[mapserver-dev] Precision about geotransform

[Lime, Steve D (DNR)]

Can you make your test case available for review? -Steve

From: mapserver-dev-bounces at lists.osgeo.org [mailto:mapserver-dev-bounces at lists.osgeo.org] On Behalf Of Mohamed Saâd HESSANE
Sent: Thursday, April 21, 2011 2:05 PM
To: thomas bonfort
Cc: mapserver-dev at lists.osgeo.org
Subject: Re: [mapserver-dev] Precision about geotransform

That exactly what I guessed. But it isn't true. If it was, a rotated map at 45°, is represented by an extent having de double of the initial size extent. A 600x600 (360000 pixels) rotated image correspond to a 848x848 (~720000) normal image. But the rotated image take 0.21s, and the normal image take 0.32s. The difference is huge.
So I read the part of the source code which is responsible of rotation. There's nothing about this. And i don't find the solution to this (I don't undestund the rotation algorithm too).

2011/4/21 thomas bonfort <thomas.bonfort at gmail.com<mailto:thomas.bonfort at gmail.com>>
Mohamed,
I'm guessing that as the underlying data drivers only accept
rectangular extents oriented northwards, they have to return more data
if there is a rotation implied. And mapserver also has to filter the
extra features out at some point, even if they do not fall on the
final map image. This could explain the overhead.

--
thomas

2011/4/21 Mohamed Saâd HESSANE <saad.hessane at gmail.com<mailto:saad.hessane at gmail.com>>:
> I don't think so. If it was, juste when a cos or a sin vanish.
> My first thought was that mapserver have to calculate more pixels in a
> rotated map than a normale map. But when i read the source code i found that
> it's the projection which rotate, not the map. So why this discimination?
> The same projection
>
>
> 2011/4/21 Stephen Woodbridge <woodbri at swoodbridge.com<mailto:woodbri at swoodbridge.com>>
>>
>> On 4/21/2011 10:32 AM, Mohamed Saād HESSANE wrote:
>>>
>>> Hy list,
>>> I have a question. I'm doing a benchmarking test of a rotated map, and i
>>> get the result in the attached file.
>>> My question is why it take more time to draw a map rotated at 45° than a
>>> map rotated at 90° or 183° ???
>>> Thank you !
>>
>> I'm only guessing here but could it be that the math required to compute
>> cos and sin is more costly in cpu cycles at those angles.
>>
>> Regardless, this is an interesting question and I love the graphic, nice
>> job on that.
>>
>> -Steve W
>> _______________________________________________
>> mapserver-dev mailing list
>> mapserver-dev at lists.osgeo.org<mailto:mapserver-dev at lists.osgeo.org>
>> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
>
>
> _______________________________________________
> mapserver-dev mailing list
> mapserver-dev at lists.osgeo.org<mailto:mapserver-dev at lists.osgeo.org>
> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/mapserver-dev/attachments/20110421/65202ea3/attachment-0001.html