LibTiff Mailing List
TIFF and LibTiff Mailing List Archive
January 2004
Previous Thread
Next Thread
Previous by Thread
Next by Thread
Previous by Date
Next by Date
Contact
The TIFF Mailing List Homepage
This list is run by Frank Warmerdam
Archive maintained by AWare Systems
Thread
2004.01.22 11:50 "Re: change fax resolution", by Joris Van Damme
> - either resolution is low, in which case I don't recommend skipping every > 2nd line since some meaningfull lines or parts of text and such may be > (almost) horizontal and just one pixel wide and skipping may in that case > lead to really undesirable results > - or either resolution is high enough, and skipping every 2nd line > probably > is almost as good as resampling + dithering despite of theoretical > difference. When I said 'line width', I actually meant line thickness, or... What is the correct word to use here? When I talked about 'either resolution is low or high enough', that may have sounded confusing, since exact resolution was specified. I actually meant more like content-related resolution. Like, it's quite possible that the images contain text of only 7 or 8 pixels high, and lines of one pixel 'thick', in wich case skipping every 2nd line is certainly not going to yield exceptable results, while it's also possible all text is eg some bold font and at least 20 pixels high, and lines are at least 3 pixels thick, in which case skipping every 2nd line should yield very nearly the same quality as resampling+dithering (I guess). Thinking about this further, I am now wondering how the original PNG images came about. Is there already a dithering step involved in creating these black and white PNG's? If it's possible to render full 8bit grayscale PNG's instead, truelly using the 8bit grayscale range, then resampling these instead and dithering afterwards will certainly yield much more exceptable quality. And even if it's not (eg because text and line drawing is rendered on the PNG without anti-aliassing), it may be worthwhile rendering the PNG at eg 800x800 resolution, black and white, resampling to 200x100 8bit brightness, and dithering to black and white next. My guess is that either of these two last options will yield considerably better quality. Joris Van Damme info@awaresystems.be http://www.awaresystems.be