PDA

View Full Version : Two Questions: Data Rate and MTC



Laura
07-08-2005, 06:20 PM
Hi Richard,
After a first pass of browsing around this forum, let me say thank you for all the detail and clarity to your discussions. I'm sure I'll be returning to the forum regularly.

I have reviewed the codec performance tests you have posted and have a question about data rate. Is there an average data rate you recommend? Or is this so processor/hard drive dependent it is not a good measure of performance? Data rate is such an easy discussion tool when clarifying quality/performance issues with client animators, I'm obviously looking for the best way to discuss Catalyst requirements to optimize computer desktop playback.

Also, I would like to see more discussion on MTC control for Catalyst. I have to show coming up with two Catalysts chasing LTC (converted to MTC by a USB device for input) sent out by Medialon, and running side by side in our set configuration to Doremi video output. I am planning on setting one Catalyst to listing to MTC and have the 2nd Catalyst sync to the first catalyst. Would you recommend having both Catalysts listen to MTC or set up by the meathod I described? Are there other recommendations you would make so that Catalyst performs it's best as compared directly to Doremi playback?

Thanks for your help,
Laura Frank
LuminousFX, LLC

samsc
07-08-2005, 06:23 PM
I have reviewed the codec performance tests you have posted and have a question about data rate. Is there an average data rate you recommend? Or is this so processor/hard drive dependent it is not a good measure of performance? Data rate is such an easy discussion tool when clarifying quality/performance issues with client animators, I'm obviously looking for the best way to discuss Catalyst requirements to optimize computer desktop playback.


what are you trying to do?
standard definition or hi-def?

and how many layers at the same time?

what is the key 'production value' you want to have most of?

is your animation - graphical or filmed live action?

samsc
07-08-2005, 07:55 PM
Also, I would like to see more discussion on MTC control for Catalyst. I have to show coming up with two Catalysts chasing LTC (converted to MTC by a USB device for input) sent out by Medialon, and running side by side in our set configuration to Doremi video output. I am planning on setting one Catalyst to listing to MTC and have the 2nd Catalyst sync to the first catalyst. Would you recommend having both Catalysts listen to MTC or set up by the meathod I described? Are there other recommendations you would make so that Catalyst performs it's best as compared directly to Doremi playback?


How many layers are you planning on needing in catalyst?
How long is the piece you need to sync?
is it hi-def or standard def?

is your production team using drop-frame timecodes and other such things?
use mtc to both catalysts. dont use sync.
---

midi time code has a whole other set of issues.

doremi's typically only do a single video stream with no fx, and have nothing else going on on their dedicated system - so sometimes its not fair to compare them.

main issues with timecode type applications are the potential for slight timing differences between computer timing's and video timings, either because someone is using drop frame timecodes or whatever- and the fact that the computer video output clock cannot be locked to anything.

also two different computer systems - of any type - in this instance - the medialon and catalyst, might have subtely different internal timings which can cause visible drop frames. for example the medialon might have computer clock that runs slightly faster or slower than the Macintosh clock. so 30fps on one computer isnt necessarily exactly 30fps on another.

also typically vga output screens do NOT run at exactly 60Hz or at 59.97 for NTSC, but differ depending on the output resolution, for example on my computer, at 1280x1024, the vga output is 60.02Hz, but at 1024x768 the refresh rate is 59.77Hz ( with respect to the computer clock ).

Some people have used timecode succesfully.
other people have had problems.
there are a number of very complex scenarios interfacing with externally clocked systems.
it is impossible to test all the combinations of equipment out there, and offer any definitive answer, under all circumstances.
interfacing pieces of equipment together always leads to surprises...

For people needing to have doremi's i did put in a control panel so that they can be controlled from catalyst via dmx.

Laura
07-08-2005, 08:07 PM
what are you trying to do?
standard definition or hi-def?

and how many layers at the same time?

what is the key 'production value' you want to have most of?

is your animation - graphical or filmed live action?

Ultimately I'm looking for the baseline answer to codec performance issues, and I think your list of questions implies that there is no easy baseline. So for the sake of discussion, I'll create an average use scenario:
Standard Def, single layer, graphical content, 800x600, photo-jpeg medium that will perform at 30fps.

samsc
07-08-2005, 08:14 PM
Ultimately I'm looking for the baseline answer to codec performance issues, and I think your list of questions implies that there is no easy baseline. So for the sake of discussion, I'll create an average use scenario:
Standard Def, single layer, graphical content, 800x600, photo-jpeg medium that will perform at 30fps.

thats is a certainty.
you will get a single layer of 800x600 photo-jpeg 50% 30fps
( you will probably get 3 layers at this rate )
---

some users will increase the photo-jpeg quality to around 65% if they are using less layers.
Above about 65%, the data rate gets quite high.

---

800x600 is slightly bigger than standard definition - which is 720x480 at ntsc.

you should get 3- 4 layers of photo-jpeg 50% at 30fps.
with a Dual 2.5Ghz Mac and 9800 graphics card.
you should get this with a single scsi drive

i just did a show with 4 layers of 1024x768 content photojpeg 50% using XServe Raid off one dual 2.5Ghz Mac.

Laura
07-08-2005, 08:26 PM
How many layers are you planning on needing in catalyst?
How long is the piece you need to sync?
is it hi-def or standard def?

is your production team using drop-frame timecodes and other such things?
use mtc to both catalysts. dont use sync.

I believe this specific case would be single layer playback of standard def graphical content, for 2.5 minutes, on Mac G5 dual hard drive systems. I will have to explore the drop-frame TC question further.


midi time code has a whole other set of issues.

doremi's typically only do a single video stream with no fx, and have nothing else going on on their dedicated system - so sometimes its not fair to compare them.

main issues with timcode type applications are the potential for slight timing differences between computer timing's and video timings, either because someone is using drop frame timecodes or whatever- and the fact that the computer video output clock cannot be locked to anything.

Some people have used timecode succesfully.
other people have had problems.
there are a number of very complex scenarios interfacing with externally clocked systems.
it is impossible to test all the combinations of equipment out there, and offer any definitive answer, under all circumstances.
interfacing pieces of equipment together always leads to surprises...

For people needing to have doremi's i did put in a control panel so that they can be controlled from catalyst via dmx.

I can see the elements in play that make this extremely complicated. Are there varilables that I can control that will optimize Catalyst performance when driven by MTC? Lower quality video content? A particular flavor of USB MTC input device?

samsc
08-08-2005, 03:05 PM
I believe this specific case would be single layer playback of standard def graphical content, for 2.5 minutes, on Mac G5 dual hard drive systems. I will have to explore the drop-frame TC question further.



you will not have any problems doing that..

2.5 minutes is short enough not to notice anything much.

sometimes peoples requirements are much more complex than this.