On 2007-09-19 21:56-0400 Hazen Babcock wrote: > Hopefully I've fixed this (shear allowed below the baseline) in v7876.
Yes, and thanks. Note, that the png and X backends of cairo (pngcairo and xcairo) have trouble with a shear aligned exactly with the baseline so the updated x28c.c (revision 7883) offsets the angle by domega = 0.05 radians to avoid that situation without noticeably affecting what the plot looks like. I am not going to put in a bug report for the cairo backends because this case is too complicated. However, I am sure I am right that it is a problem specific to the png and X back-ends for cairo because the Postscript back end of cairo (pscairo) and the PLplot core (Hershey fonts and/or non-cairo device drivers with non-Hershey fonts) did not have any trouble with domega = 0. > Ok, I see what you mean now (about character size problem for the "revolution" case) . Since the characters are the same size the only > way I can see to fix this is to change the spacing. Actually, it is not simply spacing. If you look at the first attached screenshot for the latest x28c.c, the disparity of length of string for the z=zmin case for omega = -Pi/4 (the string only goes ~ 40% to the corner) and omega = Pi/4 (the string goes well beyond the corner) is increased for this more extreme viewing angle but basically confirms the previous results. However, please also look at the shape of the "o" characters in those two strings. They are almost perfectly round (as directly viewed) which means their inferred 3D shape is strongly elliptical. Of course, the inferred 3D shape should be circular. To solve this issue the -Pi/4 (short) string needs all its characters stretched out along the axis of the string while the Pi/4 (long) string needs all its characters compressed along the axis of the string. The size of those characters perpendicular to the string axis look pretty good in both cases. Note, the screenshot also has similar rotation results for x=xmax, and y = ymax. The errors are much more subtle for those cases, but they are there as well. Also, you can make the errors for those cases arbitrarily large by picking extreme viewing angles (such as azimuth = 85 or so). Note the first attached screen shot was produced by viewing with gv the first page of results from ./x28c -dev psc -o test.ps --drvopt text=0 You can also get very similar results by running ./x28c -dev xwin The second screenshot displays the second page of the above results which shows that (a) rotation is working fine, but (b) it also shows the justification problem for Hershey fonts (as before). Those results are properly centred on the x=xmax, y = ymax, and z=zmin planes for non-Hershey fonts. Alternatively, I can get the same results for the Hershey case by negating the justification specified for plptex3. The third page of the example (not attached) shows shear in the direction of the baseline of the string) is working fine (but also illustrates the Hershey justification problem), and the fourth page of the example (not attached) shows various 3D axis titles done using plmtex3 are working fine. I am tentatively thinking of a cool-looking fifth page for x28c.c which will show a 3D character string following a complex path in 3D space with the appropately changing inclinations and shears for each character position of the string, but I am afraid it would look pretty crummy until the above "revolution" issue is sorted out. Alan __________________________ Alan W. Irwin Astronomical research affiliation with Department of Physics and Astronomy, University of Victoria (astrowww.phys.uvic.ca). Programming affiliations with the FreeEOS equation-of-state implementation for stellar interiors (freeeos.sf.net); PLplot scientific plotting software package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project (lbproject.sf.net). __________________________ Linux-powered Science __________________________ ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
