<div>Hi Krunoslav,</div> <div> Forget about english ,You and Frank are just awesome,Really Nice and Great Work !!! Thanks for sharing your work with the community.</div> <div> </div> <div>Cheers</div> <div>Max</div> <div> </div> <div><BR><B><I>Krunoslav <a_krunoslav@gmx.net></I></B> wrote:</div> <BLOCKQUOTE class=replbq style="PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #1010ff 2px solid"><BR>First of all, sorry on my bad English, I hope that next article will be<BR>understandable enough. Second, special thanks to Frank Warmerdam, who helped<BR>me a lot with handling hq tiff images.<BR><BR>I wrote this small howto, even for reminding myself, if I forgot how I did<BR>it :)<BR><BR>This howto is based on my experience with Mapguide Opensource, high quality<BR>aerial photography and FDO provider for GDAL. Images I used were
GeoTIFF<BR>images with resolution 1px=0,5m.<BR><BR>If you have large number of high-quality georeferenced images (ortophoto or<BR>other aerial images), obvious problems are:<BR><BR>- size of images (usually an area of 3x2km high-quality 24bit RGB image has<BR>over 100MB)<BR>- sacrificing quality of images to improve speed<BR>- large number of files<BR><BR>So, our goals are: <BR><BR>- to maintain highest quality of images (no compression) to show full<BR>details at closer zoom levels<BR>- to optimize loading time of images at higher zoom levels<BR><BR>Our basic approach should be to optimize images, and to reduce number of<BR>files as possible. So, to reduce number of image files, we must merge as<BR>many tiff as possible (please keep in mind that geotiff image size can’t be<BR>more than 4GB). So we must aim for the images close to 4GB. For merging, the<BR>best tool I’ve found is FWTools (http://home.gdal.org/fwtools). <BR>Since GDAL provider for raster doesn’t
support transparency, please keep in<BR>mind to merge images so that they close a rectangle area. Otherwise, you<BR>will have black background where there are no images. Now, let’s merge<BR>images: run Fwtools shell, and enter:<BR><BR>cd bin<BR>gdal_merge.bat image1.tif image2.tif image3.tif –o out.tif<BR><BR>This will merge images 1,2 and 3 to out.tif. Now we have to optimize<BR>out.tif: we will convert this image to tiled format, and to add overviews:<BR><BR>gdal_translate -co TILED=YES out.tif out_tiled.tif <BR>gdaladdo out_tiled.tif 2 4 8 16 32 64 128<BR><BR>This image is now optimized. Images should load really fast at lower zoom<BR>levels.<BR><BR>Anyway, if you have high-res images as I do (1px=0,5meters), performances<BR>will still be poor at further zoom levels. This is understandable, since<BR>Provider for raster has to read large image area, and enormous file size.<BR>But, we can use another method here:<BR><BR>Resize this high-quality tiff’s for usage
above closest zoom levels. For<BR>example, I use 1px=2meters for zoom levels 1:2000-1:10000, 1px=5meters<BR>1:50000 and 1px=10meters for levels above 1:50000. This way you will have<BR>few small tiff images, which are still good enough for those zoom levels,<BR>because at 1:50000 you really can’t see any difference between 0,5 and 10<BR>meters per pixel resolution.<BR><BR>Next, you build separate layer for each resolution. In above case, you would<BR>have four layers. Number of needed layers and zoom levels depends on images<BR>quality and on server performance. The best way is to experiment, and see<BR>what is happening.<BR>With method described above, raster layer doesn’t take more then 3 seconds<BR>to load, at any zoom level, and I think this is pretty good result.<BR><BR>All suggestions and comments are welcome.<BR><BR>-- <BR>View this message in context: http://www.nabble.com/HQ-aerial-photography-with-MGOS---howto-tf3809098s16610.html#a10780689<BR>Sent from the
MapGuide Users mailing list archive at Nabble.com.<BR><BR>_______________________________________________<BR>mapguide-users mailing list<BR>mapguide-users@lists.osgeo.org<BR>http://lists.osgeo.org/mailman/listinfo/mapguide-users<BR></BLOCKQUOTE><BR><p> 
<hr size=1>Don't pick lemons.<br>
See all the <a href="http://autos.yahoo.com/new_cars.html;_ylc=X3oDMTE0OGRsc3F2BF9TAzk3MTA3MDc2BHNlYwNtYWlsdGFncwRzbGsDbmV3Y2Fycw--">new 2007 cars</a> at <a href="http://autos.yahoo.com/new_cars.html;_ylc=X3oDMTE0OGRsc3F2BF9TAzk3MTA3MDc2BHNlYwNtYWlsdGFncwRzbGsDbmV3Y2Fycw--">Yahoo! Autos.</a>