Hi to the Jitter forum, this is my first letter and I'm working on my first Jitter patch.
I apologize for my english and Jitter vocabulary, I not a "fluent member " yet
The aim is to project huge images, one is 1280 x 640, processed by a Dual G5
the other is a double 1024x768 screen, processed by a new quad G5, projected using the two DVI/VGA output of the video card
this is piloted by a PB via ethernet
I made my first Jitter/OGL desktable to pilot the videos, mapped on several OGL objets so that i can have different ways to put it on stage.
My guess was basicly to map video on 2 videoplanes overlayed, to fade and sometimes mix different inputs using OGL, and then play with the planes or add other objects. Play mainly with shapes, position fixed background images and masks.
I thought it was reasonnable for a first project but...
It works, but the play rate is frightenning !
First of all, can anyone explain why the play rate is so different when you play a film on QT and via Jitter ? (just read it on a VP)
I built my images on after effects and calculate it in PJpeg 75, all that is perfectly red by Qt even through a Ext HD FW400.
I knew it would be different trough Jitter but the gap is huge ! Yes the CPU has to decompress the image but just like QT does I guess...
How can I make it better ? I used a get movie dim to automatically size the jit.qt.movie
I was planning to have the A film on the hard dik and the B film on an external FW800 disk, so that i never call 2 films on the same disk, but nethertheless, the rate is so low (9 ips for my first 1280 x 1024 video, getting lower and lower....), despite asking nothing else than just playing one, that I was hopeless yesterday evening!
how is loadram working ? shall I use it ? It seems that Jitter is working with my film even if i don't play it (rate 0 ?), and filling the ram until nothing is possible anymore !!!! (my videosequences can be several Go size.) How much ram do I need ?
(By the way, how can i play the same film on 2 Vplanes, with a one second delay between oneanother ?)
Is there a way to pilot the rotation of an OGL object with a speed rotation instead of an angle ? The pb i have is that when I send my rotation informations (which means a lot of informations a sec, generated by the tabledesk metro), it is briefly interrupted as soon as i ask something else to my desktable, such as loading a new film for the next sequence ....
If I calculate movies with different fps to use lowrate when it is possible (10 ips perhaps), and higher rate when it is necessary (20 ips would be probably enough), how do I configure the metro ?
I knew it would be difficult, so I planed also to work with fix Psd image as backgrounds, to free the cpu for a while and prepare the following sequence. But simply translating a plane with an image was not smooth a all ! how is loadram working with my fixed Psd images ? I sent a bang on the image only if i work on it, (if I send an offset, or move a mask.... but i didn't reach this step !) Those fixed images are mapped on a plane like the videos, using matrix with @auto
Will I allways have a rate brakedown if I load for exemple the next image of the next sequence, while playing a video ?
What about metro or qmetro in my case, overdrive or not .....
This is quite a hurry, thank you very much to anyone who could give a hand,