Issue 98252 - Slideshow quality insufficient for professional usage (all platforms: objects not round, Linux: not anti-aliased)
Summary: Slideshow quality insufficient for professional usage (all platforms: objects...
Status: ACCEPTED
Alias: None
Product: Impress
Classification: Application
Component: code (show other issues)
Version: OOo 3.0
Hardware: Unknown All
: P3 Trivial with 11 votes (vote)
Target Milestone: ---
Assignee: AOO issues mailing list
QA Contact:
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-01-19 18:04 UTC by norbert2
Modified: 2017-05-20 11:08 UTC (History)
10 users (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments
Linux (12.46 KB, image/png)
2009-01-19 18:06 UTC, norbert2
no flags Details
Windows (37.91 KB, image/png)
2009-01-19 18:06 UTC, norbert2
no flags Details
Powerpoint Windows (33.78 KB, image/png)
2009-01-19 18:07 UTC, norbert2
no flags Details
ODP (10.20 KB, application/vnd.oasis.opendocument.presentation)
2009-01-19 18:07 UTC, norbert2
no flags Details
PPT (72.50 KB, application/vnd.ms-powerpoint)
2009-01-19 18:07 UTC, norbert2
no flags Details
DEV300m 39 (Build 9378) in slideshow view running on kubuntu linux 810 intrepid ibex 32bit.png (12.44 KB, image/png)
2009-01-21 21:05 UTC, norbert2
no flags Details
cairo patch (1.07 KB, patch)
2010-06-16 00:55 UTC, thb
no flags Details | Diff
pixman patch (1.29 KB, patch)
2010-06-16 00:56 UTC, thb
no flags Details | Diff

Note You need to log in before you can comment on or make changes to this issue.
Description norbert2 2009-01-19 18:04:49 UTC
Hi,

please have a look at the attached three screen shots. (Make sure to view them
in native resolution, 100%.)

As you can see Impress WindowsXP looks like an anti aliased image but the object
is not round. It looks like connected lines with some white space between.

OOo Linux has the same problem (not round) AND it is not anti aliased.


Just look at the attached Powerpoint screen shot that shows how it should also
look like in Impress.
Comment 1 norbert2 2009-01-19 18:06:13 UTC
Created attachment 59497 [details]
Linux
Comment 2 norbert2 2009-01-19 18:06:34 UTC
Created attachment 59498 [details]
Windows
Comment 3 norbert2 2009-01-19 18:07:00 UTC
Created attachment 59499 [details]
Powerpoint Windows
Comment 4 norbert2 2009-01-19 18:07:26 UTC
Created attachment 59500 [details]
ODP
Comment 5 norbert2 2009-01-19 18:07:57 UTC
Created attachment 59501 [details]
PPT
Comment 6 wolframgarten 2009-01-20 07:54:01 UTC
Reproducible. But you might be interested to hear that this has been improved in
latest build DEV300m39 (CWS aw059, issue 28526).
Comment 7 norbert2 2009-01-21 06:43:52 UTC
Hi wg,

thanks for your response. But in Issue 28526 you say:

------- Additional comments from wg Mon Jan 5 07:39:29 +0000 2009 -------

@dengel: the changes that have been made are for edit view only. Anti-Aliasing
for slideshow is another area..
------- ------- ------- ------- ------- ------- ------- ------- -------

So what is correct? With THIS issue I mean the slideshow!

I re-open this issue and waiting for your response.
Comment 8 bryancole 2009-01-21 10:23:25 UTC
Is there are requirements spec for the Anti-aliasing feature? 

My view is that AA-rendering in the slideshow view is more important than for
edit view, so this feature should not be in a full release until both slideshow
and edit-view have AA-rendering. (otherwise it's a regression).
Comment 9 norbert2 2009-01-21 12:13:17 UTC
> so this feature should not be in a full release until both slideshow
> and edit-view have AA-rendering

I don't think so. please note that all OOo 3.0 application have no anti-aliasing
for edit-view. In Draw for example it is a big advantage to have drawings
displayed anti-aliased.

So anti-aliasing for edit-view only is better than no anti-aliasing.

But you may be right that the users could be irritated when he notices that the
slideshow looks worse than the edit-view.
Comment 10 wolframgarten 2009-01-21 12:57:54 UTC
@bryancole: you can find the spec at :
http://specs.openoffice.org/appwide/drawing_layer/SolidDragging.odt
Comment 11 Armin Le Grand 2009-01-21 13:27:45 UTC
AW: You are mixing up all kind of issues here. AA for edit view had a lot of
votes and needed intensive preparation. AA for Slideshow depends on the system
and the UNO API canvas renderer used, so it works on some systems and not on
others. Both graphic systems are pretty separated (by purpose).
So the first thing to do is to specify excactly at which version You see a
defect and on which system (and probably with which distribution, too).
I will set this issue to THB, he is the best to give a comment to the slideshow
AA problematic.
Comment 12 Armin Le Grand 2009-01-21 16:22:23 UTC
AW: Adding myself to CC
Comment 13 norbert2 2009-01-21 17:18:34 UTC
I'm not sure if this is only an AA problem. The picture also looks not round,
like connected lines.

Is it possible that there are more problems in slideshow mode? Perhaps the image
is rendered in wrong (not screen native) resolution and scaled on bitmap level
to the screen?

Please carefully(!) view the attached ooo300windows.png at native resolution
(100% scaling). It IS anti-aliased but nevertheless it looks angeled like
connected lines with some white space.

This strange effect can be better seen in the attached ooo300linux.png (kubuntu
8.10 intrepid ibex 32-bit) where AA is missing.
Comment 14 Armin Le Grand 2009-01-21 17:28:24 UTC
AW->norbert2: With which version is ooo300windows.png created? If it's OOo3.0 as
in field 'Version', then yes, it will not be round. For DrawingLayer, staring
from DEV300 m30, it will look round. That was the reason for the three-year deep
change of the needed internals. Please refer to CWS aw033 changes, tasks and
comments for that.

I cannot speak for Slideshow, though. THB?
Comment 15 norbert2 2009-01-21 17:34:27 UTC
Yes, it was created using the "official" OOo3.0 final release.

The windows screen shot has been created with a portable build of OOo 3.0.
Comment 16 norbert2 2009-01-21 17:42:00 UTC
And why does AA not work for Linux build? is this fixed in DEV300 m30, too?
Comment 17 Armin Le Grand 2009-01-21 17:42:50 UTC
AW->norbert2: So please get a DEV300 m39 and check there (as WG already hinted
correctly); it should be massively improved there for DrawingLayer.
Comment 18 norbert2 2009-01-21 17:44:32 UTC
... I will try DEV300 m30 Linux build and report the result here.
Comment 19 norbert2 2009-01-21 17:47:22 UTC
Where can I download DEV300 m39?
Comment 20 Armin Le Grand 2009-01-21 17:54:09 UTC
AW->norbert2: AA for linux build is activated from DEV300 m39 (as can be seen in
CWS aw059, the added tasks and comments)
AW->norbert2: Why DEV300 m30? Again: AA is activated staring from DEV300 m39
AW->norbert2: Where did You get DEV300 m30? Just use OpenOffice.org (see
http://download.openoffice.org/test/other.html). HTH!
Comment 21 norbert2 2009-01-21 21:05:40 UTC
Created attachment 59576 [details]
DEV300m 39 (Build 9378) in slideshow view running on kubuntu linux 810 intrepid ibex 32bit.png
Comment 22 norbert2 2009-01-21 21:06:02 UTC
Da haben wir wohl aneinander vorbeigeredet weil ich meinen Browser nicht
aktualisiert hatte. ;-)
-----------------

The edit view looks good in m39 but I cannot see improvements in slideshow mode.
Please have a look at the attached
"DEV300m 39 (Build 9378) in slideshow view running on kubuntu linux 810 intrepid
ibex 32bit.png"
Comment 23 norbert2 2009-01-21 21:08:09 UTC
While edit view is anti aliased and round slideshow view still looks like
connected lines and is NOT anti aliased.
Comment 24 norbert2 2009-01-21 21:10:54 UTC
Perhaps it looks a little bit better compared to OOo 3.0 final release but in
m39 the quality is still not acceptable.
Comment 25 Armin Le Grand 2009-01-22 10:27:17 UTC
AW->norbert2: Again: I did not talk about Slideshow as mentioned multiple times
now. But when You talk about Sliseshow, You HAVE to say on which system (also
said already: Different on different Systems, even on different distributions
partially). So, please, when writing something like "Perhaps it looks a little
bit better compared to OOo 3.0 final release but in m39 the quality is still not
acceptable", add on WHICH System and if You are talking about DrawingLayer OR
Sliseshow, else this statement is useless for us.
Also interesting (and new) for me: the *.png is NOT from graphics export, but
from a screenshot ?! Again here: System, Slideshow, DrawingLayer...(version is
know now at least) ?
As long as You have no more problems on DrawingLayer (i guess this, is NOT clear
from Your entries) i am out anyways. THB will comment on slideshow. To allow him
that, please add information about the System and the distribution You are
using. Thanks.
Comment 26 norbert2 2009-01-22 11:37:37 UTC
norbert2->aw:
> Again: I did not talk about Slideshow as mentioned multiple times
> now.

This issue is about slideshow! So i talk about slideshow.

> add on WHICH System

Just look at the screenshots file name. It contains so much information for a
file name. ;-)

> As long as You have no more problems on DrawingLayer

If DrawingLayer=EditView so I can confirm that it looks good and is anti aliased.

So again, this issue is about sideshow!
Comment 27 Armin Le Grand 2009-01-22 11:43:04 UTC
AW->norbert2: Okay, thanks for confirming. DrawingLayer == EditView is correct.
When DrawingLayer is out (and i am, too), this indeed is a Slideshow task from
now on, so THB should be in now.
BTW: How do You like DEV300 m39 (FullDrag, AA, better Selection in SW)...?
Comment 28 norbert2 2009-01-22 18:37:26 UTC
> BTW: How do You like DEV300 m39 (FullDrag, AA, better Selection in SW)...?

norbert2->AW: Nice. And I can confirm that these features all work in Linux build.
Comment 29 thb 2009-01-23 12:42:42 UTC
Ok, first off, switching on slideshow AA on Linux is merely a matter of enabling
cairo -> @mh @rt: any reason that's not yet switched on?

@norbert2: the "connected lines" problem is caused by emulated line stroking in
the drawing layer, which then passes down a split-up polygon to slideshow.

@aw: any chance to use VCL's LineInfo for stroking, when possible (I know it
will only catch a few common cases)?

Other than that, the slideshow now needs to make use of the new drawing layer
natively, such that we get exactly the same output as in edit view. I take this
part of the issue for one of the next 3.x feature release versions.
Comment 30 Armin Le Grand 2009-01-23 12:59:05 UTC
>@aw: any chance to use VCL's LineInfo for stroking, when possible (I know it
>will only catch a few common cases)?
AW->THB: In the MetaFile, all comment actions still exist (and are emuated
with big 	effort :-) using SvtGraphicStroke and relatives, so a quick
advancement would be possible now, too. Of course using primitives in slideshow
will be superior to that in maaaaany aspects...
Comment 31 rt 2009-01-23 13:14:29 UTC
> Ok, first off, switching on slideshow AA on Linux is merely a matter of enabling
> cairo -> @mh @rt: any reason that's not yet switched on?

I'm the wrong addressee here. I do not set such switches unless unless
developers ask me to do so.
Comment 32 thb 2009-01-23 14:27:50 UTC
@aw: ah, way cool, didn't know as cppcanvas currently does not interpret
GraphicStroke (it was largely broken before).
Comment 33 thb 2009-09-30 10:45:09 UTC
@af: Thanks for taking a look. You should at least use cairocanvas now.
Comment 34 clippka 2010-06-07 11:10:48 UTC
I'm adding ENABLE_CAIRO TRUE to the oracle build for unxlngi6 and see what happens.

cl->thb: anything else I have to look at?
Comment 35 thb 2010-06-07 14:40:42 UTC
@cl: there's nowadays a bit more, have a look into configure.in, search for
ENABLE_CAIRO there - FWICT also BUILD_PIXMAN needs to be set.
Comment 36 clippka 2010-06-08 14:51:24 UTC
After just setting ENABLE_CAIRO to yes and doing a complete build everything
works fine on a current suse. Added change to cws impress191.

This will enable anti aliased rendering for slideshow on linux platforms
(unxlngi6). I found some minor regressions when it comes to complex geometry
(thick dotted lines for circles, wrong scaling in some obscure metafiles) but I
think the overall benefit is more important than some issues we can fix later.
Comment 37 clippka 2010-06-10 14:24:42 UTC
verified in cws, back to qa

@cl->wg: this issue is only about unxlngi6.pro
Comment 38 clippka 2010-06-14 17:23:31 UTC
cl->thb: you where right, the evolution data source crashes if no pixman is
installed *sigh* adding that...
Comment 39 clippka 2010-06-15 09:39:29 UTC
Since adding pixmap is more work than I expected, I have to remove this from cws
impress191.

@cl->thb: why is no *.so build for pixmap? Do you see any problems adding a
pixmap.so to the OOo build?
Comment 40 clippka 2010-06-15 09:40:13 UTC
set to new
Comment 41 thb 2010-06-16 00:54:28 UTC
@cl: I recall it worked once - though we seem to patch this slightly, although I
fail to see the material difference - bit in a hurry now, so take it with a
grain of salt etc etc
Comment 42 thb 2010-06-16 00:55:22 UTC
Created attachment 70021 [details]
cairo patch
Comment 43 thb 2010-06-16 00:56:04 UTC
Created attachment 70022 [details]
pixman patch
Comment 44 thb 2010-06-16 00:58:41 UTC
@cl: and of course there's BUILD_PIXMAN, as I mentioned earlier. ;)
Comment 45 nomnex 2011-05-20 10:20:27 UTC
I am affected by this bug too on OOo 3.4. But what's going on now, since Oracle is not supporting OOo anymore?

Shall we open duplicate bugs on Libre Office?
Comment 46 Marcus 2017-05-20 11:08:22 UTC
Reset assigne to the default "issues@openoffice.apache.org".