enough. I also have a Nikon FM from circa 1983, a great camera,
    though I haven't used it in a couple of years.

    But I am interested in a DSLR. I want to see what is announced in
    February at PMA, I think it's called. Canon has said they're going
    to release 20 new digital cameras this year. I'd like to see Sigma
    announce something, too, maybe an 18 meg sensor or something. And as
    long as I'm wishing, I'd like to see another manufacturer
    (Konica-Minolta?) adopt the Foveon sensor, though I doubt it would
    be Konica-Minolta. I'd also like to be surprised by Nikon, please,
    but they seem pretty conservative and move a bit slowly.
     
    Edward Seabass, Dec 27, 2003
    #12
  13. Ok. Just avoid the review sites like dpreview, where the images are
    deliberately *not* shown at their best, rather the way they come out of
    the camera in the name of honest reporting.
    I doubt if you'd consider me unbiased, since I'm one of the people
    pointing out George's falsehoods. But my opinion is that the SD9 looks
    great for some images, and bad for some others, depending greatly on the
    subject matter.

    For myself, I'd prefer a camera that behaved more uniformly with all
    subjects. It's my judgement that shooting with a 6 MP Bayer camera and
    downsampling to 3 MP with a bit of sharpening will give consistently
    good results regardless of subject matter, while the SD9 will be hit or
    miss.

    Dave
     
    Dave Martindale, Dec 28, 2003
    #13
  14. Check back in this newsgroup 3 weeks or a month. There was someone who
    started out with a piece of software that processes RAW data from
    digital cameras under Linux, including the Sigma cameras. By disabling
    most of the Sigma-specific processing, he showed some examples of what
    the raw "RGB" data from the SD9 really looks like. Comparison with
    Sigma's SPP software output showed that even "sharpening 0" did a
    significant amount of sharpening.

    Someone else further back posted that you had to set the sharpening to
    -1 to actually get no sharpening. All of this is in the archives of the
    group that can be found at Google groups.

    Dave
     
    Dave Martindale, Dec 28, 2003
    #14
  15. True. But note the jagged line at the left edge of the grey roof area.
    I'll bet that this is really a smooth junction, as the D60 shows.
    Ok. But look at the horizontal ledge that's just below the sill of the
    main window in the image. On the D60, it looks like a single thin stone
    feature that's slightly tilted with respect to the sensor horizontal. On
    the SD9, it looks like two separate horizontal pieces with a gap in
    between. It's obvious to me that the SD9 rendition is just wrong.
    Looks like the real image detail is about the same, but the Sigma has
    more contrast (probably due to greater sharpening).
    But it shows the "tines" being irregularly spaced, with some apparently
    missing. Do you think this is what is really present on the riverboat?
    I think this is the same effect the Sigma shows in some other images
    that include rulers, where we see only 8 or 9 mm marks per cm. The SD9
    has no anti-aliasing filter and a rather small pixel fill factor, so
    whether it shows a small feature or not depends on the luck of how the
    feature aligns with a pixel centre.

    In comparison, the D60 shows a pretty uniform blur. This is the *right*
    thing for a camera to do with detail that is too small to be rendered
    accurately. At least, that's the right thing according to standard
    signal processing theory. For artistic photography, you can choose
    one or the other.
    Yes, but the bulb may be displaced left or right from the centre of the
    socket!
    The missing anti-aliasing filter in the SD9 does give images that appear
    sharper than Bayer when comparing both at their "native" resolution.
    But sometimes the detail is wrong - the riverboat "tines" are an example
    of this. The SD10's microlenses also improve the anti-aliasing to some
    extent, so the SD10 rendering of the same scene will look more like the
    D60, which is good. The SD10 has better sensitivity too, but I don't
    recall reading anything about improved colour.

    Dave
     
    Dave Martindale, Dec 28, 2003
    #15
  16. They are confusing RAW conversion with post processing. RAW converision is
    pre-processing. All cameras sharpen images (some very, very heavily) during
    the conversion from RAW CCD/CMOS data to in-camera JPEG/TIF. The SD9/10 are
    unique in that they have no in-camera sharpness setting, since all of their
    images are RAW only.

    Sigma Photo Pro, the SD9/10 RAW conversion software, gives you complete
    control over this process (unlike most cameras) and lets you sharpen, or
    blur, the RAW data during conversion to JPEG/TIF to any degree you wish.
    Think of it as the SD9/10's in-camera sharpness setting.
     
    George Preddy, Dec 28, 2003
    #16
  17. Caution on the dpreview SD9 results: they were all shot with a firmware and
    SPP version that was current when the SD9 was announced. Both were updated
    very quickly upon public availability, then again to an even greater degree
    recently. Essentially, the whole review is based on beta stuff for all
    intents and purposes, and yes, there really is a huge difference.

    For example, from page 9...
    http://img.dpreview.com/reviews/SigmaSD9/Samples/Night/IMG01299.jpg

    Now reality...
    http://www.pbase.com/image/24182797/original.jpg

    Phil has been asked to redo the entire review many times based on many
    changes to the camera, and many outright errors. Another example, the
    resolution tests (the crop series) you sited, the ones where the SD9 heavily
    outresolves the Canon 6MP, were taken with the SD9 intentionally moved
    farther away, because Phil thought crop factor was a magnifier. Its not.

    There are lots of other omissions and mistakes, some are very major, and all
    substantially handicap the SD9 (hmmmm, coincidence?). If you'd like to know
    them all, just ask. :)


     
    George Preddy, Dec 28, 2003
    #17
  18. Not talking about sharpness, but missing optical detail on Canon images.
    The best example is the construction netting around the very top of the roof
    of the tall building in the background (on the right). This is far enough
    away that Phil's intentioinal handicapping of the SD9 by moving it farther
    away from the scene has less of an effect...

    http://img.dpreview.com/reviews/SigmaSD9/Samples/Compared/outdoor/sd9_IMG01107.jpg
    http://img.dpreview.com/reviews/SigmaSD9/Samples/Compared/outdoor/d60_CRW_5303_RT8.jpg

    The D60 is blind as bat to the vertical netting detail, it only see the much
    fatter sort of horizontal elements. As are all other DSLRs Phil has tested,
    btw, including the S2 Pro in "12MP" mode.
    The roof tines show the SD9 has at least double the optical resolution.
    Much like http://www.outbackphoto.com/artofraw/raw_05/essay.html
    Color was completely reworked in the second gen Foveon chip, in the SD10.
     
    George Preddy, Dec 28, 2003
    #18
  19. The roof tines show that the SD9 sometimes portrays a feature, and
    sometimes misses it entirely. Tiny changes in position of the camera or
    the subject can make any one tine appear or disappear. You think this
    is desirable? Better to have a stable blur for detail too small to
    resolve properly.
    Several people have repeatedly pointed out that

    (a) this test did not properly adjust for the different sensor sizes of
    the Sigma and the Canon cameras, so you can't honestly compare the
    results, and

    (b) the results show only the resolution obtained with saturated colour
    test patterns, not real-world images.

    On the other hand, the *properly* conducted resolution tests on
    dpreview.com show the D60's true resolution exceeds that of the SD9,
    while the SD9 reproduces the resolution target *with the wrong content*
    at higher resolutions.

    Dave
     
    Dave Martindale, Dec 28, 2003
    #19
  20. The SD9 is clearly outresolving the D60 by about the theoretical 2.3X. The
    construction netting on the building roof on the right in the background
    also clearly shows the 6MP Canon is not competitive...

    http://img.dpreview.com/reviews/SigmaSD9/Samples/Compared/outdoor/sd9_IMG01107.jpg
    http://img.dpreview.com/reviews/SigmaSD9/Samples/Compared/outdoor/d60_CRW_5303_RT8.jpg

    Cropping isn't optical zoom, the SD9 is heavily outresoloving the Canon 1Ds
    with the same lens.
    Color imaging is all that matters.
    The D60 gets outresolved in these tests by the SD9 to the tune of 2.3 to 1
    when imaging a full color target, when imaging a B&W target the D60 only
    loses by a little. B&W targets really don't matter. This is due to the
    SD9 having 230% the number of full color sensors as the Canon 10D/D60/300D,
    not magic.
     
    George Preddy, Dec 28, 2003
    #20
    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.