Photoshop: Character Encoding of Keywords

  • 1
  • Problem
  • Updated 4 months ago
  • Solved
  • (Edited)

OS: Windows 10

Photoshop Version: CC 19.1.2

 

 

I guess there is a character encoding problem with Photoshop.

This is what I'm doing:

     -I open an existing JPG File with Photoshop which has no keywords

     -I open the panel "fileinformation" and navigate to the tab "simple"

     -I enter the text "nöice" with a german umlaut into the field "Keywords"

     -I save the changes of the file in photoshop

 

Examining the file with exiftool (or similar programs) I see that the "Keywords" added in Photoshop are

saved in the attribute "Keywords" and "Subject", but with different encodings.

Further examining those strings I get the following results for the original keyword "nöice":

Subject:

     byte sequence: 6E.C3.B6.69.63.65

     ->correct UTF8 encoding

Keywords:

     byte sequence: 6E.C3.83.C2.B6.69.63.65

     ->seems like the string was two times UTF8 encoded and is therefore malformed

 

 

When I instead create a new file in Photoshop, add the keyword "nöice" and save the file as a JPG,

everything works fine:

Subject:

     byte sequence: 6E.C3.B6.69.63.65

Keywords:

     byte sequence: 6E.C3.B6.69.63.65

 

 

Is there a way for me to fix this problem?
Photo of Christoph Trautbeck

Christoph Trautbeck

  • 3 Posts
  • 2 Reply Likes

Posted 4 months ago

  • 1
Photo of David Converse

David Converse

  • 388 Posts
  • 89 Reply Likes
As I asked in your other post, have you tested this in Bridge? You are looking for a fix, keywording there might be the answer.
Photo of Christoph Trautbeck

Christoph Trautbeck

  • 3 Posts
  • 2 Reply Likes
As requested I tested the issue in Bridge. 

The bug does not appear in Bridge.

Additionally I found out that the bug does not appear in Photoshop, when I only edit the Keywords, but not the actual image content and the image is "saved" (and not "saved as new"). 

The bug appears consistently when I either modify the image content and "save" or when I just "save as new". 
Photo of David Converse

David Converse

  • 388 Posts
  • 89 Reply Likes
I just tested this, and looked at the hex with IrfanView to be sure. "nöice" is written into the dc namespace  but not into the lr namespace at all. I had existing keywords, added in Bridge, that appear in both namespaces.
I added the keyword and made a change to the image content then saved the JPEG.

Same result looking at the raw data using File Info in both Bridge CC 2018 and PS CC 2018.
FWIW, 19.1.4 is the current release version of PS, although nothing keyword-related is listed in the release notes.
Photo of Christoph Trautbeck

Christoph Trautbeck

  • 3 Posts
  • 2 Reply Likes
I upgraded to Photoshop Version 19.1.4 and CEP Version 8. 
The bug is fixed there.

Thank you very much!