EXIF bug in PF Studio 10.1.0 (as spotted by Exiv2-0.19)

[Silversleeves]

Almost inevitably, when PhotoFiltre 10.1.0 saves a new JPEG file, and I use exiv2 on the command-line to read back the data, it looks something like the following.

Code : Tout sélectionner

Error: Directory Image: Next pointer is out of bounds; ignored.
Error: Directory Photo: IFD exceeds data buffer, cannot read next pointer.
pont-0-neuf.jpg       Exif.Image.Orientation                       Short       1  top, left
pont-0-neuf.jpg       Exif.Image.XResolution                       Rational    1  72
pont-0-neuf.jpg       Exif.Image.YResolution                       Rational    1  72
pont-0-neuf.jpg       Exif.Image.ResolutionUnit                    Short       1  inch
pont-0-neuf.jpg       Exif.Image.Software                          Ascii      21  PhotoFiltre Studio X
pont-0-neuf.jpg       Exif.Image.DateTime                          Ascii      20  2010:02:18 10:28:57
pont-0-neuf.jpg       Exif.Image.ExifTag                           Long        1  155
pont-0-neuf.jpg       Exif.Photo.ExifVersion                       Undefined   4  2.10
pont-0-neuf.jpg       Exif.Photo.PixelXDimension                   Short       1  576
pont-0-neuf.jpg       Exif.Photo.PixelYDimension                   Short       1  432

The first two lines also appear whenever I use the same exiv2 to write or delete EXIF or IPTC data to or from a JPEG file created in PhotoFiltre Studio 10.1.0.

Has this issue been addressed and fixed in a later version (that is, should I upgrade)? Or is this the first anyone's heard of PFS putting pointers "out of bounds" and not quite following the rules for IFD?

BZT

[Antonio]

For the moment this bug is not known. But i don't understand why when you remove the Exif data, The first two lines also appear.

Did you try to remove with other soft like XnView or other ?

[Silversleeves]

For the moment this bug is not known. But i don't understand why when you remove the Exif data, The first two lines also appear.

Did you try to remove with other soft like XnView or other ?

Antonio,

Thanks for the quick reply. Answering your questions in order;

* I think it's in how Exiv2 works. It must load the tags as they are in the file in its current state, which would give the same error as the -p (print existing) option, then after a "wipe" (deletion), the errors disappear along with the tags, which of course returns no errors.

* No, I hadn't. I've got XnView, plus a freeware EXIF tag eraser (that's all it does; I'm sure you've seen ones like it). I think a lot of the GUI image editors are quite "forgiving" when it comes to exif data that might be a "little off the standard." Even Photoshop only complains if the color scheme is wrong or unavailable -- or if the file image data is incomplete, but that's a whole other issue. I have yet to get these errors in XnView or (returning to the command line) Phil Harvey's Exiftools, and I think the latter has some of whatever the GUI apps do to ignore or suppress such errors.

I think I might try an upgrade to 10.2.1 or newer. The worst thing that can happen is that it won't even write an EXIF Software tag.

BZT

[Weyoun]

First, my Name is Jürgen, i'm a 27 year old guy from germany. I'm using the free Version PhotoFiltre now for many years. Last year i decided to purchase a license key for PhotoFiltre, and i don't have any regrets. It's a great program and i love it. Thx to everyone who are responsible for it.

And now to my question. I think i found a bug in PhotoFiltre. It has to do with the EXIF-Data. If you save a picture and mark the "Save EXIF-Data" check-mark it should save the original EXIF-Data of the camera, right? I'm not sure, but i think with the free version this was no problem. But now with PF-Studio this function doesn't work any more. The EXIF-Data is getting lost, if you save a picture that you worked on. Is this an already known bug?

Here's my System:

Windows 7, Core2Quad Q9400, 4 GiB Ram

Sorry if that problem was reported earlier. I searched the Forum and the only thread in english i found, was this one. I don't speak french.

[Silversleeves]

In PhotoFiltre Studio version 10.2.1 the problem of mislaid markers persists but seems to be limited to images one creates in PFS, as by way of a cut & paste from a browser or another image application, and saves to disk, not any that one opens from any media.

@Jürgen: PF Studio migrated this to Options:Saving, where you set your default save file format and tell the application whether or not you want the original creation/modification dates preserved on save. Clicking in the next check-box down from that (and clicking on OK) tells PF Studio to look in any file that supports them for EXIF and/or IPTC data, and if it finds any, you'll see the "Save Exif/IPTC data" check-box again in the Settings dialog box when you save the file. From there you can choose whether to do so in a specific file or not, as you probably did with the freeware PhotoFiltre.

BZT

[Silversleeves]

Just to let you folks know, the bug persists. No I have not checked for updates to PF Studio X since April, but it would make little sense for me to do so unless I was somehow made aware that an effort had been made at correcting this problem. To put the best possible face forward on this issue, I'm including output from a command-line utility other than Exiv2: ImageMagick.

Two files that showed the error in the former application just this morning had output data that looked like:

Code : Tout sélectionner

gae68-6860-032-004bath.jpg
ImageMagick info
	Image: gae68-6860-032-004bath.jpg
  Format: JPEG (Joint Photographic Experts Group JFIF format)
  Class: DirectClass
  Geometry: 492x745+0+0
  Resolution: 72x72
  Print size: 6.83333x10.3472
  Units: PixelsPerInch
  Type: TrueColor
  Endianess: Undefined
  Colorspace: RGB
  Depth: 8-bit
  Channel depth:
    red: 8-bit
    green: 8-bit
    blue: 8-bit
  Channel statistics:
    red:
      min: 5 (0.0196078)
      max: 255 (1)
      mean: 182.261 (0.714749)
      standard deviation: 60.178 (0.235992)
    green:
      min: 12 (0.0470588)
      max: 249 (0.976471)
      mean: 160.888 (0.630934)
      standard deviation: 69.0387 (0.27074)
    blue:
      min: 0 (0)
      max: 249 (0.976471)
      mean: 146.142 (0.573105)
      standard deviation: 81.7046 (0.32041)
  Rendering intent: Undefined
  Interlace: None
  Background color: white
  Border color: rgb(223,223,223)
  Matte color: grey74
  Transparent color: black
  Page geometry: 492x745+0+0
  Dispose: Undefined
  Iterations: 0
  Compression: JPEG
  Quality: 81
  Orientation: TopLeft
  Properties:
    create-date: 2010-07-31T04:24:11-04:00
    exif:DateTime: 2010:07:30 17:58:30
    exif:ExifImageLength: 745
    exif:ExifImageWidth: 492
    exif:ExifOffset: 122
    exif:ExifVersion: 0210
    exif:Orientation: 1
    exif:ResolutionUnit: 2
    exif:XResolution: 72/1
    exif:YResolution: 72/1
    jpeg:colorspace: 2
    jpeg:sampling-factor: 2x2,1x1,1x1
    modify-date: 2010-07-31T04:24:11-04:00
    signature: d52de37db904ee3d45b9622a4f837db040d04aeba73640f7df200f3b8f0304a9
  Profiles:
    Profile-8bim: 44 bytes
    Category[2,15]: PRN
    Credit[2,110]: Adult Empire sites
    Profile-exif: 170 bytes
  Artifacts:
    verbose: true
  Tainted: False
  Filesize: 42.5703kb
  Number pixels: 357.949kb

Exiv2 0.19.1 Error message:
Error: Directory Image: Next pointer is out of bounds; ignored.
Error: Directory Photo: IFD exceeds data buffer, cannot read next pointer.
Error: Directory Image: Next pointer is out of bounds; ignored.
Error: Directory Photo: IFD exceeds data buffer, cannot read next pointer.
Error: Directory Image: Next pointer is out of bounds; ignored.

And...

Code : Tout sélectionner

rva-089-012-4-pat2.jpg
ImageMagick info:
	Image: rva-089-012-4-pat2.jpg
  Format: JPEG (Joint Photographic Experts Group JFIF format)
  Class: DirectClass
  Geometry: 256x320+0+0
  Resolution: 72x72
  Print size: 3.55556x4.44444
  Units: PixelsPerInch
  Type: TrueColor
  Endianess: Undefined
  Colorspace: RGB
  Depth: 8-bit
  Channel depth:
    red: 8-bit
    green: 8-bit
    blue: 8-bit
  Channel statistics:
    red:
      min: 19 (0.0745098)
      max: 255 (1)
      mean: 229.159 (0.898662)
      standard deviation: 45.9802 (0.180314)
    green:
      min: 8 (0.0313725)
      max: 255 (1)
      mean: 216.398 (0.848619)
      standard deviation: 61.6198 (0.241646)
    blue:
      min: 2 (0.00784314)
      max: 255 (1)
      mean: 210.705 (0.826295)
      standard deviation: 69.6176 (0.27301)
  Rendering intent: Undefined
  Interlace: None
  Background color: white
  Border color: rgb(223,223,223)
  Matte color: grey74
  Transparent color: black
  Page geometry: 256x320+0+0
  Dispose: Undefined
  Iterations: 0
  Compression: JPEG
  Quality: 91
  Orientation: TopLeft
  Properties:
    create-date: 2010-07-31T04:24:15-04:00
    exif:DateTime: 2010:07:31 00:21:47
    exif:ExifImageLength: 320
    exif:ExifImageWidth: 256
    exif:ExifOffset: 122
    exif:ExifVersion: 0210
    exif:Orientation: 1
    exif:ResolutionUnit: 2
    exif:XResolution: 72/1
    exif:YResolution: 72/1
    jpeg:colorspace: 2
    jpeg:sampling-factor: 2x2,1x1,1x1
    modify-date: 2010-07-31T04:24:15-04:00
    signature: aabae78826bde09298ae1af70928f58e0409c5df8978752a71cd6b834f8003ce
  Profiles:
    Profile-exif: 170 bytes
  Artifacts:
    verbose: true
  Tainted: False
  Filesize: 12.7402kb
  Number pixels: 80kb

Exiv2 0.19.1 Error message:
Error: Directory Photo: IFD exceeds data buffer, cannot read next pointer.
Error: Directory Photo: IFD exceeds data buffer, cannot read next pointer.
Error: Directory Photo: IFD exceeds data buffer, cannot read next pointer.
Error: Directory Photo: IFD exceeds data buffer, cannot read next pointer.

The full text-file printout is available here for your perusal.

BZT

[Antonio]

possible, i use dExif for DElphi but this lib is not supported now

[Silversleeves]

Seems appropriate I'd have already posted dumps and outputs from ImageMagick.

One of its supporting tools, jhead, also throws an error when you use its "-purejpg" option ( similar to Exiv2's '-da' -- delete all metadata -- or Exiftools '-all' ). The error I saw follows:

Code : Tout sélectionner

>>jhead -purejpg bv0856-250.jpg

Nonfatal Error : 'bv0856-250.jpg' Illegal subdirectory link in Exif header

Maybe this gives someone a clue as to where/how to start re-coding the software "stamp" in the EXIF of supporting files.

BZT