gnuplot-beta

new demo with cyan lines

[Daniel J S. <dan...@ie...>]

On the new demo in 'pointsize.dem', the globe has cyan longitude/latitude (x11) and there are some lines and arrows apparently marking axes.  But the lines are cyan as well and things look confusing.  Is there a better color for the axes?

Dan

Re: new demo with cyan lines

[Ethan A M. <merritt@u.washington.edu>]

On Monday 20 February 2006 11:36 pm, Daniel J Sebald wrote:
> On the new demo in 'pointsize.dem', the globe has cyan 
> longitude/latitude (x11) and there are some lines and 
> arrows apparently marking axes.  But the lines are cyan as 
> well and things look confusing.  Is there a better color for the axes?   

Whatever.

But that reminds me of something else shown by the same demo.
The axes in a 3D plot with "set hidden3d" are never obscured
by the plotted surface. That tends to destroy the visual impression
of depth.  I think that in this case the axis vectors should be
handled by the hidden3d code, just like any other plot vectors.

-- 
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle 98195-7742

Re: new demo with cyan lines

[Daniel J S. <dan...@ie...>]

Ethan A Merritt wrote:
> On Monday 20 February 2006 11:36 pm, Daniel J Sebald wrote:
> 
>>On the new demo in 'pointsize.dem', the globe has cyan 
>>longitude/latitude (x11) and there are some lines and 
>>arrows apparently marking axes.  But the lines are cyan as 
>>well and things look confusing.  Is there a better color for the axes?   
> 
> 
> Whatever.
> 
> But that reminds me of something else shown by the same demo.
> The axes in a 3D plot with "set hidden3d" are never obscured
> by the plotted surface. That tends to destroy the visual impression
> of depth.

That was kind of my point.  The eye doesn't immediately grasp what that visual is supposed to be.  Something just isn't right about it.

Dan

Re: new demo with cyan lines

[<br...@ph...>]

Ethan A Merritt wrote:

> But that reminds me of something else shown by the same demo.
> The axes in a 3D plot with "set hidden3d" are never obscured
> by the plotted surface. 

Which "axes" are you talking about there?  The only standard
plot element that could fit that name, i.e. the scales on the borders,
*are* hidden by the surface.

The only things that don't get hidden by the surface are arrows.  They 
would need special handling added to hidden.dem, and it would almost 
certainly still fail to correctly hide the point, especially if the 
terminal draws its own arrows.

Re: new demo with cyan lines

[Ethan M. <merritt@u.washington.edu>]

On Thursday 23 February 2006 02:40 am, Hans-Bernhard Br=F6ker wrote:
> Ethan A Merritt wrote:
> > But that reminds me of something else shown by the same demo.
> > The axes in a 3D plot with "set hidden3d" are never obscured
> > by the plotted surface.
>
> Which "axes" are you talking about there?  The only standard
> plot element that could fit that name, i.e. the scales on the
> borders, *are* hidden by the surface.

My apologies. I had forgotten that that plot used explicit arrows
rather than the zero-axis lines.

The plot which was annoying me was actually a different one,
the one I recently constructed for rgb_variable.dem;
when Daniel mentioned the plot in pointsize.dem, I falsely
concluded that it was a parallel case.

I will construct a full example pair if you like, but my observation
is that in case such as the final plot in rgb_variable.dem, the axes
are properly occluded by other plot elements if you draw them by
including a dataset of  unit vectors in 3D, but not properly occluded
if you draw them via the "set zeroaxis" command.
Quite probably a bug in the code I added myself.

=2D-=20
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle WA

Re: new demo with cyan lines

[<br...@ph...>]

Ethan Merritt wrote:

> I will construct a full example pair if you like, but my observation
> is that in case such as the final plot in rgb_variable.dem, the axes
> are properly occluded by other plot elements if you draw them by
> including a dataset of  unit vectors in 3D, but not properly occluded
> if you draw them via the "set zeroaxis" command.

I don't think that would make a difference.  The problem is simpler: 
hidden3d doesn't try to hide points behind lines.  For actual points, 
that wouldn't be a problem: the probability of in infinitely small point 
actually coinciding with a zero-width line is zero.  The problem is that 
point symbols aren't zero-size, and lines aren't zero-width.  To fix 
that would take almost a rewrite of hidden3d.  I'm not sure we want to 
go there.

There's more wrong with those plot than just the hiding of the zeroaxis 
itself.  E.g. the direction of the tics on the "Green" (y) axis the 
wrong way round --- the tick labels are inside the plot, the ytics point 
outward, rather the opposite of what the sate of 'set ytics in' would 
suggest.

And the "Warning: Single isoline..." message from the pm3d code is 
getting rather in the way of rotating that plot via mouse.  That warning 
needs to be limited to at most a few displays per gnuplot session.