[scilab-Users] Graphic images

[Stefan Du Rietz]

On 2008-12-14 21:52, Stefan Du Rietz wrote:
--------------------
> On 2008-12-13 18:23, Sylvestre Ledru wrote:
> --------------------
>> Le samedi 13 décembre 2008 à 18:20 +0100, Stefan Du Rietz a écrit:
>>> On 2008-12-13 18:10, Sylvestre Ledru wrote:
>>> --------------------
>>>> Le samedi 13 décembre 2008 à 18:00 +0100, Stefan Du Rietz a écrit :
>>>>> I have tried to make image files of graphic windows in Scilab 4.1.2 
>>>>> (Win XP).
>>>> Did you try with Scilab 5 ?
>>>>
>>>> Sylvestre
>>>>
>>>>
>>>>
>>> No, I cannot use it, there is something with my graphic card (ATI 
>>> Radeon HD 2600 Pro AGP), which worked with everything else until I 
>>> tried to update the drivers, so I had to cancel the update...
>> OK, maybe you will be interested by:
>> http://wiki.scilab.org/Graphical_issues_with_Scilab_5.0
>>
>> I was asking about the version 5 because the graphic export completely
>> changed in this version.
>>
>> S
>>
> Thank you for the link, which I read.
> 
> I tried again to update my graphic card drivers - much
> trouble and limited success and it takes a lot of time...
> 
> I installed Scilab 5.0.3, but when I run it (it introduces
> itself as 5.0.2!) it takes three minutes to create
> a figure the first time and all my other applications slow down as well.
> 
> And the figure is worse from the beginning than in Scilab
> 4.1.2: it is upside down, dots and lines don't fit (even
> worse than after making a GIF from Scilab 4.1.2!) although
> the dots are at least round.
> 
> So, if the figure is worse from the beginning in Scilab 5,
> it does not matter if the export is better.
> 
> /Stefan
> 
> 

I have now at last succeeded in updating my graphics card
drivers, thanks to help from ATI support.

But my Scilab 5 figures do not get better. I had a faint
feeling that my problem was due to bad precision in Scilab
5. I use datenum() to convert date vectors to time numbers 
that I can plot in figures. I plot my data with "stems", 
i.e. vertical lines, terminated with round dots, from a 
baseline. Bad precision in the x values of the different 
polylines (vertical lines and dots) will result in the 
errors I described.

After having read the mail 2008-12-17 15:29 from Sylvestre:

"OK, maybe this difference is caused by the change of
precision that we introduced between Scilab 4.X & 5.X:
http://wiki.scilab.org/Scilab_precision"

I tried to make the year zero in my date vectors before
converting them with datenum() and the problem disappeared!
But in Scilab 4.1.2 it worked anyhow.

A second issue: The dotted lines are even worse in Scilab 5
than in Scilab 4.1.2 (and in Matlab...): the dots are longer
dashes. Can anyone tell me where to fix this in the source
code (I have a working function but it is very long)? 
Strangely enough this improved with  Export - GIF...

A third issue: How can you change the position of the figure
in  Export - Postscript  (Scilab 4.1.2)? I have searched for 
info on this (but probably not enough...). In Matlab you 
have the figure property 'PaperPosition'.

Regards
Stefan