[mapserver-users] Rendering Challenges

Tue May 10 10:29:15 EDT 2011

Yes, offset is intended for lines, using it on polygons is abusing it.
the buffer+difference method is probably the way to go.

note to myself and steve: the OFFSET keyword should be moved into a
GEOMTRANSFORM instead of residing in a keyword of its own.

--
thomas

On Tue, May 10, 2011 at 16:21, Brent Fraser <bfraser at geoanalytic.com> wrote:
> Thomas,
>
>  Interesting. I suspect all the polygons are digitized the same was as I
> think this is a requirement for shapefiles.  The result of using:
>
>      STYLE
>        OUTLINECOLOR 255 211 127
>        OFFSET 1.5 -99
>        WIDTH 3
>      END
>
> results in http://i54.tinypic.com/25a22k4.jpg  Close, but notice the "bleed"
> outside the polygon at the sharp corners.
>
> Thanks!
>
> Best Regards,
> Brent Fraser
>
>
> On 5/10/2011 1:38 AM, thomas bonfort wrote:
>>
>> Brent,
>>
>> those are interesting questions
>>
>> On Mon, May 9, 2011 at 21:51, Brent Fraser<bfraser at geoanalytic.com>
>>  wrote:
>>>
>>>  While testing v6, I can up with the following list of things I couldn't
>>> do
>>> (see http://i56.tinypic.com/8vvprn.png for a graphic):
>>>
>>> 1. Render lines inside polygon boundary (with or without opacity)
>>
>> Along with Steve's buffer proposition, you can already more or less do
>> that with OFFSET , except you don't have the notion of inside/outside,
>> but left/right. If all your polygons are digitized in the same
>> direction (i.e. clockwise or anticlockwise) this can be done with:
>>
>> STYLE
>>   #FILL
>> END
>> STYLE
>>    OUTLINECOLOR r g b
>>    OFFSET 1.5 -99
>>    WIDTH 3
>>    OPACITY 50
>> END
>>
>>
>>
>>> 2. Place polygon annotation on/inside boundary line
>>> (http://trac.osgeo.org/mapserver/ticket/1592 point 8.)
>>
>> not possible in a single step, but by using the polygon layer a second
>> time in a TYPE LINE layer for labelling should get you close.
>>
>>> 3. Rotate anno if required to fit inside a polygon
>>
>> would be a nice addition, ANGLE AUTO for polygon labels. We should
>> probably give some way to define if the label should always follow the
>> longest polygon diagonal, or if it should be first placed horizontally
>> and fall back to automatic angling only if the horizontal one does not
>> fit.
>>
>>> 4. Place polygon annotation outside polygon if anno doesn't fit inside
>>
>> The fine tuning on how to place the label text compared to the feature
>> would require new mapfile keywords, but that would seem possible to
>> implement. I'm not sure these kind of automatic adjustments would
>> really enhance the rendered quality though.
>>
>>> 5. Place line anno above (or below) when using ANGLE FOLLOW
>>
>> This is something I would like to add to. There's a paper detailing
>> the underlying algorithm here:
>> http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.24.6970
>>
>>> 6. Rotate vector symbols to follow lines (a font might work but is a pain
>>> to
>>> create)
>>
>> how is this different than what we already have with a negative GAP?
>> On a side note, vector symbols are currently rather ill-defined in
>> mapserver and would need to be tightened up before being able to use
>> them for precise symbol placement like here.
>>
>>> 7. User specified character spacing
>>
>> SIZE AUTO for labels in polygon layers would be a nice addition. We'd
>> have to be able to define how the size of the font varies in
>> accordance with the spacing that is applied between characters.
>> Before going down that road, we shall have to implement the
>> refactoring proposed in http://trac.osgeo.org/mapserver/ticket/3611 so
>> as not to clutter the glyph rendering functions of each renderer.
>>
>>
>>> 8. Erase lines (but not other objects) under annotation
>>
>> This seems rather involved. It would be difficult to keep the ordering
>> of layers, as the modified feature could only be drawn during the
>> labelcache phase, i.e. after all other features in the map.
>>
>>>  My goal was to use the published shp files
>>>
>>> (http://ftp2.cits.rncan.gc.ca/pub/canvec/50k_shp/092/h/canvec_092h06_shp.zip)
>>> without modification and see how close I could get to the published PDF
>>>
>>> (http://ftp2.cits.rncan.gc.ca/pub/cantopo/50k_geopdf/092/h/cantopo_092h06_geopdf.zip).
>>>
>>>  Any suggested rendering techniques are appreciated.  I'll file tickets
>>> for
>>> those problems deemed worthy.
>>
>> So, most of these enhancements seem possible, some were already on my
>> laundry-list of rendering enhancements I'd like to tackle. Creating
>> enhancement tickets for those features is a first step, and funding
>> them is the second ;)
>>
>> regards,
>>
>> thomas
>>
>>> --
>>> Best Regards,
>>> Brent Fraser
>>>
>>>
>>> _______________________________________________
>>> mapserver-users mailing list
>>> mapserver-users at lists.osgeo.org
>>> http://lists.osgeo.org/mailman/listinfo/mapserver-users
>>>
>
>