Adobe Bicubic Smoother upsize vs Genuine Fractals

Discussion in 'Digital Photography' started by John Navas, Nov 27, 2007.

  1. John Navas

    John Navas Guest

    Here's a comparison of Adobe Bicubic Smoother resizing in current
    versions of Photoshop against Genuine Fractals v5. For a large amount of
    realistic detail, I started with a 400x400 crop of leaves from my
    Panasonic DMC-FZ8, and then upsampled to 200% (565x565), 300% (700x700),
    and 400% (810x810), in a single step from the original in each case,
    cropping back to 400x400 for inspection. (All images are 100% 1:1
    crops.) Final comparison image was saved at JPEG quality 11 in
    Photoshop.

    <http://img265.imageshack.us/my.php?image=resizingns7.jpg>

    200%: I see no real difference

    300%: Genuine Fractals may be a tiny bit better, but may also be a tiny
    bit harsher, and any difference is too small for me to care about

    400%: Genuine Fractals is a bit better, but the difference may not be
    visible in a print

    Bottom line: I think Adobe Bicubic Smoother is fine for most purposes,
    and wouldn't bother with Genuine Fractals except for very large
    upsizing.

    --
    Best regards,
    John Navas
    Panasonic DMC-FZ8 (and several others)
     
    John Navas, Nov 27, 2007
    #1
    1. Advertising

  2. John Navas

    John Navas Guest

    Bicubic Smoother vs Genuine Fractals vs Alienskin Blowup

    Here's an updated comparison of Adobe Bicubic Smoother upsizing (in
    current versions of Photoshop) versus Genuine Fractals 5.0.3 versus
    Alienskin Blowup 1.0.2. For a large amount of realistic detail, I
    started with a 400x400 crop of leaves from my Panasonic DMC-FZ8, and
    then upsampled to 200% (565x565), 300% (700x700), and 400% (810x810), in
    a single step from the original in each case, cropping back to 400x400
    for comparison. (All images are 100% 1:1 crops.) Final comparison image
    was saved at JPEG quality 10 in Photoshop.
    http://img137.imageshack.us/img137/412/resizingtestsxh0.jpg
    My take is that Blowup has a somewhat plastic look that increases with
    increasing resizing. Genuine Fractals seems to have the most detail in
    larger resizing, but to me it looks more like artifacting than real
    detail. I personally think Bicubic Smoother is doing the best job
    overall.
    --
    Best regards,
    John Navas
    Panasonic DMC-FZ8 (and several others)
     
    John Navas, Nov 28, 2007
    #2
    1. Advertising

  3. John Navas

    Scott W Guest

    Re: Bicubic Smoother vs Genuine Fractals vs Alienskin Blowup

    John Navas wrote:
    > Here's an updated comparison of Adobe Bicubic Smoother upsizing (in
    > current versions of Photoshop) versus Genuine Fractals 5.0.3 versus
    > Alienskin Blowup 1.0.2. For a large amount of realistic detail, I
    > started with a 400x400 crop of leaves from my Panasonic DMC-FZ8, and
    > then upsampled to 200% (565x565), 300% (700x700), and 400% (810x810), in
    > a single step from the original in each case, cropping back to 400x400
    > for comparison. (All images are 100% 1:1 crops.) Final comparison image
    > was saved at JPEG quality 10 in Photoshop.
    > http://img137.imageshack.us/img137/412/resizingtestsxh0.jpg
    > My take is that Blowup has a somewhat plastic look that increases with
    > increasing resizing. Genuine Fractals seems to have the most detail in
    > larger resizing, but to me it looks more like artifacting than real
    > detail. I personally think Bicubic Smoother is doing the best job
    > overall.


    It took me a while to realize what you are calling 400% is in area not
    linear dimensions.

    You might want to use a number of images for your source before coming
    to any hard conclusions. And image with a mix of low contrast content
    mixed with some hard edges would be interesting.

    Scott
     
    Scott W, Nov 28, 2007
    #3
  4. John Navas

    John Navas Guest

    Re: Bicubic Smoother vs Genuine Fractals vs Alienskin Blowup

    On Wed, 28 Nov 2007 09:27:17 -1000, Scott W <> wrote
    in <474dc11c$0$24288$>:

    >John Navas wrote:
    >> Here's an updated comparison of Adobe Bicubic Smoother upsizing (in
    >> current versions of Photoshop) versus Genuine Fractals 5.0.3 versus
    >> Alienskin Blowup 1.0.2. For a large amount of realistic detail, I
    >> started with a 400x400 crop of leaves from my Panasonic DMC-FZ8, and
    >> then upsampled to 200% (565x565), 300% (700x700), and 400% (810x810), in
    >> a single step from the original in each case, cropping back to 400x400
    >> for comparison. (All images are 100% 1:1 crops.) Final comparison image
    >> was saved at JPEG quality 10 in Photoshop.
    >> http://img137.imageshack.us/img137/412/resizingtestsxh0.jpg
    >> My take is that Blowup has a somewhat plastic look that increases with
    >> increasing resizing. Genuine Fractals seems to have the most detail in
    >> larger resizing, but to me it looks more like artifacting than real
    >> detail. I personally think Bicubic Smoother is doing the best job
    >> overall.

    >
    >It took me a while to realize what you are calling 400% is in area not
    >linear dimensions.


    The percentage is the amount of *image* (re)sizing, correct usage, and
    the exact methodology is in what you quoted. Using the percentage for
    both dimensions would have been misleading.

    >You might want to use a number of images for your source before coming
    >to any hard conclusions. And image with a mix of low contrast content
    >mixed with some hard edges would be interesting.


    I've actually tested lots of different images. I just don't have the
    time to present it all here. This particular image seemed most
    representative, which is why I used it. Feel free to contribute similar
    results for whatever images you think would be interesting additions.

    --
    Best regards,
    John Navas
    Panasonic DMC-FZ8 (and several others)
     
    John Navas, Nov 28, 2007
    #4
  5. John Navas

    John Navas Guest

    PhotoZoom Pro vs Bicubic Smoother vs Genuine Fractals vs Alienskin Blowup

    My comparison expanded to include PhotoZoom Pro, all at 400% (both x and
    y doubled). My own take is that Bicubic Smoother has a bit more detail
    and a bit less blurring than PhotoZoom Pro.
    Adobe Bicubic Smoother (PSE6):
    <http://img86.imageshack.us/my.php?image=4810bicubicsmoothersa9.png>
    PhotoZoom Pro (x-spline-xl):
    <http://img81.imageshack.us/my.php?image=resizingtestsxh00mw2.png>
    Genuine Fractals (5.0.3):
    <http://img81.imageshack.us/my.php?image=7810genuinefractalshy0.png>
    Alienskin Blowup (1.0.2):
    <http://img128.imageshack.us/my.php?image=a810blowupho1.png>
    --
    Best regards,
    John Navas
    Panasonic DMC-FZ8 (and several others)
    --
    Best regards,
    John Navas
    Panasonic DMC-FZ8 (and several others)
     
    John Navas, Nov 28, 2007
    #5
  6. John Navas

    Scott W Guest

    Re: Bicubic Smoother vs Genuine Fractals vs Alienskin Blowup

    John Navas wrote:
    > On Wed, 28 Nov 2007 09:27:17 -1000, Scott W <> wrote
    > in <474dc11c$0$24288$>:
    >
    >> John Navas wrote:
    >>> Here's an updated comparison of Adobe Bicubic Smoother upsizing (in
    >>> current versions of Photoshop) versus Genuine Fractals 5.0.3 versus
    >>> Alienskin Blowup 1.0.2. For a large amount of realistic detail, I
    >>> started with a 400x400 crop of leaves from my Panasonic DMC-FZ8, and
    >>> then upsampled to 200% (565x565), 300% (700x700), and 400% (810x810), in
    >>> a single step from the original in each case, cropping back to 400x400
    >>> for comparison. (All images are 100% 1:1 crops.) Final comparison image
    >>> was saved at JPEG quality 10 in Photoshop.
    >>> http://img137.imageshack.us/img137/412/resizingtestsxh0.jpg
    >>> My take is that Blowup has a somewhat plastic look that increases with
    >>> increasing resizing. Genuine Fractals seems to have the most detail in
    >>> larger resizing, but to me it looks more like artifacting than real
    >>> detail. I personally think Bicubic Smoother is doing the best job
    >>> overall.

    >> It took me a while to realize what you are calling 400% is in area not
    >> linear dimensions.

    >
    > The percentage is the amount of *image* (re)sizing, correct usage, and
    > the exact methodology is in what you quoted. Using the percentage for
    > both dimensions would have been misleading.


    When re-sizing Photoshop figures 200% is in both directions, so this can
    be taken either way.

    >> You might want to use a number of images for your source before coming
    >> to any hard conclusions. And image with a mix of low contrast content
    >> mixed with some hard edges would be interesting.

    >
    > I've actually tested lots of different images. I just don't have the
    > time to present it all here. This particular image seemed most
    > representative, which is why I used it. Feel free to contribute similar
    > results for whatever images you think would be interesting additions.


    Here is a lot of other methods with more test images.
    http://www.americaswonderlands.com/digital_photo_interpolation.htm

    Some do a bit better then others, but over all you are far better off
    starting with enough pixels.

    Some of the methods do well with sharp edges but mess up the eye pretty bad.

    Here is a rather extensive paper on Image Interpolation
    http://www.math.ucla.edu/~wittman/thesis/Poral2.pdf

    It would be interesting to test the best interpolation method against
    bicubic starting with a bit more resolution.

    My guess is that gains are small.

    Scott
     
    Scott W, Nov 28, 2007
    #6
  7. John Navas

    John Navas Guest

    Re: Bicubic Smoother vs Genuine Fractals vs Alienskin Blowup

    On Wed, 28 Nov 2007 12:50:25 -1000, Scott W <> wrote
    in <474df0ba$0$2339$>:

    >John Navas wrote:


    >> I've actually tested lots of different images. I just don't have the
    >> time to present it all here. This particular image seemed most
    >> representative, which is why I used it. Feel free to contribute similar
    >> results for whatever images you think would be interesting additions.

    >
    >Here is a lot of other methods with more test images.
    >http://www.americaswonderlands.com/digital_photo_interpolation.htm
    >Some do a bit better then others, but over all you are far better off
    >starting with enough pixels.
    >Some of the methods do well with sharp edges but mess up the eye pretty bad.
    >Here is a rather extensive paper on Image Interpolation
    >http://www.math.ucla.edu/~wittman/thesis/Poral2.pdf
    >It would be interesting to test the best interpolation method against
    >bicubic starting with a bit more resolution.
    >My guess is that gains are small.


    Yep. Here are some good additional resources:
    <http://www.general-cathexis.com/interpolation.html>
    <http://www.ronbigelow.com/articles/interpolation/interpolation.htm>

    --
    Best regards,
    John Navas
    Panasonic DMC-FZ8 (and several others)
     
    John Navas, Nov 29, 2007
    #7
  8. John Navas

    Bill Tuthill Guest

    John Navas wrote:
    > Here's a comparison of Adobe Bicubic Smoother resizing in current
    > versions of Photoshop against Genuine Fractals v5...
    >
    > http://img265.imageshack.us/my.php?image=resizingns7.jpg
    >
    > 300%: Genuine Fractals may be a tiny bit better, but may also be a tiny
    > bit harsher, and any difference is too small for me to care about
    > ...
    > Bottom line: I think Adobe Bicubic Smoother is fine for most purposes,
    > and wouldn't bother with Genuine Fractals except for very large upsizing.


    Nice work, but it's too bad you didn't compare Adobe Bicubic Smoother
    to a really good upsampling algorithm such as that in Qimage,
    or data-dependent Lanczos filters in SAR from general-cathexis.com.
     
    Bill Tuthill, Nov 29, 2007
    #8
  9. On Wed, 28 Nov 2007 22:52:53 -0800, Bill Tuthill
    <> wrote:

    >John Navas wrote:
    >> Here's a comparison of Adobe Bicubic Smoother resizing in current
    >> versions of Photoshop against Genuine Fractals v5...
    >>
    >> http://img265.imageshack.us/my.php?image=resizingns7.jpg
    >>
    >> 300%: Genuine Fractals may be a tiny bit better, but may also be a tiny
    >> bit harsher, and any difference is too small for me to care about
    >> ...
    >> Bottom line: I think Adobe Bicubic Smoother is fine for most purposes,
    >> and wouldn't bother with Genuine Fractals except for very large upsizing.

    >
    >Nice work, but it's too bad you didn't compare Adobe Bicubic Smoother
    >to a really good upsampling algorithm such as that in Qimage,
    >or data-dependent Lanczos filters in SAR from general-cathexis.com.


    Keeping in mind too that extrapolation is different than
    interpolation. Results found upsizing are not applicable to
    downsizing, and vice-versa. Bicubic usually makes a mess of
    things in downsizing. Lanczos to the rescue.
     
    trent kingston, Nov 29, 2007
    #9
  10. John Navas

    Ron Hunter Guest

    Re: Bicubic Smoother vs Genuine Fractals vs Alienskin Blowup

    Scott W wrote:
    > John Navas wrote:
    >> Here's an updated comparison of Adobe Bicubic Smoother upsizing (in
    >> current versions of Photoshop) versus Genuine Fractals 5.0.3 versus
    >> Alienskin Blowup 1.0.2. For a large amount of realistic detail, I
    >> started with a 400x400 crop of leaves from my Panasonic DMC-FZ8, and
    >> then upsampled to 200% (565x565), 300% (700x700), and 400% (810x810), in
    >> a single step from the original in each case, cropping back to 400x400
    >> for comparison. (All images are 100% 1:1 crops.) Final comparison image
    >> was saved at JPEG quality 10 in Photoshop.
    >> http://img137.imageshack.us/img137/412/resizingtestsxh0.jpg
    >> My take is that Blowup has a somewhat plastic look that increases with
    >> increasing resizing. Genuine Fractals seems to have the most detail in
    >> larger resizing, but to me it looks more like artifacting than real
    >> detail. I personally think Bicubic Smoother is doing the best job
    >> overall.

    >
    > It took me a while to realize what you are calling 400% is in area not
    > linear dimensions.
    >
    > You might want to use a number of images for your source before coming
    > to any hard conclusions. And image with a mix of low contrast content
    > mixed with some hard edges would be interesting.
    >
    > Scott


    I agree. I suspect some methods would do better with one type of image.
    For instance, Genuine Fractals would probably do better with scenes
    like trees, grass, etc. Still, they all look pretty good in the
    examples, although at 400% the GF image began to show some noise.
     
    Ron Hunter, Nov 29, 2007
    #10
  11. John Navas

    John Navas Guest

    On Thu, 29 Nov 2007 02:03:36 -0600, trent kingston
    <> wrote in
    <>:

    >On Wed, 28 Nov 2007 22:52:53 -0800, Bill Tuthill
    ><> wrote:
    >
    >>John Navas wrote:
    >>> Here's a comparison of Adobe Bicubic Smoother resizing in current
    >>> versions of Photoshop against Genuine Fractals v5...
    >>>
    >>> http://img265.imageshack.us/my.php?image=resizingns7.jpg
    >>>
    >>> 300%: Genuine Fractals may be a tiny bit better, but may also be a tiny
    >>> bit harsher, and any difference is too small for me to care about
    >>> ...
    >>> Bottom line: I think Adobe Bicubic Smoother is fine for most purposes,
    >>> and wouldn't bother with Genuine Fractals except for very large upsizing.

    >>
    >>Nice work, but it's too bad you didn't compare Adobe Bicubic Smoother
    >>to a really good upsampling algorithm such as that in Qimage,
    >>or data-dependent Lanczos filters in SAR from general-cathexis.com.


    Thanks. I was sticking to readily-available plugins for Photoshop.

    >Keeping in mind too that extrapolation is different than
    >interpolation. Results found upsizing are not applicable to
    >downsizing, and vice-versa. Bicubic usually makes a mess of
    >things in downsizing. Lanczos to the rescue.


    YMMV: I've gotten good downsizing results from Bicubic Sharper.

    --
    Best regards,
    John Navas
    Panasonic DMC-FZ8 (and several others)
     
    John Navas, Nov 29, 2007
    #11
  12. John Navas

    Scott W Guest

    John Navas wrote:
    > On Thu, 29 Nov 2007 02:03:36 -0600, trent kingston
    > <> wrote in
    > <>:
    >
    >> On Wed, 28 Nov 2007 22:52:53 -0800, Bill Tuthill
    >> <> wrote:
    >>
    >>> John Navas wrote:
    >>>> Here's a comparison of Adobe Bicubic Smoother resizing in current
    >>>> versions of Photoshop against Genuine Fractals v5...
    >>>>
    >>>> http://img265.imageshack.us/my.php?image=resizingns7.jpg
    >>>>
    >>>> 300%: Genuine Fractals may be a tiny bit better, but may also be a tiny
    >>>> bit harsher, and any difference is too small for me to care about
    >>>> ...
    >>>> Bottom line: I think Adobe Bicubic Smoother is fine for most purposes,
    >>>> and wouldn't bother with Genuine Fractals except for very large upsizing.
    >>> Nice work, but it's too bad you didn't compare Adobe Bicubic Smoother
    >>> to a really good upsampling algorithm such as that in Qimage,
    >>> or data-dependent Lanczos filters in SAR from general-cathexis.com.

    >
    > Thanks. I was sticking to readily-available plugins for Photoshop.
    >
    >> Keeping in mind too that extrapolation is different than
    >> interpolation. Results found upsizing are not applicable to
    >> downsizing, and vice-versa. Bicubic usually makes a mess of
    >> things in downsizing. Lanczos to the rescue.

    >
    > YMMV: I've gotten good downsizing results from Bicubic Sharper.
    >
     
    Scott W, Nov 29, 2007
    #12
  13. John Navas

    Scott W Guest

    John Navas wrote:
    > On Thu, 29 Nov 2007 02:03:36 -0600, trent kingston
    > <> wrote in
    > <>:
    >
    >> On Wed, 28 Nov 2007 22:52:53 -0800, Bill Tuthill
    >> <> wrote:
    >>
    >>> John Navas wrote:
    >>>> Here's a comparison of Adobe Bicubic Smoother resizing in current
    >>>> versions of Photoshop against Genuine Fractals v5...
    >>>>
    >>>> http://img265.imageshack.us/my.php?image=resizingns7.jpg
    >>>>
    >>>> 300%: Genuine Fractals may be a tiny bit better, but may also be a tiny
    >>>> bit harsher, and any difference is too small for me to care about
    >>>> ...
    >>>> Bottom line: I think Adobe Bicubic Smoother is fine for most purposes,
    >>>> and wouldn't bother with Genuine Fractals except for very large upsizing.
    >>> Nice work, but it's too bad you didn't compare Adobe Bicubic Smoother
    >>> to a really good upsampling algorithm such as that in Qimage,
    >>> or data-dependent Lanczos filters in SAR from general-cathexis.com.

    >
    > Thanks. I was sticking to readily-available plugins for Photoshop.
    >
    >> Keeping in mind too that extrapolation is different than
    >> interpolation. Results found upsizing are not applicable to
    >> downsizing, and vice-versa. Bicubic usually makes a mess of
    >> things in downsizing. Lanczos to the rescue.

    >
    > YMMV: I've gotten good downsizing results from Bicubic Sharper.


    I get good results from Bicubic as well, down sample and then use a bit
    of USM.

    I seem to have more problems with Lanczos, using IrfanView.

    Here are my test images, the 1000 x 1000 image was my starting image, I
    down sided to 345 x 345 pixels in both cases.

    http://www.pbase.com/konascott/downsampling

    If anyone want to try their hand at down sampling the 1000 x 1000 image
    free free to use it.

    Scott
     
    Scott W, Nov 29, 2007
    #13
  14. "Scott W" <> wrote:
    >
    > I seem to have more problems with Lanczos, using IrfanView.
    >
    > Here are my test images, the 1000 x 1000 image was my starting image, I
    > down sided to 345 x 345 pixels in both cases.
    >
    > http://www.pbase.com/konascott/downsampling
    >
    > If anyone want to try their hand at down sampling the 1000 x 1000 image
    > free free to use it.


    It was more fun with the maximally disoptimal bar pattern....

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Nov 30, 2007
    #14
  15. John Navas

    Scott W Guest

    David J. Littleboy wrote:
    > "Scott W" <> wrote:
    >> I seem to have more problems with Lanczos, using IrfanView.
    >>
    >> Here are my test images, the 1000 x 1000 image was my starting image, I
    >> down sided to 345 x 345 pixels in both cases.
    >>
    >> http://www.pbase.com/konascott/downsampling
    >>
    >> If anyone want to try their hand at down sampling the 1000 x 1000 image
    >> free free to use it.

    >
    > It was more fun with the maximally disoptimal bar pattern....
    >
    > David J. Littleboy
    > Tokyo, Japan
    >
    >

    Yeah, but people got hung up on that and it is not very real world.

    But anyone who wants to play with that one here it is.
    http://www.pbase.com/konascott/image/69543104/original
    The challenge is to down sample by a factor of 4, linear.
    The bar patter should disappear if there is no aliasing, but you can't
    get a sharp looking down sized image without some aliasing.

    The only method that I have run across that is way bad is nearest
    neighbor, which is bad beyond believe.

    Scott
     
    Scott W, Nov 30, 2007
    #15
  16. Scott W <> wrote:
    > John Navas wrote:
    >> On Thu, 29 Nov 2007 02:03:36 -0600, trent kingston
    >> <> wrote in
    >> <>:
    >>> On Wed, 28 Nov 2007 22:52:53 -0800, Bill Tuthill
    >>> <> wrote:


    > I get good results from Bicubic as well, down sample and then use a bit
    > of USM.


    > I seem to have more problems with Lanczos, using IrfanView.


    > Here are my test images, the 1000 x 1000 image was my starting image, I
    > down sided to 345 x 345 pixels in both cases.


    > http://www.pbase.com/konascott/downsampling


    > If anyone want to try their hand at down sampling the 1000 x 1000 image
    > free free to use it.


    When I downloaded it it turned out to be an 800x800 image, not
    1000x1000. When I used Irfanview's Lanczos resampling resizer on the
    800x800 image I could only get it to look like yours if I ticked the
    "sharpen" box. Left unticked it produced a result very hard to
    distinguish from your Bicubic resample.

    --
    Chris Malcolm DoD #205
    IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
    [http://www.dai.ed.ac.uk/homes/cam/]
     
    Chris Malcolm, Nov 30, 2007
    #16
  17. John Navas

    John Navas Guest

    On Thu, 29 Nov 2007 11:34:11 -1000, Scott W <> wrote
    in <474f3055$0$24310$>:

    >John Navas wrote:


    >> YMMV: I've gotten good downsizing results from Bicubic Sharper.

    >
    >I get good results from Bicubic as well, down sample and then use a bit
    >of USM.


    Please note I'm referring to Bicubic *Sharper* (as recommended by Adobe
    for downsizing), which works better (for me at least) than straight
    Bicubic (or Bicubic Smoother) for downsizing.

    --
    Best regards,
    John Navas
    Panasonic DMC-FZ8 (and several others)
     
    John Navas, Nov 30, 2007
    #17
  18. John Navas

    Bill Tuthill Guest

    trent kingston wrote:
    >>> Bottom line: I think Adobe Bicubic Smoother is fine for most purposes,
    >>> and wouldn't bother with Genuine Fractals except for very large upsizing.

    >>
    >> Nice work, but it's too bad you didn't compare Adobe Bicubic Smoother
    >> to a really good upsampling algorithm such as that in Qimage,
    >> or data-dependent Lanczos filters in SAR from general-cathexis.com.

    >
    > Keeping in mind too that extrapolation is different than
    > interpolation. Results found upsizing are not applicable to
    > downsizing, and vice-versa. Bicubic usually makes a mess of
    > things in downsizing. Lanczos to the rescue.


    Yes, I agree with you that simple Lanczos (as in freeware Irfanview)
    outperforms Adobe Bicubic Sharper for downsampling.
     
    Bill Tuthill, Nov 30, 2007
    #18
  19. John Navas

    Bill Tuthill Guest

    Scott W wrote:
    >
    > I seem to have more problems with Lanczos, using IrfanView.
    >
    > Here are my test images, the 1000 x 1000 image was my starting image, I
    > down sided to 345 x 345 pixels in both cases.
    >
    > http://www.pbase.com/konascott/downsampling


    So is the bicubic image with Photoshop bicubic Sharper,
    and the Lanczos image with Irfanview?

    The bricks in the chimney look better with Lanczos,
    but the wood-slat lines are smoother in bicubic.
     
    Bill Tuthill, Nov 30, 2007
    #19
  20. John Navas

    Guest

    On Nov 30, 3:57 pm, Bill Tuthill <> wrote:
    > Scott W wrote:
    >
    > > I seem to have more problems with Lanczos, using IrfanView.

    >
    > > Here are my test images, the 1000 x 1000 image was my starting image, I
    > > down sided to 345 x 345 pixels in both cases.

    >
    > >http://www.pbase.com/konascott/downsampling

    >
    > So is the bicubic image with Photoshop bicubic Sharper,
    > and the Lanczos image with Irfanview?
    >
    > The bricks in the chimney look better with Lanczos,
    > but the wood-slat lines are smoother in bicubic.


    Like Bill (and the name-changing moron) I find Lanczos gives the best
    results generally, but when you get the sort of stair-stepping that
    Scott's example shows, you may need to resample in a few steps. I got
    similar problems when I used Lanczos *in one hit*.. But when I used
    four steps, and then lightly sharpened, I got this:
    http://www.marktphoto.com/examples/scottw_example.jpg

    Note the lack of stair stepping along the top chimney edge. No, it
    doesn't *look* as sharp, but I think the 'sharpness' in Scott's
    versions is *not* real detail - to my taste the sharpening is
    overdone. I didn't try Bicubic - maybe it would give similar
    results. But I think the results will differ depending on the image
    content and on the ratio of downsampling. For that image, I would
    imagine I could probably get a little more sharpness out of the final
    result if I carefully sharpened at each step, but I can't be bothered
    right now... (O;

    It's an interesting exercise, and I think a lot of images on the web
    (including some of mine!) suffer from poor downsampling. Bart VDW has
    an interesting page on it here, but I think he oversharpens too...
    http://www.xs4all.nl/~bvdwolf/main/foto/down_sample/example1.htm

    (and no, that *isn't* the concentric rings one!)
     
    , Nov 30, 2007
    #20
    1. Advertising

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

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. speedburn

    genuine fractals ser no req'd thanks

    speedburn, Jul 23, 2003, in forum: Digital Photography
    Replies:
    3
    Views:
    421
    Webkatz
    Jul 25, 2003
  2. Click

    Genuine Fractals questions

    Click, Jul 27, 2003, in forum: Digital Photography
    Replies:
    2
    Views:
    471
    Bill Hilton
    Jul 27, 2003
  3. Re: Genuine fractals trial

    , Jul 31, 2003, in forum: Digital Photography
    Replies:
    4
    Views:
    566
    Bart van der Wolf
    Aug 2, 2003
  4. Lisa Horton

    16x20 prints from 10D, Genuine Fractals?

    Lisa Horton, Sep 3, 2003, in forum: Digital Photography
    Replies:
    5
    Views:
    723
    Bill Hilton
    Sep 4, 2003
  5. Hans-Georg Michna

    Upscaling Programs - Qimage, Genuine Fractals, pxl SmartScale

    Hans-Georg Michna, Oct 25, 2003, in forum: Digital Photography
    Replies:
    16
    Views:
    12,105
    Hans-Georg Michna
    Oct 26, 2003
Loading...

Share This Page