Discussion:
[MayaVi-users] High-resolution graphics output
h***@cnrs-orleans.fr
2004-02-17 07:09:03 UTC
Permalink
Does anyone have a good recipe to produce high-resolution graphics
output from MayaVi? It has to be suitable for submission to scientific
journals.

I tried GL2PS, but I am rather disappointed by the quality of the
output. My IsoSurface becomes a totally flat projection of uniform
colour. The image exports seem to produce only screen resolution.
RenderMan might be an option, but a rather expensive one (though I
might consider it if someone can confirm that I do get high-quality
output that way).

Konrad.
Gerard Gorman
2004-02-17 07:27:17 UTC
Permalink
I think this is a general problem of getting the really nice images we
see on our screens to reproduce nicely on paper and power-point type
presentations. I had this problem when I wanted to include the edges of
my computational mesh - they just appeared blurry. I think that unless
you have a vectorised postscript then your choice is between one bitmap
or another. So I'd suggest two things. First choose a bit map output
that gives you the right balance between quality and size (tiff worked
well for me - then I edited it and converted it to other formats using
gimp). Secondly, remember that what you see on the screen will not in
general reproduce the same on paper, so you have to experiment. For
example - when I was visualising mesh edges varying the opacity worked
well for visualising on screen but adjusting the line width worked much
better for reproduction on paper.

Cheers
g
Post by h***@cnrs-orleans.fr
Does anyone have a good recipe to produce high-resolution graphics
output from MayaVi? It has to be suitable for submission to scientific
journals.
I tried GL2PS, but I am rather disappointed by the quality of the
output. My IsoSurface becomes a totally flat projection of uniform
colour. The image exports seem to produce only screen resolution.
RenderMan might be an option, but a rather expensive one (though I might
consider it if someone can confirm that I do get high-quality output
that way).
Konrad.
-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
MayaVi-users mailing list
https://lists.sourceforge.net/lists/listinfo/mayavi-users
--
----------------------------------------------------------
Gerard Gorman
Applied Modelling and Computation Group
Earth Science and Engineering
Imperial College
Prince Consort Road Tel. 00 44 +207 594 9323
London SW7 2BP Fax. 00 44 +207 594 9321
U.K. o~o A good slogan beats a good solution.
-----------------------w-v-w------------------------------
h***@cnrs-orleans.fr
2004-02-17 07:48:01 UTC
Permalink
Post by Gerard Gorman
computational mesh - they just appeared blurry. I think that unless
you have a vectorised postscript then your choice is between one
bitmap or another.
I'd rather have something vectorised... Screen resolution bitmaps are
not of much use, I need something like 600 dpi at least for the final
output.
Post by Gerard Gorman
edited it and converted it to other formats using gimp). Secondly,
remember that what you see on the screen will not in general reproduce
the same on paper, so you have to experiment. For example - when I was
visualising
Indeed, that is hard to avoid. But first I need a good basis for
experimenting.

For molecular visualizations (my main business), I use VMD (a
specialized molecular viewer) and ask it to produce POV-Ray output,
which I then edit by hand for best results. That works quite well.

Konrad.
Hans Fangohr
2004-02-17 07:47:00 UTC
Permalink
Konrad,

nice to read from you hear --- I've been well impressed with your work
(Scientific Python and MMTK)!
Post by h***@cnrs-orleans.fr
Does anyone have a good recipe to produce high-resolution graphics
output from MayaVi? It has to be suitable for submission to scientific
journals.
I tried GL2PS, but I am rather disappointed by the quality of the
output. My IsoSurface becomes a totally flat projection of uniform
colour. The image exports seem to produce only screen resolution.
RenderMan might be an option, but a rather expensive one (though I
might consider it if someone can confirm that I do get high-quality
output that way).
We have certainly used the Renderman format to raytrace the file
subsequently to get high-quality data. There may have been the need to
edit the renderman file slightly (I think the light source was in the
'wrong' position --- let me know if you need details on this).

A faster solution is to create a bitmap with a much higher resolution
than you have on screen. You can achieve this using
Options->Preferences->Mangnification factor for saving images. Choose
3 or 5 or so, and then save your presentation as a bitmap (png or
whatever is suitable). I'd suggest you try this first; it may well do
what you need and it is very easily done.


I think we have also used the gl2ps without the problems you reported
but maybe our visualisation was less demanding -- I would have to
double check this.

Hope this helps,

Hans
Post by h***@cnrs-orleans.fr
-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
MayaVi-users mailing list
https://lists.sourceforge.net/lists/listinfo/mayavi-users
-------------------------------------------------
Dr Hans Fangohr

Computational Engineering & Design Research Group
School of Engineering Sciences
University of Southampton
Southampton, SO17 1BJ
United Kingdom

Location: Building 25, Room 1033
phone : +44 (0) 23 8059 8345
fax : +44 (0) 23 8059 7082
email : ***@soton.ac.uk
-------------------------------------------------
Richard Boardman
2004-02-17 08:11:01 UTC
Permalink
Hi Konrad,
Post by h***@cnrs-orleans.fr
Does anyone have a good recipe to produce high-resolution graphics
output from MayaVi? It has to be suitable for submission to scientific
journals.
The method I've always used for this is increasing the magnification
factor for saving in Options > Preferences to around 3 for figures-on-A4
size (~6 inches wide), and bashing it up to 7 or so for a full-size A3
print. Make sure you don't have anything covering the render area in
MayaVi though since it works on a screen capture basis.

Incidentally, because of the nature of the capture, if you have a
high-quality graphics card which does decent real-time GL effects (one of
the top-end GeForce cards or so) then this improves the quality of the
captured image in this instance.
Post by h***@cnrs-orleans.fr
I tried GL2PS, but I am rather disappointed by the quality of the
output. My IsoSurface becomes a totally flat projection of uniform
colour. The image exports seem to produce only screen resolution.
RenderMan might be an option, but a rather expensive one (though I
might consider it if someone can confirm that I do get high-quality
output that way).
Renderman is a method I've used in the past, however I found I needed to
play a little with the lighting configuration. One can get good results.

Incidentally, Renderman might not be as expensive as you think. The Blue
Moon Rendering Tools may well still be available as a freebie on the
internet, despite the Pixar people getting upset about someone creating a
free thingy which does what their expensive tools do and asking them to
get rid of it. Try googling for BMRT, and see what comes up these days.

Hope this helps,

Richard Boardman o o o o o
Computational Engineering and Design Group o o o o o
School of Engineering Sciences o o o o o
University of Southampton o o o o o
United Kingdom o o o o o
Konrad Hinsen
2004-02-17 09:57:02 UTC
Permalink
Thanks to all who made suggestions on this. I wasn't aware of the
magnification factor, that could already prove sufficient for my purposes.

Concerning RenderMan, I found a renderer that is (at least temporarily)
distributed for free:

http://www.3delight.com/index.htm

A quick test shows that it works with MayaVi output, though lightning must
indeed be improved. I can't say any more at this moment because my experience
with RenderMan files is limited to one single try :-) But perhaps this is of
interest to others as well.

Konrad.
--
-------------------------------------------------------------------------------
Konrad Hinsen | E-Mail: ***@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
Rue Charles Sadron | Fax: +33-2.38.63.15.17
45071 Orleans Cedex 2 | Deutsch/Esperanto/English/
France | Nederlands/Francais
-------------------------------------------------------------------------------
Arnd Baecker
2004-02-17 10:31:02 UTC
Permalink
Post by Konrad Hinsen
Concerning RenderMan, I found a renderer that is (at least temporarily)
http://www.3delight.com/index.htm
A quick test shows that it works with MayaVi output, though lightning must
indeed be improved. I can't say any more at this moment because my experience
with RenderMan files is limited to one single try :-) But perhaps this is of
interest to others as well.
But beware that there is the danger that it gets
withdrawn as with BMRT - (since then I stopped investing time
in using non-open-source projects)

Another, but open source, candidate is aqsis
http://aqsis.sourceforge.net/index.php

Best,

Arnd
Prabhu Ramachandran
2004-02-17 18:02:03 UTC
Permalink
KH> I tried GL2PS, but I am rather disappointed by the quality of
KH> the output. My IsoSurface becomes a totally flat projection of
KH> uniform colour. The image exports seem to produce only screen
KH> resolution. RenderMan might be an option, but a rather
KH> expensive one (though I might consider it if someone can
KH> confirm that I do get high-quality output that way).

I find that GL2PS usually does a fair job, however, I have not
rendered very complex scenes and generated PS for them. How did you
compile GL2PS? What version? Does this particular image you are
rendering have a large number of colors? What platform are you
running this on?

cheers,
prabhu
h***@cnrs-orleans.fr
2004-02-17 19:39:02 UTC
Permalink
Post by Prabhu Ramachandran
I find that GL2PS usually does a fair job, however, I have not
rendered very complex scenes and generated PS for them. How did you
compile GL2PS? What version? Does this particular image you are
rendering have a large number of colors? What platform are you
running this on?
I just enabled GL2PS in VTK 4.4.0, I don't know which version of GL2PS
that is. I use it on a Macintosh.

The image doesn't have many colors but it has many shades of a single
color. It is simply an IsoSurface of some potato-shape object (more
precisely, a low-resolution electron density map for a protein). The
image rendered by GL2PS has only a uniform colour, which means that all
the relief is gone.

Konrad.
--
------------------------------------------------------------------------
-------
Konrad Hinsen | E-Mail:
***@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
Rue Charles Sadron | Fax: +33-2.38.63.15.17
45071 Orleans Cedex 2 | Deutsch/Esperanto/English/
France | Nederlands/Francais
------------------------------------------------------------------------
-------
Prabhu Ramachandran
2004-02-18 11:33:03 UTC
Permalink
Post by Prabhu Ramachandran
I find that GL2PS usually does a fair job, however, I have not
rendered very complex scenes and generated PS for them. How
did you compile GL2PS? What version? Does this particular
image you are rendering have a large number of colors? What
platform are you running this on?
KH> I just enabled GL2PS in VTK 4.4.0, I don't know which version
KH> of GL2PS that is. I use it on a Macintosh.

That should be fine.

KH> The image doesn't have many colors but it has many shades of a
KH> single color. It is simply an IsoSurface of some potato-shape
KH> object (more precisely, a low-resolution electron density map
KH> for a protein). The image rendered by GL2PS has only a uniform
KH> colour, which means that all the relief is gone.

Thats surprising. I've tried with a few simple scenes (and some
not-so-simple ones) and it does seem to render it close to whats seen
on screen. Its not perfect, but is pretty good. Its ideal for
line-contour plots. Even changing the lighting (press the 'l' key to
configure the lighting) affects the PS/PDF generated. Perhaps viewing
the PS in GV might not be as colorful as rendered on screen. You
might want to try rendering a PDF file and seeing how it looks on
acroread? Of course, you can't really compare it to a scene rendered
using povray/renderman.

regards,
prabhu
Konrad Hinsen
2004-02-18 13:28:04 UTC
Permalink
Post by Prabhu Ramachandran
the PS in GV might not be as colorful as rendered on screen. You
might want to try rendering a PDF file and seeing how it looks on
acroread? Of course, you can't really compare it to a scene rendered
using povray/renderman.
I tried PDF first (it's better supported by MacOS than EPS these days, but the
result is exactly the same.

I didn't go through all the options, though, just the default.

Konrad.
--
-------------------------------------------------------------------------------
Konrad Hinsen | E-Mail: ***@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
Rue Charles Sadron | Fax: +33-2.38.63.15.17
45071 Orleans Cedex 2 | Deutsch/Esperanto/English/
France | Nederlands/Francais
-------------------------------------------------------------------------------
Loading...