PyMOL Molecular Graphics System

yvonne leduc wrote:
> I would like some confirmation or explanation about the differences I=20
> see, and the way I understand it.
Hello Yvonne,

The command "ray" does not care about the size of your display, it may=20
even be used in batch mode on a computer that does not have a monitor=20
attached to it at all.

To prevent the "cropping"  or "zooming" on images created by the "ray"=20
command you have to make sure that the aspect ratio of the width and=20
height of the ray command are the same as the aspect ratio of your=20
current viewport.

Example 1)
	if you will be rendering (thats what the command ray does)
	figures of 2000 x 2000 you should choose a viewport of
	say 700x700, then:

	2000 / 2000 =3D 1 =3D 700 / 700

    - so you should execute the command:

	viewport 700, 700

    - and then fine tune your figure and then run the ray command:

	ray 2000, 2000


Exmaple 2)
     - you want images at 3000 Width x 2000 Height

	3000 / 2000 =3D 1.5

    - your limiting display size is your height so
      try for instance height =3D 768

	width =3D height * 1.5 =3D 768 * 1.5 =3D 1152

    - but 1152 is wider than your display (1024) so lets try height =3D 5=
00

	width =3D 500 * 1.5 =3D 750 (which is good now, or at least it fits)

   - so you would issue the command

	viewport 750, 500

   - then ray 3000, 2000

> And the reason for all this is ...
> ... is a *.png image created with ray 1000,750 going to be good enough=20
> for publication standards?  Is there much benefit in going higher?  wha=
t=20
> causes the limits, if any?

The journal usually specifies the resolution at which the images will be=20
printed.

If the journal says images print at 300 dpi (Dots Per Inch) and your=20
figure will have a final printed width of 3 inches then you want to=20
render images with a width of 900.

The publisher will usually scale the picture with a very good software=20
so even if it has to be "enlarged" by a few pixels the result will be=20
good. But if it has to be shrinked you may run into trouble with labels,=20
thin lines etc.

Hope this helped a bit.

Zac
--=20
Ezequiel PANEPUCCI, Ph.D. - Institut Pasteur
Plate-Forme 6 - Cristallog=E9n=E8se et Diffraction des Rayons X
Telephone: +33 (0)1 44 38 94 57
Portable:  +33 (0)6 13 14 39 06

On Thursday 20 January 2005 15:37, yvonne leduc wrote:
> [snip]
> And the reason for all this is ...
> ... is a *.png image created with ray 1000,750 going to be good enough
> for publication standards?  Is there much benefit in going higher?  what
> causes the limits, if any?

Journals typically demand at least 600 dpi (dots per inch) resolution for 
figures. So if you want a figure 1 inch (2.54 cm) to a side, it will have to 
be 600 pixels to a side. A typical figure is about five by five centimeters, 
so that would call for a 1200x1200 pixel image.

I have never run into any practical limitations using the PyMOL raytracer, 
which produces images of the requested size independent of the screen or 
window size, even if the image becomes so large it cannot be displayed in the 
PyMOL window (using "ray sizex,sizey" and "png filename").

Hope this helps,

-- 
Lieven Buts
Ultrastructure Laboratory
Vrije Universiteit Brussel

Thanks very much to all who answered, for the detailed and helpful 
responses to my questions.

Yvonne

Building on the two previous posts, is it necessary to use 'viewport' 
before executing 'ray'?

this:
    ray 2000, 2000

instead of:
    viewport 700, 700
    ray 2000, 2000

Thanks,
Doug

On Jan 20, 2005, at 10:11 AM, Ezequiel H Panepucci wrote:

>    - so you should execute the command:
>
> 	viewport 700, 700
>
>    - and then fine tune your figure and then run the ray command:
>
> 	ray 2000, 2000

Douglas Kojetin wrote:
> Building on the two previous posts, is it necessary to use 'viewport' 
> before executing 'ray'?
> 
> this:
>    ray 2000, 2000
> 
> instead of:
>    viewport 700, 700
>    ray 2000, 2000

I didn't quite understand your post. The viewport command prior to ray 
tracing is just so you have on the interactive graphics windows the same 
aspect-ratio of WidthxHeight that you will have on your ray command. You 
only do it once before starting to work on the picture, or if you decide 
to ray the image with a different width and height.

This helps preventing cropping the image accidentally and only realizing 
this after waiting 6 minutes or so for the ray tracing to finish.

The original_400x400.jpg file attached was generated with the png 
command *without* ray tracing on a viewport of 400 x 400.

The ray_700x800.jpg file attached was generated with the png *with* the 
ray tracing command "ray 700, 800".

You notice on the ray image that the cartoon is cropped.

I did some heavy loss compression to reduce the file sizes, usually 
pymol's ray command does a better job than that ;-)

Zac