Re: Possible new feature for next Photoshop

Discussion in 'Digital Photography' started by Trevor, Oct 11, 2011.

  1. Trevor

    Trevor Guest

    "Savageduck" <savageduck1@{REMOVESPAM}me.com> wrote in message
    news:2011101020474736716-savageduck1@REMOVESPAMmecom...
    > It seems this "Removal of blur" filter could possibly be included in a
    > future Photoshop release.
    > < http://gizmodo.com/5848371/photoshop-will-end-blurry-pics-forever >


    I love this : "Keep in mind that this won't fix your out of focus images..."
    then says "...no more ruined personal photos".

    Given that as many photo's are often ruined by being out of focus, or
    commonly focused on the wrong spot with many people not knowing how to use
    autofocus properly, I'd say the latter claim is fanciful unless they can get
    another filter to fix all those out of focus shots too!

    Trevor.
     
    Trevor, Oct 11, 2011
    #1
    1. Advertising

  2. David J Taylor, Oct 17, 2011
    #2
    1. Advertising

  3. "Savageduck" <savageduck1@{REMOVESPAM}me.com> wrote in message
    news:2011101700335750073-savageduck1@REMOVESPAMmecom...
    > On 2011-10-17 00:20:08 -0700, "David J Taylor"
    > <> said:
    >
    >>> More of the proposed "deblurring technology" came up in discussion on
    >>> G+ today and this A-B comparison was shared.
    >>>
    >>> < http://blogs.adobe.com/photoshopdotcom/files/2011/10/Plaza.png >
    >>>
    >>>
    >>> --
    >>> Regards,
    >>> Savageduck

    >>
    >> Seems far too good to be true. Was the original image RAW or JPEG?
    >>
    >> Cheers,
    >> David

    >
    > Who the Hell knows?
    >
    > --
    > Regards,
    >
    > Savageduck


    It could be important. In the blurred image there are areas where there
    appears to be insufficient information to extract the detail shown. A RAW
    image taken at low ISO would have more brightness levels, and may
    therefore show more blurred detail when the contrast range is severely
    expanded, but an 8-bit JPEG image would not have those fine brightness
    levels.

    The technique may require a very "clean" RAW input image, to work well.

    Also note that the original blur appears to be in a uniform direction
    across the image, possibly with a period without blur as well. No twist
    to the blur, and no deviation from a straight line.

    I welcome a blur reduction facility in the program, but wonder how useful
    it will be to real-world blurred images such as I might produce!

    Cheers,
    David
     
    David J Taylor, Oct 17, 2011
    #3
  4. > This was a press release without details of the process, or algorithm. I
    > would guess that given the complexity of such a correction that
    > processing a from RAW would certainly make sense. However I do not know
    > that was or was nor the case for this demonstration of the technology.


    Thanks, I didn't know it was a press release.

    > Again, I might be making a false assumption here, but it seems that this
    > level of correction would call for a RAW file to work on. I am also sure
    > that once this technology matures it will eventually find itself in the
    > consumer levels of Photoshop such as elements and/or permit this process
    > with JPEGs.


    If so, it will be most welcome!

    > As far as requiring a "clean" RAW file, I would just point to the
    > before image, and note that it is not particularly "clean".


    It seemed "clean" to me, but I didn't look at carefully.

    > Personally I shoot RAW and any blur I seem to produce is not as extreme
    > as that shown, so I could possibly benefit at some time.
    >
    > --
    > Regards,
    > Savageduck


    Certainly it was extreme. I may yet be converted to RAW!

    Cheers,
    David
     
    David J Taylor, Oct 17, 2011
    #4
  5. Trevor

    Martin Brown Guest

    On 17/10/2011 08:20, David J Taylor wrote:
    >> More of the proposed "deblurring technology" came up in discussion on
    >> G+ today and this A-B comparison was shared.
    >>
    >> < http://blogs.adobe.com/photoshopdotcom/files/2011/10/Plaza.png >
    >>
    >>
    >> --
    >> Regards,
    >> Savageduck

    >
    > Seems far too good to be true. Was the original image RAW or JPEG?
    >
    > Cheers,
    > David


    Even if it was 16 bit raw data it looks far too good to be true and most
    probably is - this looks much more like marketing's idea of what it
    might do rather than the effect of deblurring on a real image.

    Blurring is easy but the inverse is notoriously difficult - more so if
    the blurring function itself is unknown.

    The recovered image does not show any of the characteristic artefacts of
    state-of-the-art deconvolution methods (blind or otherwise).

    (examine point sources against dark areas or vice-versa)

    eg. Spotlight top left above pink & turquoise dresses
    Black dots on building in middle distance/lights on right hand wall.

    It is unkind to pixel peep like this, but extraordinary claims require
    extraordinary proof.

    --
    Regards,
    Martin Brown
     
    Martin Brown, Oct 17, 2011
    #5
  6. > Even if it was 16 bit raw data it looks far too good to be true and most
    > probably is - this looks much more like marketing's idea of what it
    > might do rather than the effect of deblurring on a real image.
    >
    > Blurring is easy but the inverse is notoriously difficult - more so if
    > the blurring function itself is unknown.
    >
    > The recovered image does not show any of the characteristic artefacts of
    > state-of-the-art deconvolution methods (blind or otherwise).
    >
    > (examine point sources against dark areas or vice-versa)
    >
    > eg. Spotlight top left above pink & turquoise dresses
    > Black dots on building in middle distance/lights on right hand wall.
    >
    > It is unkind to pixel peep like this, but extraordinary claims require
    > extraordinary proof.
    >
    > --
    > Regards,
    > Martin Brown


    Thanks for your input, Martin. My personal impression was that the
    blurred image was produced from a sharp original. It doesn't match with
    my experience at all. You may well be right about it being simply to show
    what image blurring looks like, rather than an honest example of the
    program's capabilities. Marketing half-truths.

    I think that anyone buying the software and expecting results like those
    will be rather disappointed, putting it mildly!

    Cheers,
    David
     
    David J Taylor, Oct 17, 2011
    #6
  7. Trevor

    Martin Brown Guest

    On 17/10/2011 15:33, bugbear wrote:
    > bugbear wrote:
    >> Martin Brown wrote:
    >>> On 17/10/2011 08:20, David J Taylor wrote:
    >>>>> More of the proposed "deblurring technology" came up in discussion on
    >>>>> G+ today and this A-B comparison was shared.
    >>>>>
    >>>>> < http://blogs.adobe.com/photoshopdotcom/files/2011/10/Plaza.png >
    >>>>>
    >>>>>
    >>>>> --
    >>>>> Regards,
    >>>>> Savageduck
    >>>>
    >>>> Seems far too good to be true. Was the original image RAW or JPEG?
    >>>>
    >>>> Cheers,
    >>>> David
    >>>
    >>> Even if it was 16 bit raw data it looks far too good to be true and most
    >>> probably is - this looks much more like marketing's idea of what it
    >>> might do rather than the effect of deblurring on a real image.
    >>>
    >>> Blurring is easy but the inverse is notoriously difficult - more so if
    >>> the blurring function itself is unknown.
    >>>
    >>> The recovered image does not show any of the characteristic artefacts of
    >>> state-of-the-art deconvolution methods (blind or otherwise).
    >>>
    >>> (examine point sources against dark areas or vice-versa)
    >>>
    >>> eg. Spotlight top left above pink & turquoise dresses
    >>> Black dots on building in middle distance/lights on right hand wall.
    >>>
    >>> It is unkind to pixel peep like this, but extraordinary claims require
    >>> extraordinary proof.

    >>
    >> If these were (say) samples that had been synthetically motion blurred
    >> you might gets results this good.
    >>
    >> IIRC motion blur is fully reversible.


    No. It is slightly easier to invert than some other smoother functions.
    That is not the same thing at all.
    >
    > http://research.microsoft.com/en-us/um/redmond/groups/ivm/imudeblurring/
    >
    > BugBear


    These are genuine semi-blind deconvolutions and show exactly the sort of
    artefacts that I would expect to see if you know where to look. The
    easiest one to see is on the carpark shot - note how the black and grey
    cars in the background shed a visible ringing outline onto the clean
    white walls and floor behind them. That is what you would expect to see
    when a modern deconvolution code was run against a blurred image.

    The metal sculpture also shows ringing clearly in the bright sky.

    The images on this Microsoft link are genuine enough.
    I have my doubts about the other from Photoshops press release.

    --
    Regards,
    Martin Brown
     
    Martin Brown, Oct 17, 2011
    #7
  8. Trevor

    Martin Brown Guest

    On 17/10/2011 16:52, Savageduck wrote:
    > On 2011-10-17 06:02:33 -0700, Martin Brown
    > <|||newspam|||@nezumi.demon.co.uk> said:
    >
    >> On 17/10/2011 08:20, David J Taylor wrote:
    >>>> More of the proposed "deblurring technology" came up in discussion on
    >>>> G+ today and this A-B comparison was shared.
    >>>>
    >>>> < http://blogs.adobe.com/photoshopdotcom/files/2011/10/Plaza.png >
    >>>>
    >>>>
    >>>> --
    >>>> Regards,
    >>>> Savageduck
    >>>
    >>> Seems far too good to be true. Was the original image RAW or JPEG?
    >>>
    >>> Cheers,
    >>> David

    >>
    >> Even if it was 16 bit raw data it looks far too good to be true and
    >> most probably is - this looks much more like marketing's idea of what
    >> it might do rather than the effect of deblurring on a real image.
    >>
    >> Blurring is easy but the inverse is notoriously difficult - more so if
    >> the blurring function itself is unknown.
    >>
    >> The recovered image does not show any of the characteristic artefacts
    >> of state-of-the-art deconvolution methods (blind or otherwise).
    >>
    >> (examine point sources against dark areas or vice-versa)
    >>
    >> eg. Spotlight top left above pink & turquoise dresses
    >> Black dots on building in middle distance/lights on right hand wall.
    >>
    >> It is unkind to pixel peep like this, but extraordinary claims require
    >> extraordinary proof.

    >
    > Just so the image I linked to and the comments made are not taken out of
    > context, here is what appeared in the photoshop.com blog. What cannot be
    > denied is the sensationalized promotional purpose of this release to the
    > public.
    > <
    > http://blogs.adobe.com/photoshopdotcom/2011/10/behind-all-the-buzz-deblur-sneak-peek.html


    FE, FI, FO, FUM!
    I smell the weasel words of marketeers

    Namely:
    "The before image below has a blur caused by camera shake. The after
    image shows the type of magic that can occur when the right algorithm is
    applied using Jue’s new prototype."

    Note the phrase used here is "the type of magic that *CAN* occur" (my
    emphasis) it does not say "the result of using". Colour me sceptical.

    --
    Regards,
    Martin Brown
     
    Martin Brown, Oct 17, 2011
    #8
  9. David J Taylor, Oct 17, 2011
    #9
  10. > Just so the image I linked to and the comments made are not taken out of
    > context, here is what appeared in the photoshop.com blog. What cannot be
    > denied is the sensationalized promotional purpose of this release to the
    > public.
    > <
    > http://blogs.adobe.com/photoshopdotcom/2011/10/behind-all-the-buzz-deblur-sneak-peek.html
    >>

    >
    >
    > --
    > Regards,
    > Savageduck


    Thanks for that URL, it does help put things in context. I also noticed
    the very particular use of English. Now look at the lower image pair:

    http://blogs.adobe.com/photoshopdotcom/files/2011/10/deblur-poster.png

    See the artefacts on the edges? That I am happy to believe has been
    deblurred in software, and such a tool will be very welcome, providing
    that people understand the limitations.

    Cheers,
    David
     
    David J Taylor, Oct 17, 2011
    #10
    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. Alan Browne

    Next (the other next) Gen "DVD" storage

    Alan Browne, May 27, 2005, in forum: Digital Photography
    Replies:
    13
    Views:
    493
    Iain Laskey
    May 31, 2005
  2. Annika1980

    Best Photoshop Feature You've Never Heard Of?

    Annika1980, Dec 11, 2005, in forum: Digital Photography
    Replies:
    2
    Views:
    362
    Annika1980
    Dec 12, 2005
  3. Giuen
    Replies:
    0
    Views:
    978
    Giuen
    Sep 12, 2008
  4. Bruce

    Re: Possible new feature for next Photoshop

    Bruce, Oct 11, 2011, in forum: Digital Photography
    Replies:
    157
    Views:
    2,110
    Wolfgang Weisselberg
    Jan 18, 2012
  5. PeterN

    Re: Possible new feature for next Photoshop

    PeterN, Oct 12, 2011, in forum: Digital Photography
    Replies:
    10
    Views:
    424
Loading...

Share This Page