VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Wed Feb 17 2010 - 19:57:08 CST

Hi Dan,
  The quality of the final video output depends strongly on
the compression ratio and the limitations (if any) of the
video codec you're using.

Have you read through the VMD images and movies tutorial yet?

If you're making movies that are low resolution, e.g. 320x240,
then MPEG-1 works fine. Once you get up to 640x480 or higher,
you really must begin using MPEG-2, MPEG-4, or other codecs
that are better suited to high resolution video streams,
and you'll also need to adjust the bitrate from something like
5 megabits/sec for DVD-like 640x480, up to > 15 megabits/sec or more
if you start doing HD movies. To make movies in the more sophisticated
formats, you may need to set the VMD movie maker to generate the raw
images only, and then use a high-quality commercial movie encoder on
the generated frames rather than using the 'ppmtompeg' tool the
the VMD movie maker launches by default.

Another thing you can do to get better results with the same bit rate
and/or resolution is to set the video the frame rate to 24 frames per
second rather than 30. This will result in fewer, but higher quality,
frames per second.

Beyond the choice of video codec and bitrates, some other things
that can have a strong effect on the quality of the resulting movie are:
  1) Minimum visible feature size, line width, point size, etc.
     In general, you should choose line, point, and other geometry sizes
     so that when displayed in the final image, they span at least 2 pixels
     in both X and Y directions. If you get 1-pixel sized objects in the
     scene, they'll often get blurred into the surrounding colors, and/or
     they invite various types of compression artifacts.
     
  2) Choice of colors. When you make images for movies, you want to
     keep in mind that unlike the on-screen OpenGL display
     (which uses an RGB color space), most movie encoders are designed
     to perform well for natural colors, which are well represented by
     luminance-chrominance color spaces (e.g. YUV, YCrCb, etc).
     Unfortunately, the entire RGB color space often won't fit within
     the color gamut of these video-oriented color spaces, and the result
     is that during compression, you'll get color shifting artifacts
     where the original RGB image went beyond the color gamut of the
     target codec's color space. It is often a good idea to avoid the
     use of saturated primary colors like bright red, green, and blue.
     If you choose somewhat moderated colors, they will compress better
     than the pure colors do and have fewer resulting artifacts.

  3) You'll get better movie encoding results from movie sequences where
     less change occurs per frame. It's better to have a slow,
     smooth rotation, than a rapid jerky one, as it uses less bits and
     compresses better.

  4) USE ANTIALIASING. If your GPU doesn't have antialiasing enabled
     by default in the OpenGL driver, or VMD is unable to turn it on
     for you, then you should make sure that you use the "half-size rescaling"
     options in the VMD movie maker plugin. This will take the original
     incoming video frame and average groups of 4 pixels together into
     a single output pixel, reducing the resolution significantly, but
     avoiding aliasing artifacts which tend to compress poorly. You'll want
     to start out with a high-res VMD window size when you make the movie
     if you use this option.

If you follow some of these suggestions, I think you'll get better
results. And yes, you may also want to use VdW rather than points.

Cheers,
  John Stone
  vmd_at_ks.uiuc.edu

On Wed, Feb 17, 2010 at 05:51:03PM -0600, Dan Lussier wrote:
> Hi,
>
> I am trying to make high quality renders/video on the Mac from my MD
> simulations. I am using the 'Movie Maker' in VMD but have been
> disappointed with the final appearance once I push the still frames
> into a movie (already tried differing movie formats).
>
> Another person in my lab made great looking movies on Windows using
> Video Mach which got me to thinking about this issue.
>
> Also - does anyone know if using point rather than vdW representation
> could be a source of the poor image quality?
>
> If anyone on OS X has a good work flow for making high quality movies
> from atom/particle simulations I'd really appreciate it if you could
> post it.
>
> Many thanks!
>
> Dan Lussier 

-- 
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email: johns_at_ks.uiuc.edu                 Phone: 217-244-3349
  WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078