Re: Comparing plots to Matlab

[bpabbott]

On Dec 03, 2010, at 12:59 PM, Jordi Gutiérrez Hermoso <address@hidden> wrote:

2010/12/2 bpabbott <address@hidden>:
> On Dec 02, 2010, at 12:36 PM, Jordi Gutiérrez Hermoso <address@hidden>
> wrote:
>
> > Why are there scanlines or something like it in the fltk plots that
> > have colour patches?
>
>
> My understanding is that OpenGL triangulates all patches and then draws the
> triangles. By "scanlines" do you mean artifacts of that process?

I think so. It shows up in the gl2ps plot, but not on the on-screen
fltk plot. It looks sort of like when you convert a photograph, say of
a person's face, into an .eps file. Do you need me to draw and point
out explicitly point out what I'm referring to?

 

No need. I just wanted to be sure I understood you correctly.

> Perhaps there is some other solution for producing output that will hide the
> triangular mesh?

I can't think of anything other than glReadPixels. I also ask again,
is a vector image format the right format for these pictures? The 3d
surfaces produced by print also look a little weird with the fltk
backend. I didn't run Søren's script (I'm sorry, but where is the
latest version again?), but I imagine this also produces really large
image files.

- Jordi G. H.

 

Just to clarify the vector/bitmap point for those interested ... the png output isn't vector based. However, it is being produced using ghostscript from a postscript image. After the conversion using ghostscript the OpenGL triangulation remains visible.

Regarding glReadPixels, I'm unfamiliar with it, and can't comment on what would be required to make use of it for output. Using glReadPixels for output via print() will require changes as some formats will rely upon imagemagick, and the remaining on ghostscript, for conversion.

Soren has only posted a single version. jwe latter mentioned the changes he used to get the FLTK to (mostly) run.

Ben