[postgis-devel] [raster] Implementations of rotation and pixel size.

[Bborie Park]

Hey David,

I was wondering if you could add the hard limit of 90 degrees for
RASTER_setRotation.  In writing what will become ST_Intersects(raster,
raster), I was doing some testing on setting skew and what I realized
is that as the numbers for skew (X and Y) get larger and larger, you
get infinitely closer to plus/minus 90 degrees but will never there.
As such, setting a limit saying that the rotation must be less than 90
degrees sounds reasonable to me.

-bborie

On Thu, Aug 18, 2011 at 4:19 PM, dustymugs <dustymugs at gmail.com> wrote:
> Hey David,
>
> I've looked at the patch attached to #482 and find no issues with it.
> Everything (merging, compilation, regression testing and checking for memory
> leaks) looks good.
>
> The only thing I changed with the patch remove the diff for GNUmakefile.in,
> which is outside the raster directory.
>
> I have no recommendation to setting a hard limit of 90 degrees for
> RASTER_setRotation.  It makes sense to me but I'd hope that Pierre chimes
> in.
>
> -bborie
>
> On 08/18/2011 02:36 PM, David Zwarg wrote:
>>
>> Greetings raster devs,
>>
>> I took a day to implement the outstanding tickets that I had for some of
>> the
>> accessors related to rotated rasters:
>>
>>    - #928 - Implement ST_PixelWidth
>>     <http://trac.osgeo.org/postgis/ticket/928>
>>    - #929 - Implement ST_PixelHeight
>>    <http://trac.osgeo.org/postgis/ticket/929>
>>    - #936 - Implement ST_Rotation
>>    <http://trac.osgeo.org/postgis/ticket/936>
>>    - #482 - Implement
>> ST_SetRotation<http://trac.osgeo.org/postgis/ticket/482>
>>
>> If someone could take a moment to review the patches attached to those
>> tickets, that would be helpful. One of the things that I did not do was to
>> limit the rotation amount in ST_SetRotation. I was reading more on the
>> world
>> file, and it sounds like a raster shouldn't be rotated more than 90
>> degrees
>> -- should that be a hard limit in ST_Rotation?  If the rotation exceeds 90
>> degrees, does the affine transformation go haywire? In addition, the
>> UpperLeft position should realistically be changed. Otherwise, you could
>> technically define different rasters that cover the same area, but with
>> different UpperLeft coordinates (imagine if the geotransform put the
>> UpperLeft pixel at the lower right, and applied a 180 degree rotation).
>>
>> Thanks,
>> David
>>
>>
>>
>>
>> _______________________________________________
>> postgis-devel mailing list
>> postgis-devel at postgis.refractions.net
>> http://postgis.refractions.net/mailman/listinfo/postgis-devel
>
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-devel
>