Jpegs and Metadata

This area contains the messages from the old Yahoo gcmac group after the port.
Post Reply
Brian Jones
Posts: 0
Joined: Sat Jun 23, 2007 2:55 am

Jpegs and Metadata

Post by Brian Jones »

I have old family photo jpegs and would like to add metadata such as Caption, Description, Name(s), Place, Date. I would like to put jpegs on a website which family could search by Name, for example, and browse, print etc the images found. Has any member done that? I'd welcome help with metadata. I have GC 9.6.1 and Henke manual. Brian Jones
thorstenlemke
Posts: 0
Joined: Fri Dec 30, 2016 12:00 pm

Re: Jpegs and Metadata

Post by thorstenlemke »

Dear Brian,please view my short tutorial movie at: http://www.lemkesoft.org/help/caption.mp4 Thorsten I have old family photo jpegs and would like to add metadata such as Caption, Description, Name(s), Place, Date. I would like to put jpegs on a website which family could search by Name, for example, and browse, print etc the images found.Has any member done that? I'd welcome help with metadata. I have GC 9.6.1 and Henke manual.Brian Jones
William H. Magill
Posts: 0
Joined: Wed Jun 23, 2004 8:23 am

New development - Exorcising photographic ghosts - A flaw that affec

Post by William H. Magill »

Once the algorithm is published next month at the Computer Vision and Pattern Recognition conference, this might be an option to be included in GC, just as red-ey fixing. From the Economist: May 23rd 2015 : From the print edition: Science and technology Exorcising photographic ghosts Double take THOSE who visit the Louvre, in Paris, hoping to get a good picture of Leonardo da Vinci’s masterpiece, “Mona Lisa”, are destined for disappointment. Their photos will almost certainly be ruined by reflections from the bulletproof glass that protects the work from vandals. Indeed, any snap taken through glass risks bearing a ghostly image of the scene behind the photographer, or of the photographer himself. A polarising filter or a lens shield can ameliorate the problem, but YiChang Shih of the Massachusetts Institute of Technology (MIT) and his colleagues think they have found a way to abolish it altogether. Instead of preventing superimposed images forming when a picture is taken, Dr Shih thinks he can edit out the unwanted ghost-image after the fact. Something similar can already be done for things like the red-eye that once marred portraits taken with the assistance of a flash. But removing reflections is trickier than removing red-eye. Scanning an image for small, matched red circles, and then deleting them, is easy. A photo spoiled by reflections provides no such obvious short cuts. There is, however, one thing which Dr Shih realised that a piece of image-editing software could get hold of. This is that what a camera sees reflected from a piece of glass is not one image but two: one from the front of the pane and one from the back. If a pane is of uniform thickness, these reflections will be identical, but offset from one another by a fixed angle and distance. The photograph therefore consists of three superposed images, not two—but this makes the problem easier rather than harder to solve, for if the replicated elements of the two ghost-images can be identified, they can be subtracted from the photograph to leave the image the photographer is interested in. This is still no simple task, but some assumptions about probability, together with a bit of computing power, can help. The probability part comes from letting the computer sift through a “learning set” of many photographs, building up a statistical view of the world based on tiny patches of each picture (in this case, squares eight pixels a side). The result is a bucket of likelihoods that a particular patch of the picture under scrutiny corresponds to shapes, edges or textures found in the learning set, and thus in the real world. The computer then creates combinations of thousands of patches that form trial images, which it attempts to combine in threes (with the crucial clue that two of the images must be offset repeats) to form something which matches the original photo. If it succeeds, the non-repeated member of the trio should be what the photographer was attempting to capture. As the team describe in a paper they will present next month at the Computer Vision and Pattern Recognition conference, held just across the river from MIT, in Boston, their software can indeed separate the desired image from the reflected one. That should improve the quality of many a holiday snap. Occasionally, though, it may be the reflected image that is the desired one. Forensic-science types, the team say, are showing an interest in the technology for just this reason. Reflections might, unbeknown to the photographer, cast light on malfeasant activity. The technique is not yet perfect. In wide-angle shots, for example, the degree to which the reflections are offset may vary across a picture’s width in a way that the team’s mathematics do not yet account for. But it looks as though it is just a matter of time before Louvre-goers can, at the tap of a button, get the photograph of Lisa Gherardini that they want. T.T.F.N. William H. Magill # iMac11,3 Core i7 [2.93GHz - 8 GB 1067MHz] OS X 10.10.3 # Macmini6,1 Intel Core i5 [2.5 Ghz - 4GB 1600MHz] OS X 10.10.3 magill@icloud.com magill@mac.com whmagill@gmail.com
Brian Jones
Posts: 0
Joined: Sat Jun 23, 2007 2:55 am

Re: Jpegs and Metadata

Post by Brian Jones »

This was v useful and I have been happily captioning and keywording people photos by IPTC ever since. I have now run into a problem, no doubt my own fault. Using Browser Search to look for a person by entering a keyword (surname-firstname pair), I am now getting strange results. I am finding jpegs which don't have any of the keywords and some without the keywords are being found twice. Surnames or firstnames are in the Captions, though. If I hover over the word Search I get "Fast search files in the current folder by file name". I want to search by Keyword only but can't see how to achieve this. Does GC9 do fuzzy searches? If so, is it possible to switch to exact searches? I have Googled and searched the Henke manual. I would welcome your advice. Brian Jones > On 1 Jun 2015, at 07:43, Thorsten Lemke lemke@lemkesoft.de [gcmac] <gcmac@yahoogroups.com> wrote: > > Dear Brian, > > > please view my short tutorial movie at: http://www.lemkesoft.org/help/caption.mp4 > > Thorsten >> >> >> I have old family photo jpegs and would like to add metadata such as Caption, Description, Name(s), Place, Date. I would like to put jpegs on a website which family could search by Name, for example, and browse, print etc the images found. >> >> Has any member done that? I'd welcome help with metadata. I have GC 9.6.1 and Henke manual. >> >> Brian Jones >> > > >
thorstenlemke
Posts: 0
Joined: Fri Dec 30, 2016 12:00 pm

Re: Jpegs and Metadata

Post by thorstenlemke »

Dear Brian,please e-mail a screenshot on what you entered.Thorsten This was v useful and I have been happily captioning and keywording people photos by IPTC ever since.I have now run into a problem, no doubt my own fault.Using Browser Search to look for a person by entering a keyword (surname-firstname pair), I am now getting strange results.I am finding jpegs which don't have any of the keywords and some without the keywords are being found twice. Surnames or firstnames are in the Captions, though. If I hover over the word Search I get "Fast search files in the current folder by file name". I want to search by Keyword only but can't see how to achieve this.Does GC9 do fuzzy searches? If so, is it possible to switch to exact searches?I have Googled and searched the Henke manual.I would welcome your advice.Brian Jones
Brian Jones
Posts: 0
Joined: Sat Jun 23, 2007 2:55 am

Re: Jpegs and Metadata

Post by Brian Jones »

Thanks, Thorsten.  Not quite sure what you mean but I attach a screenshot of the result of entering the keyword "Smith Audrey" (quotes not used in Search).THe Caption on the file "5_of_us_crop.jpg" which is found once is:Children of Derek Cyril Harrill (son of Robert Collins Harrill) and Audrey Harrill nee Smith, L-R: Christine Sarah Harrill (b 1951), Lindsey Jane Harrrill (b 1950), Elaine Susan Harrill (b 1960), Elizabeth Anne Harrill (b 1949), Robert Harrill (b 1954).  (5_of_us_crop.jpg)The Keywords are:Harrill Christine; Harrill; xxxThe caption on all_the_family.jpg which is found twice is:Derek Cyril Harrill (1926-2004) and wife Audrey P H Harrill nee Smith (1926-2009), children and grand-daughter.  L-R: Background grand-daughter Claire Louise Smith (b 1973);  Standing: Arthur Jim Smith  (b 1946), Christine ' Chris ' Sarah Smith nee Harrill (b 1951), Richard Trott (b 1959), Elaine Susan Langley nee Harrill (b 1960), Lindsey Jane Phillips nee Harrill (b 1950), Audrey, Clive A Phillips, Derek.  Front: Michael J Smith (b 1948), Elizabeth Anne Smith nee Harrill (b 1949), Melanie Butler Harrill nee Eagle, Robert Harrill (b 1954).  (AlanWedding.jpg)I ran Graphic Converter First Aid just in case.The keywords are:Harrill Derek; Harrill Audrey; Harrill Christine; Smith Christine; Harrill; Smith; xxxIt's as if Search is finding "Audrey" and "Smith" separately, rather than tied together as a pair.The files are not duplicated in the original folder.I hope I haven't done something silly.  It has happened.Brian On 7 Dec 2015, at 15:45, Thorsten Lemke lemke@lemkesoft.de [gcmac] <gcmac@yahoogroups.com> wrote:Dear Brian,please e-mail a screenshot on what you entered.ThorstenThis was v useful and I have been happily captioning and keywording people photos by IPTC ever since.I have now run into a problem, no doubt my own fault.Using Browser Search to look for a person by entering a keyword (surname-firstname pair), I am now getting strange results.I am finding jpegs which don't have any of the keywords and some without the keywords are being found twice. Surnames or firstnames are in the Captions, though. If I hover over the word Search I get "Fast search files in the current folder by file name". I want to search by Keyword only but can't see how to achieve this.Does GC9 do fuzzy searches? If so, is it possible to switch to exact searches?I have Googled and searched the Henke manual.I would welcome your advice.Brian Jones
thorstenlemke
Posts: 0
Joined: Fri Dec 30, 2016 12:00 pm

Re: Jpegs and Metadata [1 Attachment]

Post by thorstenlemke »

Hello Brian,please click onto the + sign to limit the search to the keywords.Thorsten Thanks, Thorsten.  Not quite sure what you mean but I attach a screenshot of the result of entering the keyword "Smith Audrey" (quotes not used in Search).
Brian Jones
Posts: 0
Joined: Sat Jun 23, 2007 2:55 am

Re: Jpegs and Metadata

Post by Brian Jones »

Thanks, Thorsten. Until now I've not needed to go to the "+". My two-word keywords are unique to the IPTC Keywords field. It's strange that each of 10 jpegs (out of 900) are being found twice, yet the keyword "Smith Audrey", for example, is present in each only once and only in the Keywords field. I tried "+" but it made no difference. Still doubles. Something exists in these 10 jpegs which is invisible to me but not to GC9. But what? I realise that this may not be a GC9 problem but you may have advice or may have come across this before. Brian > On 8 Dec 2015, at 06:25, Thorsten Lemke lemke@lemkesoft.de [gcmac] <gcmac@yahoogroups.com> wrote: > > Hello Brian, > > > please click onto the + sign to limit the search to the keywords. > > Thorsten
thorstenlemke
Posts: 0
Joined: Fri Dec 30, 2016 12:00 pm

Re: Jpegs and Metadata

Post by thorstenlemke »

Dear Brian,please e-mail two images zipped to lemke@lemkesoft.de for testing.Thorsten Thanks, Thorsten.Until now I've not needed to go to the "+". My two-word keywords are unique to the IPTC Keywords field. It's strange that each of 10 jpegs (out of 900) are being found twice, yet the keyword "Smith Audrey", for example, is present in each only once and only in the Keywords field.I tried "+" but it made no difference. Still doubles.Something exists in these 10 jpegs which is invisible to me but not to GC9. But what?I realise that this may not be a GC9 problem but you may have advice or may have come across this before.Brian
Post Reply