I think we need to talk some more in depth about some basic imman targets. Imagine some peeple sending in the pictures they have taken during the last scout camp. The pictures are sent to the scout's office and are archived (put into a drawer :-).
When we have successfully finished imman, things would not change a lot. The same people will send in the same boring pictures. The difference is that those pictures are now (bulk) imported and indexed using imman which keeps a reference to the original media for later search and retrieval. But what will the people in the office search for? Having a look at Iggo's earlier post in which he suggests which image meta data fields to index, I found some fields that might be of interest in a search: For example the date when the image was taken. This could be used to arrange images on a timeline. The problem is that the most other interesting fields would not contain meaningful information (place taken, description, keywords) because those information is not added by the camera software.
We need to add the metadata before the metadata gets indexed by lucene or we need a possibility to edit the pictues metadata. The meta information must then be deleted from the index and readded afterwards. A tool for editing the metadata will not be able to change the meta data of the original pictures because they reside on read only media in most cases. But what if one wants to use image "bar" to publish it in "foo" magazine? Of course some imman search will be done. The result is a number identifying the original image. But this image does not contain the meta data that was added after the image was taken but before indexing :-(
Maybe we should keep a copy of the images that are indexed that we could later use for exporting the images with it's full meta data information?
Right now I think that I have some sort of understanding of the "add an image to the index" workflow but I am not sure what happens, if one has found an image.
Flo
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
as we talked on Wednesday here is my summary.
There is no change of the workflow needed. Somebody gets a bunch of (digital) images. The meta information of each image has to be extracted. I suggest to use following scheme:
1) The original image file (e.g. *.jpg): my_wonderful_hat.jpg
2) File which contains the iptc information has to be created: my_wonderful_hat.iptc.txt
3) File which contains the other meta information has to be created: my_wonderful_hat.txt
4) A JPEG-thumbnail e.g. for a preview has to be created: my_wonderful_hat-mini.jpg
Now we have the original image file which is leaving untouched and three sidecar files.
In my opinion the only manipulation of metainformation is needed for the iptc data file. In the first step there has to be something like a controlled and formalized vocabulary (thesaurus) for assigning keywords consistently to all scouting related images (your example). As you posted it the manipulation of the iptc data has to be done prior starting the lucene indexing. To make it clear not the metadata directly gathered out of the image file hast to be put to lucene, lucene has to read the files my_wonderful_hat.iptc.txt and my_wonderful_hat.txt!
I don't understand really yout thoughts about pblishing the images, the iptc data doesn't need to change you will always be the author of this image. Just the exploitation right could be sold (if it's that you are worry about).
You last clause is very interesting. Indeed we should provide a function like "export image with all iptc information included". This is not trivial if you store just image files in RAW-Format (like I do) or an other format which isn't capable of including iptc information. What we could do is to export the original image file with the iptc file for not-iptc-capable image formats and for images like jpeg we add the iptc-information and save this new file (in case of jpeg) as my_wonderful_hat-iptc.jpg
Iggo
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi there,
I think we need to talk some more in depth about some basic imman targets. Imagine some peeple sending in the pictures they have taken during the last scout camp. The pictures are sent to the scout's office and are archived (put into a drawer :-).
When we have successfully finished imman, things would not change a lot. The same people will send in the same boring pictures. The difference is that those pictures are now (bulk) imported and indexed using imman which keeps a reference to the original media for later search and retrieval. But what will the people in the office search for? Having a look at Iggo's earlier post in which he suggests which image meta data fields to index, I found some fields that might be of interest in a search: For example the date when the image was taken. This could be used to arrange images on a timeline. The problem is that the most other interesting fields would not contain meaningful information (place taken, description, keywords) because those information is not added by the camera software.
We need to add the metadata before the metadata gets indexed by lucene or we need a possibility to edit the pictues metadata. The meta information must then be deleted from the index and readded afterwards. A tool for editing the metadata will not be able to change the meta data of the original pictures because they reside on read only media in most cases. But what if one wants to use image "bar" to publish it in "foo" magazine? Of course some imman search will be done. The result is a number identifying the original image. But this image does not contain the meta data that was added after the image was taken but before indexing :-(
Maybe we should keep a copy of the images that are indexed that we could later use for exporting the images with it's full meta data information?
Right now I think that I have some sort of understanding of the "add an image to the index" workflow but I am not sure what happens, if one has found an image.
Flo
Hi Flo,
as we talked on Wednesday here is my summary.
There is no change of the workflow needed. Somebody gets a bunch of (digital) images. The meta information of each image has to be extracted. I suggest to use following scheme:
1) The original image file (e.g. *.jpg): my_wonderful_hat.jpg
2) File which contains the iptc information has to be created: my_wonderful_hat.iptc.txt
3) File which contains the other meta information has to be created: my_wonderful_hat.txt
4) A JPEG-thumbnail e.g. for a preview has to be created: my_wonderful_hat-mini.jpg
Now we have the original image file which is leaving untouched and three sidecar files.
In my opinion the only manipulation of metainformation is needed for the iptc data file. In the first step there has to be something like a controlled and formalized vocabulary (thesaurus) for assigning keywords consistently to all scouting related images (your example). As you posted it the manipulation of the iptc data has to be done prior starting the lucene indexing. To make it clear not the metadata directly gathered out of the image file hast to be put to lucene, lucene has to read the files my_wonderful_hat.iptc.txt and my_wonderful_hat.txt!
I don't understand really yout thoughts about pblishing the images, the iptc data doesn't need to change you will always be the author of this image. Just the exploitation right could be sold (if it's that you are worry about).
You last clause is very interesting. Indeed we should provide a function like "export image with all iptc information included". This is not trivial if you store just image files in RAW-Format (like I do) or an other format which isn't capable of including iptc information. What we could do is to export the original image file with the iptc file for not-iptc-capable image formats and for images like jpeg we add the iptc-information and save this new file (in case of jpeg) as my_wonderful_hat-iptc.jpg
Iggo