PDA

View Full Version : Crossfade glitches after long cues



micpool@mac.com
20-03-2006, 10:57 AM
I have a show running on 2 catalyst systems. 1 of the systems is a standard uk rental configuration and outputs 2 800x600 feeds. All material is encoded photo jpeg at medium quality at 25fps. All clips play unity scale and, unless stills, play oncefw>0.

Most of the show consists on crossfading video on both outputs simultaneously. In the main this has worked flawlessly but there are 2 points in the show where the system has regularly glitched. Both these cues are crossfades after the system has been playing out a long 8 minute clip for about 6 minutes. The long clips are of very active surface water reflections. The outgoing video has regularly paused at this point for a few frames or one or both screens has blanked for a second or so.during the crossfade

The first remedy I applied was to preload the incoming clips. This improved reliability but didn?t eradicate the problem. I then re compressed these clips using less than medium quality, the data rate is now well under 1MB/s I then changed the preloaded clips to stills by preloading and only using the inframe in the cue. Effectively then I am crossfading 2 preloaded still frames IN with 2 running video clips OUTt. I thought this had sorted it but on Saturday the outgoing video froze for a second at the point the crossfade started.

I am convinced this behaviour is related to the length of running video preceding the crossfade rather than due to some vagaries of the rental system because the other system on the show (Catalyst 3.31 on a G5 Quad running 10.4.4 with QT7 etc) has glitched, very occassionally, momentarily at exactly the same points.

One other surprising aspect of this is that when the system glitched badly (i.e before I converted the incoming cues to stills) and blanked the screens one of the projectors actually lost sync (the resolution message came up on screen when it regained it a couple of seconds later.) The projectors are fed with Smart E vga to CAT5 systems. I can understand the system not being able to find a frame to output but not this disturbing the output sync.

Finally, the only feature of Catalyst I have difficulty getting along with is the fact that it outputs black (or more accurately a transparent layer) whenever it has any delay in loading a clip . If it didn?t do this you could just jump from clip to clip (or still to still) on a single layer when all you wanted were hard cuts which would make programming so much easier. Why can?t the system say to itself ? have I got a new valid video frame to place on this layer?? and just hold on to what it has when the answer is no. If I don?t tell it I want a black screen I don?t expect it to give me one, ever!

Mic

samsc
20-03-2006, 11:57 AM
One other surprising aspect of this is that when the system glitched badly (i.e before I converted the incoming cues to stills) and blanked the screens one of the projectors actually lost sync (the resolution message came up on screen when it regained it a couple of seconds later.) The projectors are fed with Smart E vga to CAT5 systems. I can understand the system not being able to find a frame to output but not this disturbing the output sync.

Mic

there is nothing in catalyst that could cause this to happen.

this might be a power glitch caused by a lighting or video cue.

certainly nothing in software would cause projectors to loose sync.

suggest you look for other factors related to staging or other equipment



Finally, the only feature of Catalyst I have difficulty getting along with is the fact that it outputs black (or more accurately a transparent layer) whenever it has any delay in loading a clip . If it didn?t do this you could just jump from clip to clip (or still to still) on a single layer when all you wanted were hard cuts which would make programming so much easier. Why can?t the system say to itself ? have I got a new valid video frame to place on this layer?? and just hold on to what it has when the answer is no. If I don?t tell it I want a black screen I don?t expect it to give me one, ever!

Mic

the problem here is that loading a new movie takes a variable ammount of time.
and so if the effect is to switch from one clip running to another.
suggest cue programming strategy like this-

preload the movie clips required into different layers.-- once they are loaded they play frames very quickly.
use "play if intensity >0" play modes on the layers you want to switch between.
then simply move the intensity to 0% if you want to pause it, and 100% for it to play.

this might be a simpler solution for you.

-- control the cutting by using intensity.
this should product the most seemless transitions.

Its much better to preload the required movies onto different layers - than it is to load a new movie onto the same layer.
loading a new movie onto a layer can take too much time

----

its also the computer resources required on older machines to hold copies of the last movie loaded when the next one loads...

micpool@mac.com
20-03-2006, 12:59 PM
Thanks for your prompt reply.

I think the main thrust of my initial problem might have got obscured by my later points.

Leaving aside the loss of sync issue (which I accept is unrelated if Catalyst couldn't have caused it):

I am unclear as to whether you are saying that when Catalyst is overloaded it should only drop frame rate and pause momentarily, or whether blank output can be expected under some overload conditions.

I still have a glitch, which has caused screen blanking, that occurs at 2 specific points in the show which have the common condition that they have been playing long cues before a crossfade.

We have replicated these problems outside show running when all other systems have been off so I don't think its a power or RF issue.

The crossfades on these cues are also quite long (20s) and I appreciate that, as originally programmed, for the duration of these crossfades Catalyst was playing out 4 layers of 800x600 photojpeg. But having changed the incoming footage to inframe it is only playing moving video on 2 of the 4 layers involved for the crossfade duration which should be well within the system capability.


Thanks

Mic

BTW the performance of 3.31 on the G5 Quad has been very good. I am playing out 1024x768 Photo jpegs on screen1 and 800x600 jpegs on screen 2 and everything is pretty solid during simultaneous crossfades on both screens.

samsc
20-03-2006, 02:24 PM
I still have a glitch, which has caused screen blanking, that occurs at 2 specific points in the show which have the common condition that they have been playing long cues before a crossfade.


i can think of nothing in software that would case a screen to go blank.
nothing - no overload - or anything. it would never do that.
unless something is going wierd in your dmx.


suggest you check the dmx input/output and check whether there isnt some funky dmx error, that the console is not outputting zero for a brief moment.

in my experience it would not be impossible that you have encountered some kindof math error in the lighting console....

---

as far as the monitor loosing sync- that HAS to be a hardware issue...
nothing in any software would ever turn the video output from the entire computer off.

Spam Butterfly
20-03-2006, 03:04 PM
I'm not sure I've ever seen a Mac 'lose' a display under any conditions. However, it might be something else in the chain...? Perhaps some spurious RF like walkie talkies?

Hugh

samsc
20-03-2006, 11:59 PM
have you noticed the mac loose sync at any other time?

it is a new graphics card...

is it still doing this?

micpool@mac.com
21-03-2006, 10:09 PM
We have now completely eliminated the Catalyst system as the cause of this problem!

Sorry, it was just a series of unfortunate coincidences where every thing we did to reduce the load on the Catalyst systems seemed to directly correlate to an improvement in the time between glitch incidents and the fact that this is a show with very little heavy mains equipment or motors in a modern purpose built venue with no history of power supply or RF problems

We are now trying to work out whether the projectors or the Smart E Cat5 convertors are the problem, and the source of the interference.

Thanks for reponding so quickly and firmly but politely pointing us away from the Catalysts being the likely source of the problem.

Mic

samsc
22-03-2006, 07:00 AM
thanks mike.

it didnt have the profile of a computer or software problem.

i have worked in this field long enough to check for other types of things like interference and power issues.