[mapserver-dev] Re: Re:#3265 - Scale dependent layer switching error with Mercator

[relet]

Hi Brent,

Thank you for your comment. I followed your discussion on the proj4 list. 

I realize that I have implemented a workaround around the real problem,
which will become harder for the application of measuring distances. And it
seems that if we had the ground scale factor, the two applications would
have a potential for conflict when it comes to the accuracy of the ground
scale factor over tiles:


Brent Fraser wrote:
> 
>    A better approximation would be ask the projection library for the
> point 
> scale factor at the center of the current (or perhaps average the four 
> corners?), then multiply the projecton scale to get a "ground" scale. 
> 

I assume that in my case you could use the center/four corners of the entire
map extent to get a decent approximation and circumvent the issues for a
tiled service, unless you use world extents. 

I think there are other ways to approach the "disappearing layers at tile
borders" issue, such as for example to render more features than necessarily
visible, and filtering with MINFEATURESIZE instead of MINSCALEDENOM to get a
more coherent picture - that would switch from a hard line along a tile
border to a more natural "fading out" when things get squeezed for
projection reasons. Then again, other applications will require all of
certain features to be visible at a given scale level. 

All in all I agree that having access to a ground scale factor would improve
things. 

Best regards,
Thomas
-- 
View this message in context: http://osgeo-org.1803224.n2.nabble.com/Re-3265-Scale-dependent-layer-switching-error-with-Mercator-tp4966902p5034618.html
Sent from the Mapserver - Dev mailing list archive at Nabble.com.