Pierre Jasmin
75
COW Coins
2
Premiere Pro Coins
2
VEGAS Pro Coins
18
RE:Vision Effects Coins
Forum Replies Created
-
Meanwhile back at techsupport….
All fine now?
Pierre
-
Meanwhile back at techsupport….
All fine now?
Pierre
-
When you run the installer it asks you where you want to put the manual.
There is also some introduction video tutorials https://help.revisionfx.com/album/20/
and these are at well here at the cow in the tutorials section.
-
When you run the installer it asks you where you want to put the manual.
There is also some introduction video tutorials https://help.revisionfx.com/album/20/
and these are at well here at the cow in the tutorials section.
-
The distance of a ping pong ball in motion from a frame to the other might be too large for an automatic process.
We would need to see a short sequence – techsupport at revisionfx dot comAlso I am not clear if you looked at the tutorials Lori did for Twixtor in the Tutorials section here at the cow.
Pierre
RE:Vision Effects -
The distance of a ping pong ball in motion from a frame to the other might be too large for an automatic process.
We would need to see a short sequence – techsupport at revisionfx dot comAlso I am not clear if you looked at the tutorials Lori did for Twixtor in the Tutorials section here at the cow.
Pierre
RE:Vision Effects -
Just arrived here via an unrelated search… some corrections here:
One reason is the version of aerender you are using is probably not implementing properly -MP flag (aerender works like MP off for me in render queue in CS5.0, bug filed).
If you send N aerender each working on a segment (e.g. frame 1-100 for first, 101-200 for second…), you should see about the same render timing as MultiProcessing in render queue. On 12 CPU to give you an example sending 12 aerenders at once is 3-4 times faster then one. That is each frame renders 3-4 slower but you have 12 times less frames per render instances.
Since compositing by definition implies a lot of images in, the main issue with multi-processing is you become quickly IO bounded – that is most of the time is spent waiting for images (as you now have 24 processes instead of 1 reading from the same disk, the bottleneck is not necessarily processing power). You sort of need to scale the hard disk bandwidth (e.g. TBs raid) to match all that CPU. Also it quickly gets worse if all these renders at once move the data over the network.
That said, I would say if a machine is also used interactively then always go for less but faster CPU. For a render-only monster, some attention required at the IO level for this to scale properly. The ideal scenario for a multi-user setup could be (and AE might not be totally there yet) to have the app cache frames requested to an SSD when in the interactive app and maintain the media assets on a big/fast RAID attached to the render monster.
Pierre
https://www.revisionfx.com -
There is a mode in the Deinterlacer called Motion Estimated Blend which will do what you want. Just drop your footage, apply that with the default (making sure the field order menu is correct). No need for Twixtor. The bit you read about using FieldsKit before Twixtor is for certain retiming cases problems.
Pierre
-
There is a mode in the Deinterlacer called Motion Estimated Blend which will do what you want. Just drop your footage, apply that with the default (making sure the field order menu is correct). No need for Twixtor. The bit you read about using FieldsKit before Twixtor is for certain retiming cases problems.
Pierre
-
Would be worth to send to a short section (to techsupport at revisionfx dot com) to see if anything is possible. I might be totally off here as I am not seeing the video but maybe there is a way to key out the sparkles to do it in two passes or something?
Pierre