When VMD creates the output file it will try to match the current view and screen size. For the most part it does a good a job but there are some problems. The colors in the final raster image seem to be faded when compared to the screen. We may be able to correct this in future versions by artificially increasing the color intensity. In the meantime, this can be remedied with a program like xv or display which will brighten and sharpen the image.
The eye position seems to be slightly different between the screen and raster images. This occurs because the Screen Height and Screen Width (see here) values are not propagated to the output file, so the programs have slightly different transformation matrices. We have added an extra scaling factor, determined by trial and error, which reduces this effect for the default screen sizes, but the factor is only valid if the parameters are not changed.
The lights are supposed to be positioned in the output file as they are on the screen but not all formats support the number of lights available in VMD. In addition, there seems to be a bug in the code that determines the correct light positions. This can be changed by editing the output file by hand.
A new option in VMD draws cylinders with flat disks covering the ends. Some of the renderers may not draw the capped ends.
We have primarily tested the Raster3D output and have little experience with the other programs. The only problem we have found with Raster3D is if there are too many objects to render (this could easily occur when using `Tube' or `Ribbon' on large molecules), it will not render anything.