Well after some testing it is now being suggested that if we make our graphics imports to 10:1 instead of 1:1 SD Avid mxf it will speed up our system. For three years we've always imported graphics and stills at 1:1 so we don't have to reimport for online etc... Now that we had our unity issues I've mentioned and sorted those out with a new media net unity connected with fibre instead of ethernet it is being suggested that the sluggish performance we had on very nested and complex sequences will be improved dramatically by importing our graphics at 10:1 instead of how we have always done it 1:1 SD. Perhaps this ties into the previous suggestion that we not render at 1:1 but render to 10:1. Now I'm sure our 3.1.1 ish media composers are rendering to same as source so I would think 10:1 media would render to 10:1 but the graphics at 1:1 would render at 1:1. I'm of the opinion the sluggish behavior was primarily due to complex nested sequences with many Pan and Zoom unrendered stills. One of the problem sequences would not save in a bin without breaking it in half. Once I had finished color correction and rendered all the stills in Pan and Zoom individually not a top down render like offline does I could save the whole sequence. This leads me to believe it is something about all the pointers and I believe the terms is objects that is the main problem here. All but the complex shows have worked fine. I checked bin info and saw an item count but nothing about object count. Is object the correct term and if so how does one asses that. I just can't buy that 1:1 graphic imports are drastically slowing down our system, especially after 3 years of relative stable performance except on really complex sequences which I see in all the facilities I work in. Anybody know how to check object count?
--- In Avid-L2@yahoogroups.com, John Moore <bigfish@...> wrote:
>
> Okay after all the issues we had switching over from an older unity with the separate chassis to a media net 5 with fibre connections instead of ethernet all is working. Of course are heavily layered sequences with 19 track nests are still slow and kludgey. This is just the nature of big layers nested in big layer. We've got someone claiming that the sluggishness is do to the fact that we render to 1:1 instead of 10:1. This makes no sense to anyone but this individual who has proven to be less than accurate in the past. In fact greater compression is often more taxing on the CPU. Of course 1:1 SD that we online in is more data but less processing. The resolution of the renders has little or nothing to do with sluggish behavior on a complex sequence. This whole thing sounds like BS from the ground up. I would love to hear others opinions on the statement that, "Sluggish sequence behavior on a complex sequence is a result of rendering to
> 1:1 instead of 10:1." The sequences are cut from multigroups that are 10:1 and the graphics are 1:1 all are SD. Looking for conformation or denial that this has anything to do with it.
>
> John Moore
>
> Barking Trout Productions
>
> Studio City, CA
>
> bigfish@...
>
> [Non-text portions of this message have been removed]
>
No comments:
Post a Comment