After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 90969 - Blending modes (other than normal) fail
Blending modes (other than normal) fail
Status: RESOLVED DUPLICATE of bug 86290
Product: GIMP
Classification: Other
Component: General
1.x
Other Linux
: Normal major
: 2.0
Assigned To: GIMP Bugs
Daniel Egger
Depends on:
Blocks:
 
 
Reported: 2002-08-16 15:54 UTC by airfullbete
Modified: 2002-11-02 23:39 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
The resulting GIMP image (1.03 KB, application/octet-stream)
2002-08-16 15:55 UTC, airfullbete
Details

Description airfullbete 2002-08-16 15:54:53 UTC
When I use other blending modes than normal it happens sometimes (very
often) that the result is not correct and shows several vertical lines.
Here is an example:

start the Gimp
File / New
Ok.
select the Bucket Fill
Choose Multiply mode
Click in the image.
The result is attached here.

This phenomenon does not appear when the blending is done between two
layers. It happens when using a tool or between a layer and a floating object.

AirBete.
Comment 1 airfullbete 2002-08-16 15:55:55 UTC
Created attachment 10539 [details]
The resulting GIMP image
Comment 2 Sven Neumann 2002-08-27 15:25:31 UTC
I can reproduce this. It seems to be a bug in the MMX routines since
it doesn't happen when GIMP is started with the --no-mmx command-line
option.
Comment 3 Sven Neumann 2002-08-27 15:27:44 UTC
This might be a duplicate of bug #86290.
Comment 4 Garry R. Osgood 2002-11-02 17:21:46 UTC
CVS as of November 01 2002 11:00 GMT
I can reproduce this. (mmx enabled) Furthermore, it appears only with
a background layer lacking an alpha channel. Following
'Layer->Alpha->Add Alpha Channel' the miscreant behavior goes away.
Comment 5 Sven Neumann 2002-11-02 17:30:54 UTC
This is good news as it seems to imply that the bug is not in the MMX
code but in the code that decides which routines to apply. Garry, do
you plan to investigate this further? Would be nice to get this fixed
soon.
Comment 6 Garry R. Osgood 2002-11-02 23:39:22 UTC
Confirm it is a duplicate of #86290. Both bugs exhibit same maligned
rendering condition, which arises from invocation of 
layer_multiply_mode_mmx () via combine_sub_region ( ) in paint_func.c,
which is a common control flow to both bugs. Suspect
layer_multiply_mode_mmx () is not correctly handling the 2 sources
when such has three and four byte pixel depths, respectively. Issue
disappears when both sources are at same depth; and, of course, the 
issue disappears with mmx disabled, as observed in separate ways
in both bugs. 

*** This bug has been marked as a duplicate of 86290 ***