[PD] Gem testing

shree bigswift at ufl.edu
Fri Nov 5 20:54:22 CET 2004

i will try these suggestions this weekend chris, thanks for the reply


On Friday, November 5, 2004, at 02:46  PM, chris clepper wrote:

> On Nov 4, 2004, at 4:40 PM, shreeswifty wrote:
>> 4 - 3.5 GB MOVIES with a patch mixing and alpha channeling the films 
>> and moving the stuff with midi
>> With 090, it stutters and skips because i assume it's trying to play 
>> back the audio on the Quicktime as well?
>> With 888pre the behavior is a little better
>> With the April 087 CVS they all play back smooth as silk, but of 
>> course no audio, which is fine for me
> Those findings go counter to the work we did improving the player 
> objects.  0.87 has some big problems playing long clips which were 
> fixed by late last year/early this year (around 0.888).  The current 
> version is designed to play multiple DV clips at once very smoothly 
> using 'auto 1' and 'rate $1', which is something 0.87 did not do well.
> One thing to try is toggling 'hiquality 0/1' since I did change how 
> that defaulted.  Also, use pix_movie if you don't have any processing 
> to do as 'client_storage' works better than with pix_film + 
> pix_texture.  With 4 clips going, client_storage will make a 
> noticeable difference.
> You mention the alpha channel and doing any RGBA work on the CPU is 
> going to slow everything down 3-400% or more because of the YUV->RGB 
> conversion.  Do any alpha manipulations using OpenGL, and keep in mind 
> that nearly every alpha option can be done in YUV apart from actually 
> storing the alpha (in many cases the alpha is used as a surrogate Y 
> anyway).
>> i notice the SIZE of the Gem.pd_darwin is now near 17MBs and the 
>> CVS87 is near 3MBs
> That is due to the static linking of the text libraries on OSX.  Also, 
> many objects were added, and the code improvements like loop unrolling 
> add some size as well.
>> I am on OSX 10.3 g4 laptop 1024 all that.
> On my 1Ghz laptop, I can play 3 NTSC DV clips smoothly using 
> pix_movie, and a fourth would be possible if I wasn't using the 
> internal 4200 rpm drive.
> cgc
Patrick Pagano, B.S., M.F.A. candidate
Research And Development Assistant
Digital Worlds Institute
University Of Florida
(352) 294-2082

More information about the Pd-list mailing list