Compression in TIF files: disk space vs. performance

Andreas Neumann neumann at KARTO.BAUG.ETHZ.CH
Mon May 15 16:54:56 EDT 2006


Thank you Ed and Armin for your suggestions.

I will give a try and report how it works.

Sounds reasonable to compress a number of lines or tiles rather than the 
whole image at once.

Thanks,
Andreas

Armin Burger wrote:

> Andreas,
>
> I made once quite good experiences with scanned b&w images using the 
> Fax Encoding (CCITT) Group 4 TIFF compression. The images had no 
> overviews, but were also just shown at a scale that was reasonable for 
> the details on them. The compression was typically good, something 
> like 3 to 5 times.
>
> Armin
>
> Andreas Neumann wrote:
>
>> Hello list,
>>
>> I have a question on compression in tiff files.
>>
>> I read that for best mapserver performance it is recommended to store 
>> tiff files on the server in uncompressed format. We currently have a 
>> disk space problem and need to serve a large number of raster images. 
>> Is there any compression technique that allows to save diskspace 
>> while maintaining a reasonable or good performance? Or is compression 
>> not recommended at all?
>>
>> Another question: we need to serve 1bit (b+w) rasterimages that are 
>> delivered as gray/rgb jpegs later because we want to use resampling. 
>> What is the best color depth to store the original images. I noticed 
>> that when keeping them in 1bit color depth one cannot create pyramids 
>> using gdaladdo. Is it better to store them in 1 bit with no pyramids 
>> or in 8bit gray value with pyramids?
>>
>> Thank you for any help on my questions,
>>
>> All the best,
>> Andreas
>>


-- 
----------------------------------------------
Andreas Neumann
Institute of Cartography
ETH Zurich
Wolfgang-Paulistrasse 15
CH-8093  Zurich, Switzerland

Phone: ++41-44-633 3031, Fax: ++41-44-633 1153
e-mail: neumann at karto.baug.ethz.ch
www: http://www.carto.net/neumann/
SVG.Open: http://www.svgopen.org/
Carto.net: http://www.carto.net/



More information about the mapserver-users mailing list