Talk:Exchangeable image file format
From Wikipedia, the free encyclopedia
While trying to reorganize the external links I couldn't quite decide what to do with the ones listed under 'Display'. I mean, most of the programs/libraries listed under 'Extraction' can also display the Exif information. Martin Geisler 23:13, 11 Dec 2004 (UTC)
- Seeing as Wikipedia is not an internet directory, we really don't need to have links to every bit of software out there that handles EXIF data. Most of the links should go, IMO. Chuck 17:19, August 24, 2005 (UTC)
-
- Agreed (again ☺). I recommend you delete Manipulation, Extraction, and Display link sections. Barefootguru 20:10, 24 August 2005 (UTC)
[edit] Wikimedia wish
Wikimedia could display automatically the EXIF data in the pages for uploaded images.
- And soon after the above note, MediaWiki started doing so :-) - David Gerard 20:12, 20 August 2007 (UTC)
[edit] Image rotation issue
"Also, be aware that many low-end Microsoft Windows programs rotate the image degrading JPEG quality. If you want to rotate the image without losing image quality look for programs like jpegtran or exiftran."
What does that have to do with EXIF? Should it be in this article at all? Chuck 21:52, August 18, 2005 (UTC)
- It has zero relevance and should be removed. Barefootguru 05:26, 19 August 2005 (UTC)
-
- Done. I also rewrote the rest of that section, which, even after deleting the quoted sentences above, was still about twice as long as it needed to be for the information it conveyed. Chuck 17:12, August 24, 2005 (UTC)
-
-
- Great. I further shortened it! BTW, watch your capitalisation of Exif. While I personally think it should be uppercased, the spec defines it as Exif. Barefootguru 20:10, 24 August 2005 (UTC)
-
[edit] Camera Makernotes
Howdy. I'm the author of a library for decoding Exif data in Java. In practice, the most complex areas of Exif surrounds the variations of non-standardised camera manufacturer makernote data. Is this something that should be documented in Wikipedia? I imagine the content would consist of lists of codes, and discussion about each manufacturer's implementations (e.g. incorrect directory offsets, alternative endianess, etc). If people agree that this sort of reference material belongs here, then please advise upon how I might add it. Some of it would be quite volumous, and I'd expect it belongs on another page. Seeking suggestions! --Drewnoakes 17:28, 16 May 2006 (UTC)
- In my opinion, some general discussion along those lines would be appropriate, but an extensive list of details about each individual manufacturer's non-standard implementation might be a bit much. Is there an external site which covers this? Perhaps a general discussion of just a paragraph or two, along with a link to the external page for more details, might be appropriate. Chuck 17:30, 17 May 2006 (UTC)
-
- IMHO a detailed reference wouldn't be bad in a separate article, the same way there're lots of lists/comparisons articles. --Outlyer 15:56, 29 July 2006 (UTC)
- I agree that a discussion of this could be useful. I would help to contribute here where I can Boardhead 15:39, 11 September 2006 (UTC) (Phil Harvey, ExifTool author)
- If we can find the documentation, detailing manufacturers' individual variations on Exif would be fantastically useful information to provide. I'm not sure it's actually all gathered in one place anywhere else. The individual variations can be infuriating, e.g. Canon's - David Gerard 20:15, 20 August 2007 (UTC)
[edit] Windows XP damages exif tags
The reference dates back to 2001. Is this assertion still valid?
Answer: Yes the assertion is still valid.
Rotating images in Windows XP still causes metadata damage. Suprisingly the Makernote seems to survive. Other JPEG Application segments are rearranged, but otherwise left intact. The following tags are deleted :
* Orientation * Special Processing (Custom Rendered) * Exposure Mode * White Balance * Auto white balance * Digital Zoom Ratio * Equivalent Focal Length In 35mm Film * Scene Capture Type * Gain Control * Contrast * Saturation * Sharpness * Subject Distance Range * Interoperability IFD
Using the "Properties Summary Tab" of Windows XP does far worse damage - to EXIF, XMP, and Photoshop IRB (including IPTC) metadata. Some metadata will remain intact, some will be modified, some moved to other places, and some deleted altogether. See this forum post for more information
- Not true; please read it again. Applying changes with either OK or Apply will damage the Exif data, but simply viewing and canceling the dialog box will not damage the Exif data. jdstroy 05:50, 13 July 2007 (UTC)
[edit] Linux Instructions
On Linux, a subset of Exif data can be seen by right clicking the file and selecting properties. Most Linux image viewers would give the full set of Exif data. Under KDE? Gnome? Maybe someone should mention the root library that allows accessing Exif data? --Mdwyer 15:13, 28 June 2006 (UTC)
- I use GNOME and it does, just tested KDE and XFCE and don't seem to show info. I'm udpating the article (and also replacing Linux with UNIX). --Outlyer 15:40, 29 July 2006 (UTC)
[edit] screenshot
is this gnome under linux? Image:Eog_EXIF.png
it looks like mac, and the caption says: Eye of Gnome Screenshot showing Exif data of an image on a Linux system
--200.55.101.214 00:12, 29 June 2006 (UTC)
- If you look closely, It's actually (confusingly) Gnome running with an OS X-lookalike theme. —Xavid 08:46, 5 July 2006 (UTC)
-
- Should we replace it by a screenshot that use a more standard theme for Gnome? I would says yes. -- hub 14:45, 9 September 2006 (UTC)
[edit] Image only?
I incidently noticed that EXIF metadatas do not only concern 'image file format used by digital cameras', while testing a small script to rename my camera files by using their EXIF metadata. While batch-working on the files contained in a directory, it attempted to work on an Excel file. Sure enough, I did a 'exiftool blabla.xls', and here came a couple of metadatas, among which a 'MIME Type : image/vnd.fpx'. Is it worth mentioning that in the article, or why not finding a pointer to a HOWTO about taking Excel pics with your average DSLR :-) --Olivier Debre 14:33, 25 August 2006 (UTC)?
- Exif is only about images. Actually only about JPEG and TIFF. Unlike XMP. But ExifTool do more that just Exif...., including XMP -- hub 14:46, 9 September 2006 (UTC)
- ExifTool is recognizing FlashPix-formatted information in your XLS file. Very interesting, I didn't realize this existed. The EXIF specification actually contains a whole section on FlashPix information, so this could actually be considered EXIF... Boardhead 15:16, 6 February 2007 (UTC)
[edit] Rotation, EXIF tags and XMP
I'd like the article to mention that Adobe's XMP is probably the 'new EXIF' and is a vast improvement in every regard. I believe it has an open license.
Also, there's considerable ambiguity about the Exif orientation tags. If you rotate an image some editors add new tags, others revise the existing tags. If there are two sets of tags some editors get confused, others don't. (GraphicConverter (Mac) complains about two tags, but Apple iPhoto seems to create them.)
- There's a number of things that XMP does poorly with regard to metadata. It is certainly not an improvement in every regard:
- It is not designed for any kind of binary metadata, hence anything large (such as large tables of settings, authentication hashes, or thumbnails) have to be encoded with something like base64
- The text format is not well suited for small images, since it takes up far more space than the equivalent EXIF metadata. This could also be a problem for JPEG images where the largest allowable segment is 64kb.
- It is still not widely distributed (especially in cameras) and is really only well supported in Adobe products.
- XML is a very complex format, and even though XMP restricts use of some XML features, it is still overly complex. This may not be a problem for large companies with deep pockets, but it is a big problem for individuals or small companies developing small applications. Also, as a result of this complexity, it is easy to generate valid XMP which can not be read by your average utility. Boardhead 13:47, 8 February 2007 (UTC)
- The orientation tag in Exif is a problem, as are the thumbnails, as often programs will try to preserve the Exif information by simply copying it to the new image. There is no clear answer to this, as in a simple program, it is much easier to copy the Exif block intact than to parse, change and encode it. In these cases there is the choice of having out of date Exif, or having no Exif info at all. --Ozhiker 11:03, 7 November 2006 (UTC)
[edit] Problems section: Possible original research
I tagged this problem as "original research" and "citation needed":
- The JPEG and standard TIFF file formats supported by the standard are both limited to 24 bit colour. Many modern cameras can capture significantly more data than this but can only make use of it in a proprietary 'raw' format, hence the Exif standard helps reduce the motivation for camera manufacturers to provide higher colour-depth capabilities.
The rewrite clarifies some details (albeit at the expense of making section longer), but misses the point. The problem wasn't use of "POV" language, but was that the statement was clearly speculation- unattributed and thus possibly original research (i.e. one person's opinion). That was why it was "POV"; not the use of language. I should have made that clearer in my comment; sorry.
I would appreciate references to some reputable voices in the photo industry backing up this speculation. Thank you.
Fourohfour 11:34, 15 November 2006 (UTC)
- The Exif standard specifically states that colour depth is always 24 bits.[1] Many modern cameras can capture significantly more data than this (e.g. the Nikon D70 captures 36 bits of colour per pixel). Since Exif/DCF files cannot represent this colour depth, many manufacturers have developed proprietary, non-compatible RAW image formats.
- This is what it looks like now. Just stick to the bare facts and leave any speculation out of the article. If there is doubt about the accuracy or verifiability of some information then be bold and remove it. Everything is still in the old revisions, you can't break anything by being bold. It is likely that no one will fix it for you anyway, especially if they need to look for hidden comments in the article. MartinDK 13:07, 26 November 2006 (UTC)
- On the contrary, the people they were aimed at (i.e. editors of that section) *will* see them; specifically someone wanting to remove the tags, or misinterpreting the reason they were put there (as happened before).
- Although I'm happy with the removal in this case, the "nothing is lost" argument is misleading. The info is still there, but not obvious to anyone not looking for it, and after some time will need reworked to go back into a changed article. Fourohfour 14:03, 27 November 2006 (UTC)
- I would agree with that point of view. However, the problem is that usually when articles are tagged nothing really happens. Very few people bother to actually fix the article. There is a huge backlog of pages tagged as one or the other thing. In some cases half the article consists of tags. That's why I think you should just be bold, especially when the original research was so blatant as in this case. MartinDK 08:04, 28 November 2006 (UTC)
- I give the original authors, or people who actually care enough, the chance to fix and/or reference these things. Someone normally removes the uncited stuff soon enough anyway, if that doesn't happen. Fourohfour 11:41, 29 November 2006 (UTC)
- I would agree with that point of view. However, the problem is that usually when articles are tagged nothing really happens. Very few people bother to actually fix the article. There is a huge backlog of pages tagged as one or the other thing. In some cases half the article consists of tags. That's why I think you should just be bold, especially when the original research was so blatant as in this case. MartinDK 08:04, 28 November 2006 (UTC)
- This is what it looks like now. Just stick to the bare facts and leave any speculation out of the article. If there is doubt about the accuracy or verifiability of some information then be bold and remove it. Everything is still in the old revisions, you can't break anything by being bold. It is likely that no one will fix it for you anyway, especially if they need to look for hidden comments in the article. MartinDK 13:07, 26 November 2006 (UTC)
[edit] exif 2.21
There's a EXIF 2.21 on the canon EOS 350D and 400D. nothing in the article about this newer version. --70.111.218.254 20:08, 25 November 2006 (UTC)
- I have a copy of a 2.2.1 draft, but I believe the official 2.2.1 version is under the control of the ISO, and hence not accessible to your average person (unless you can afford it). This is likely the reason why this version is not well distributed. Boardhead 19:58, 24 February 2007 (UTC)
-
- I had a look on http://www.iso.org and found no standards and only one document (ISO Bulletin, Feb 2003 PDF), which just mentions Exif as an influence on the JPEG 2000 metadata structure. Do you have any more detail on the ISO connection, which would mean it's not an unmaintained specification? - David Gerard 20:22, 20 August 2007 (UTC)
[edit] EXIF Capitalization
The article states that "Exif" is the official capitalization on the acronym according to the specs. However, specs for EXIF were translated to English from a Japanese source document, and the specs don't explicitly mention presentation of the acronym. I have a hard time believing that using "Exif" was an intentional departure. Googling "EXIF capitalization" turns up a bunch of open-source code that seems to be standardizing on "Exif", but I couldn't find anything else authoritative. Anyone know more about this? Oasisbob 21:12, 31 December 2006 (UTC)
- The JEITA website uses "Exif" even on Japanese pages, so it seems very likely that the Japanese version of the spec uses that capitalization too. In any case, the English translation is an official one, and it uses "Exif" all the way through, so it's clearly intentional. (I don't understand what you mean by an "intentional departure". Departure from what?) --Zundark 16:14, 6 February 2007 (UTC)
- While I believe that either Exif or EXIF is acceptable, I think the WikiPedia page should use the same form as the official Exif documentation. However, earlier today the article was changed to "EXIF" throughout. I vote to change it back. Boardhead 20:01, 24 February 2007 (UTC)
[edit] Main template added to Exif
I've added the Main template to the Exif section. Though that is not necessarily a child of this article, it is a topic that delves in more detail then this one. David Spalding (☎ ✉ ✍) 17:55, 10 August 2007 (UTC)
- You're referring to your edit of Image file formats. Dicklyon 18:16, 10 August 2007 (UTC)
[edit] jhead
I added a link to jhead (http://www.sentex.net/~mwandel/jhead/) to the Applications for Editing Exif section. However, it was removed with the comment "revert personal-page link". I'd like to put it back since it is a useful tool for editing Exif data, but wanted to discuss it first. --Ishi Gustaedr 13:36, 4 September 2007 (UTC)
- It's generally not a good idea to link to personal pages, and especially not those with software downloads. If the software is useful and credible, link to an independent page that says so. Dicklyon 15:46, 4 September 2007 (UTC)
-
- Makes sense. Would linking to either the freshmeat page at http://freshmeat.net/projects/jhead/ or the Free Software Directory page at http://directory.fsf.org/jhead.html be acceptable? I think you can list yourself in freshmeat so it's not really independent, but it does allow comments and ratings. I'm not sure what the criteria is for being listed in directory.fsf.org. (BTW, the link for ExifTool is to a personal page, although it is a personal page at a university as opposed to a hosting site.) —Preceding unsigned comment added by Ishi Gustaedr (talk • contribs) 16:55, 4 September 2007 (UTC)
-
-
- Those might be OK. Here's a site with link, description, and review. Dicklyon 17:14, 4 September 2007 (UTC)
-
[edit] GEOLOCATION SECTION
It should be noted that even with a GPS card, the geolocation information will be limited to the location of the camera (lat, long, alt), and not of the image itself. Unless the camera has some fairly accurate sensor for determining azimuth the information is not very useful to the user. Furthermore most consumers don't "see" the world in this geographic framework. —Preceding unsigned comment added by 12.110.112.30 (talk) 14:20, 12 September 2007 (UTC)
[edit] Merger proposal
Someone has proposed merging Camera filename structure into this article. This makes no sense at all, as it has nothing to do with Exif. --Zundark (talk) 10:38, 25 January 2008 (UTC)