Hi, I made a very short test with Mapnik 2.2.0 demo maps. Mapnik can create png images either with AGG or with Cairo and I compared the 256 colour demo maps obtained by running the Mapnik c++ version of "rundemo" test program. Image sizes were
AGG 112 kb, went down to 104 kb by optimizing with Irfan view PNGOUT plugin Cairo 105 kb, went down to 97 kb by optimizing with Irfan view PNGOUT plugin Perhaps it would be worth having a try with Cairo 8-bit png for small output size if we can use it with Mapserver. -Jukka Rahkonen- ________________________________ Lähettäjä: [email protected] [[email protected]] käyttäjän Richard Greenwood [[email protected]] puolesta Lähetetty: 7. kesäkuuta 2013 23:33 Vastaanottaja: mapserver Aihe: [mapserver-users] Fwd: RFC99: Remove GD support in 7.0 Sorry - posting back to list... ---------- Forwarded message ---------- From: Richard Greenwood <[email protected]<mailto:[email protected]>> Date: Fri, Jun 7, 2013 at 2:32 PM Subject: Re: [mapserver-users] RFC99: Remove GD support in 7.0 To: thomas bonfort <[email protected]<mailto:[email protected]>> On Fri, Jun 7, 2013 at 2:00 PM, thomas bonfort <[email protected]<mailto:[email protected]>> wrote: (posting back on list) On 7 June 2013 21:24, Richard Greenwood <[email protected]<mailto:[email protected]>> wrote: I have continued to use GIF rather than 8bit PNG because the GIF is smaller. Often by as much as 15%. Is that expected? Is there anyway to get PNG down to GIF size? I certainly understand the rationale for eliminating GD but I am discouraged by the prospect of larger image sizes. I don't think that the gif vs. png format is relevant concerning file size when encoding the same pixel data. However, the quantity of information to encode in an antialiased image rendering is higher than in a non antialiased one, thus the higher file size when using agg/png8 vs. gd/gif. Even if we were to add a an agg/gif format, you would still be seeing this size overhead. Yes, I think I sort of knew that but thanks for clarifying it. To put this bluntly, if the 15% overhead is important for you, you would have to stick with gd aliased rendering and the 6.4 release. I would also like to put this in context: a 15% overhead compared to the 2001 (2005?) outputs does not seem like a big deal given the evolution of available bandwidth since that time. I work in some pretty rural areas where bandwidth is still limited. And when considering bandwidth we need to keep mobile applications in mind alos. MapServer has always been synonymous with speed and from the user experience perspective speed is determined by the whole pipeline. We've been supporting this technologically obsolete rendering mode for many years now, but imo it's time to move on. I completely respect that. I was mainly wondering if there was a way to reduce the anti-aliasing (and consequently the rendered image file size) when using AGG, which I'm sure sounds like a pretty silly question given the "A" in AGG. Thanks for your reply and for all of the work that you do on the MapServer project. Rich regards, thomas Rich On Fri, Jun 7, 2013 at 7:27 AM, thomas bonfort <[email protected]<mailto:[email protected]>> wrote: Devs and Users, Please have a look at RFC99 (http://mapserver.org/development/rfc/ms-rfc-99.html). I am particularly interested in use-cases that would not be supported if GD were to be removed. cheers, Thomas _______________________________________________ mapserver-users mailing list [email protected]<mailto:[email protected]> http://lists.osgeo.org/mailman/listinfo/mapserver-users -- Richard Greenwood [email protected]<mailto:[email protected]> www.greenwoodmap.com<http://www.greenwoodmap.com> -- Richard Greenwood [email protected]<mailto:[email protected]> www.greenwoodmap.com<http://www.greenwoodmap.com> -- Richard Greenwood [email protected]<mailto:[email protected]> www.greenwoodmap.com<http://www.greenwoodmap.com> _______________________________________________ mapserver-users mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/mapserver-users
