Smart DeBlur

Discussion in 'Digital Photography' started by philo, Oct 26, 2012.

  1. philo

    philo Guest

    philo, Oct 26, 2012
    #1
    1. Advertising

  2. philo

    Martin Brown Guest

    On 26/10/2012 22:50, Alan Browne wrote:
    > On 2012.10.26 17:00 , philo wrote:
    >> free download
    >>
    >> https://github.com/Y-Vladimir/SmartDeblur/downloads

    >
    > Remarkable - good to get usable out of crap, but useless for a good
    > quality photo. See:
    >
    > http://www.dailymail.co.uk/sciencet...d-app-blurs-fuzzy-images-amazing-results.html


    You can do better than that with the right algorithm.

    But it isn't new though. The technique has been available since the
    early 1980's on sufficiently powerful computers. What has changed is
    that domestic PCs are now better than supercomputers of yesteryear.

    http://www.maxent.co.uk/example_1.htm

    Is an example of car number plate deblurring from 1980.

    Also appeared in the Times about the same time.

    --
    Regards,
    Martin Brown
    Martin Brown, Oct 27, 2012
    #2
    1. Advertising

  3. philo

    philo Guest

    On 10/27/2012 07:55 AM, Martin Brown wrote:
    > On 26/10/2012 22:50, Alan Browne wrote:
    >> On 2012.10.26 17:00 , philo wrote:
    >>> free download
    >>>
    >>> https://github.com/Y-Vladimir/SmartDeblur/downloads

    >>
    >> Remarkable - good to get usable out of crap, but useless for a good
    >> quality photo. See:
    >>
    >> http://www.dailymail.co.uk/sciencet...d-app-blurs-fuzzy-images-amazing-results.html
    >>

    >
    > You can do better than that with the right algorithm.
    >
    > But it isn't new though. The technique has been available since the
    > early 1980's on sufficiently powerful computers. What has changed is
    > that domestic PCs are now better than supercomputers of yesteryear.
    >
    > http://www.maxent.co.uk/example_1.htm
    >
    > Is an example of car number plate deblurring from 1980.
    >
    > Also appeared in the Times about the same time.
    >



    I have been playing with Smart DeBlur and found it very CPU intensive
    so it makes experimenting with it a bit slow. So far I have not gotten
    any decent results but will keep trying.

    About five years ago I got a few shots of Les Paul at the Iridium that
    were a bit blurred due to slow shutter speed /low light. Obviously there
    will be no opportunity for a second chance.
    philo, Oct 29, 2012
    #3
  4. philo

    Martin Brown Guest

    On 29/10/2012 11:05, philo wrote:
    > On 10/27/2012 07:55 AM, Martin Brown wrote:
    >> On 26/10/2012 22:50, Alan Browne wrote:
    >>> On 2012.10.26 17:00 , philo wrote:
    >>>> free download
    >>>>
    >>>> https://github.com/Y-Vladimir/SmartDeblur/downloads
    >>>
    >>> Remarkable - good to get usable out of crap, but useless for a good
    >>> quality photo. See:
    >>>
    >>> http://www.dailymail.co.uk/sciencet...d-app-blurs-fuzzy-images-amazing-results.html
    >>>
    >>>

    >>
    >> You can do better than that with the right algorithm.
    >>
    >> But it isn't new though. The technique has been available since the
    >> early 1980's on sufficiently powerful computers. What has changed is
    >> that domestic PCs are now better than supercomputers of yesteryear.
    >>
    >> http://www.maxent.co.uk/example_1.htm
    >>
    >> Is an example of car number plate deblurring from 1980.
    >>
    >> Also appeared in the Times about the same time.

    >
    > I have been playing with Smart DeBlur and found it very CPU intensive
    > so it makes experimenting with it a bit slow. So far I have not gotten
    > any decent results but will keep trying.


    Such algorithms are typically 50-200x slower than unsharp masking.

    > About five years ago I got a few shots of Les Paul at the Iridium that
    > were a bit blurred due to slow shutter speed /low light. Obviously there
    > will be no opportunity for a second chance.


    Look for a specular highlight near the subject you want to get in focus
    and isolate that as the trial point spread function. Also work on the
    smallest image you can get away with to practice eg 128x128.

    --
    Regards,
    Martin Brown
    Martin Brown, Oct 29, 2012
    #4
  5. philo

    philo Guest

    On 10/29/2012 09:45 AM, Martin Brown wrote:
    > On 29/10/2012 11:05, philo wrote:
    >> On 10/27/2012 07:55 AM, Martin Brown wrote:
    >>> On 26/10/2012 22:50, Alan Browne wrote:
    >>>> On 2012.10.26 17:00 , philo wrote:
    >>>>> free download
    >>>>>
    >>>>> https://github.com/Y-Vladimir/SmartDeblur/downloads
    >>>>Times about the same time.

    >>
    >> I have been playing with Smart DeBlur and found it very CPU intensive
    >> so it makes experimenting with it a bit slow. So far I have not gotten
    >> any decent results but will keep trying.

    >
    > Such algorithms are typically 50-200x slower than unsharp masking.
    >
    >> About five years ago I got a few shots of Les Paul at the Iridium that
    >> were a bit blurred due to slow shutter speed /low light. Obviously there
    >> will be no opportunity for a second chance.

    >
    > Look for a specular highlight near the subject you want to get in focus
    > and isolate that as the trial point spread function. Also work on the
    > smallest image you can get away with to practice eg 128x128.
    >



    OK will do
    philo, Oct 29, 2012
    #5
  6. philo

    Bruce Guest

    philo <" philo"@privacy.net> wrote:

    >On 10/29/2012 09:45 AM, Martin Brown wrote:
    >> On 29/10/2012 11:05, philo wrote:
    >>> On 10/27/2012 07:55 AM, Martin Brown wrote:
    >>>> On 26/10/2012 22:50, Alan Browne wrote:
    >>>>> On 2012.10.26 17:00 , philo wrote:
    >>>>>> free download
    >>>>>>
    >>>>>> https://github.com/Y-Vladimir/SmartDeblur/downloads
    >>>>>Times about the same time.
    >>>
    >>> I have been playing with Smart DeBlur and found it very CPU intensive
    >>> so it makes experimenting with it a bit slow. So far I have not gotten
    >>> any decent results but will keep trying.

    >>
    >> Such algorithms are typically 50-200x slower than unsharp masking.
    >>
    >>> About five years ago I got a few shots of Les Paul at the Iridium that
    >>> were a bit blurred due to slow shutter speed /low light. Obviously there
    >>> will be no opportunity for a second chance.

    >>
    >> Look for a specular highlight near the subject you want to get in focus
    >> and isolate that as the trial point spread function. Also work on the
    >> smallest image you can get away with to practice eg 128x128.
    >>

    >
    >
    >OK will do



    Perhaps it would have been better if philo had tested this software
    *before* recommending it on here?
    Bruce, Oct 29, 2012
    #6
  7. Martin Brown <|||newspam|||@nezumi.demon.co.uk> wrote:

    > On 26/10/2012 22:50, Alan Browne wrote:
    >> On 2012.10.26 17:00 , philo wrote:
    >>> free download
    >>>
    >>> https://github.com/Y-Vladimir/SmartDeblur/downloads

    >>
    >> Remarkable - good to get usable out of crap, but useless for a good
    >> quality photo. See:
    >>
    >> http://www.dailymail.co.uk/sciencetech/article-2221466/SmartDeblur-

    Download-app-blurs-fuzzy-images-amazing-results.html
    >
    > You can do better than that with the right algorithm.
    >
    > But it isn't new though.



    A few years ago I tried out a deconvolution program, Unshake, which
    sharpens not only out of focus stuff but other also undoes various types of
    camera shake and motion blur. You can still get it at
    http://www.zen147963.zen.co.uk/
    which also has articles explaining the principles very well.
    Gordon Freeman, Oct 30, 2012
    #7
  8. philo

    Guest Guest

    philo <" philo"@privacy.net> wrote:
    > On 10/29/2012 09:45 AM, Martin Brown wrote:
    >> On 29/10/2012 11:05, philo wrote:
    >>> On 10/27/2012 07:55 AM, Martin Brown wrote:
    >>>> On 26/10/2012 22:50, Alan Browne wrote:
    >>>>> On 2012.10.26 17:00 , philo wrote:
    >>>>>> free download
    >>>>>>
    >>>>>> https://github.com/Y-Vladimir/SmartDeblur/downloads
    >>>>> Times about the same time.
    >>>
    >>> I have been playing with Smart DeBlur and found it very CPU intensive
    >>> so it makes experimenting with it a bit slow. So far I have not gotten
    >>> any decent results but will keep trying.

    >>
    >> Such algorithms are typically 50-200x slower than unsharp masking.
    >>
    >>> About five years ago I got a few shots of Les Paul at the Iridium that
    >>> were a bit blurred due to slow shutter speed /low light. Obviously there
    >>> will be no opportunity for a second chance.

    >>
    >> Look for a specular highlight near the subject you want to get in focus
    >> and isolate that as the trial point spread function. Also work on the
    >> smallest image you can get away with to practice eg 128x128.
    >>

    >
    >
    > OK will do


    Any apps that will do the same on an iPhone?
    Guest, Oct 31, 2012
    #8
  9. philo

    PeterN Guest

    On 10/29/2012 10:45 AM, Martin Brown wrote:
    > On 29/10/2012 11:05, philo wrote:
    >> On 10/27/2012 07:55 AM, Martin Brown wrote:
    >>> On 26/10/2012 22:50, Alan Browne wrote:
    >>>> On 2012.10.26 17:00 , philo wrote:
    >>>>> free download
    >>>>>
    >>>>> https://github.com/Y-Vladimir/SmartDeblur/downloads
    >>>>
    >>>> Remarkable - good to get usable out of crap, but useless for a good
    >>>> quality photo. See:
    >>>>
    >>>> http://www.dailymail.co.uk/sciencet...d-app-blurs-fuzzy-images-amazing-results.html
    >>>>
    >>>>
    >>>>
    >>>
    >>> You can do better than that with the right algorithm.
    >>>
    >>> But it isn't new though. The technique has been available since the
    >>> early 1980's on sufficiently powerful computers. What has changed is
    >>> that domestic PCs are now better than supercomputers of yesteryear.
    >>>
    >>> http://www.maxent.co.uk/example_1.htm
    >>>
    >>> Is an example of car number plate deblurring from 1980.
    >>>
    >>> Also appeared in the Times about the same time.

    >>
    >> I have been playing with Smart DeBlur and found it very CPU intensive
    >> so it makes experimenting with it a bit slow. So far I have not gotten
    >> any decent results but will keep trying.

    >
    > Such algorithms are typically 50-200x slower than unsharp masking.
    >
    >> About five years ago I got a few shots of Les Paul at the Iridium that
    >> were a bit blurred due to slow shutter speed /low light. Obviously there
    >> will be no opportunity for a second chance.

    >
    > Look for a specular highlight near the subject you want to get in focus
    > and isolate that as the trial point spread function. Also work on the
    > smallest image you can get away with to practice eg 128x128.
    >


    I found that if you set feathering high, it runs faster. As soon as I
    get power back I would try getting the best image possible with
    feathering at factory level. Then, as a second step reduce feathering to
    zero.



    --
    Peter
    PeterN, Nov 1, 2012
    #9
    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. jr

    Problems with EAP-TLS with smart cards

    jr, Jul 26, 2004, in forum: Wireless Networking
    Replies:
    1
    Views:
    2,698
    drewbono
    Jun 23, 2006
  2. nsmurthy
    Replies:
    0
    Views:
    712
    nsmurthy
    Aug 13, 2004
  3. erha
    Replies:
    0
    Views:
    1,071
  4. Nick
    Replies:
    0
    Views:
    893
  5. wjva
    Replies:
    1
    Views:
    511
    Bryce
    Aug 20, 2003
Loading...

Share This Page