[Dennis Couzin]

Thanks John and Rafael,
My question naively lumped the effects under the Motion tab with filters. Sorry.

The Filters tab does work rationally. Filters can be listed in any order and they are applied in that order.
But under the Motion tab, while it looks like rotation comes before crop, it doesn’t. The Motion tab is special.

OK, using Rafael’s suggestion I can get rotation before crop by using Basic 3D in the Filters tab for rotation; Crop in the Motion tab for crop.

But what if someone really did want crop before rotation. This is what came from using the Motion tab for both. However Rafael warns against using Rotation in the Motion tab for quality reasons.
Luckily it is possible to do it all with Filters. Fake crop each edge by applying the Basic 3D filter twice (the first time moving the center P pixels and the second time -P pixels), and then apply Basic 3D another time for the rotation.

Is every effect under the Motion tab also achievable with some filter? If not, what if I want that effect to be be done before some filter?

Rafael’s warning about the quality problem when using Rotation in the Motion tab is a shocker. Does the FCP User Guide warn us? Is there no ban on landmines for software?

[Dennis Couzin]

Sorry for the lame example of the mirrored shot. A better example would be a cut to slow motion or a cut to the same shot with something inserted into it.
Years ago, low budget films would shift quality at the beginning and end of every effect. To save money, they optically printed just the effect, not the whole shot. Recompression today is not as lossy as optical printing was then, but I think “Recompress All Frames” is to solve that same problem.
Incidentally, recompression of a compressed video doesn’t necessarily change it at all. It depends on the type of compression. There’s a fun experiment for testing what decode/code does for a particular codec. Make a QT with the “Recompress All Frames” option. Import it to the timeline and make another QT with the “Recompress All Frames” option. After several iterations, do you see a difference with the original?

[Dennis Couzin]

Tom, doesn’t your explanation imply that Apple used the wrong term? Rerendering all frames doesn’t mean recompressing all frames.
I thought Apple offered the “Recompress all frames” option in order make the QT export more uniform in quality. For example, suppose I capture DV and set the timeline as DV. Suppose the edit goes from from a shot to the same shot mirrored left to right. The second shot required FCP rendering, so it will be recompressed DV in the export. There will be an appearance difference between the consecutive shots unless “Recompress all frames” is chosen.

[Dennis Couzin]

Thanks Sean. I posted all my specs except the disk because three days ago I didn’t have the slightest suspicion that disk read time might be the limiting factor. (Common knowledge isn’t universal knowledge.) So, Samsung’s 175 MB/s figure is untrue. Can we trust the HD103UJ for at least 50 MB/s? Is software RAID 0 efficient? Can I stick two HD103UJ’s into the MacPro, joined in RAID 0 with Disk Utility, to get 100 MB/s? That would be a cheap and sufficient solution for me for now.

[Dennis Couzin]

Alexander, doesn’t that depend on the codecs?
I’m starting with DV-PAL, so the conversion to 8-bit uncompressed 4:2:2 should be pretty nearly harmless. Doesn’t Compressor have to decompress the DV-PAL anyhow? Compressor can’t apply its frame controls to the DV-PAL directly. There are no pixels to play with until the video is uncompresed.
There might be small loss involved with the color subsampling. I don’t know how FCP is converting the 4:2:0 to 4:2:2. I suspect that Compressor needs 4:4:4 to best apply its “flow technology”. So Compressor has to transform the 4:2:2 to 4:4:4. Maybe the FCP conversion from 4:2:0 to 4:2:2 followed by the Compressor conversion from 4:2:2 to 4:4:4 isn’t exactly the same as Compressor’s conversion from 4:2:0 to 4:4:4. Or maybe it is.

We are losing track of the original question. It is whether Compressor renders some effects better than FCP does, so that any rendered output from FCP, in any codec, is less desirable than letting Compressor render those effects.

[Dennis Couzin]

Sean, am I misreading the Samsung HD103UJ spec : “Data Transfer Rate / Media to/from Buffer(Max.) 175 MB/sec”?
I don’t think bubble memory has as fast read/write as good hard disk drives.

[Dennis Couzin]

“The clips will run a 4% faster, which is unnoticeable”
This is generally true for the picture speed but it’s debatable for the sound speed. Sing a scale. 4% speedup is 2/3 of the way from ‘te’ to ‘do’. A fraction of the population has near enough to absolute pitch to notice this error. Especially if this is an ethnographic film, why instill bias?
You can use Cinema Tools to conform the 24 fps picture to 25 fps, but you should use sound resampling software to actually convert the 24 fps sound to 25 fps without pitch changes.
Isn’t Alex Elkins’s suggestion simpler?

[Dennis Couzin]

Alexander, I partly agree. The Nattress flow chart shows what can happen when FCP renders its own effects. The native footage is first uncompressed, then effected, then compressed to native codec. This recompression usually adds artifacts. Using the “Export QuickTime Movie” command can make such a recompressed file. So feeding that QuickTime file to Compressor is less good than exporting straight from the FCP timeline to Compressor.

The “Export QuickTime Movie” command has an option “Recompress All Frames.” When this option is deselected, (I think) only the frames involved in effects — the ones that needed to be uncompressed from native — will get compressed to native. So these are the frames that are disadvantaged. They might be the whole film, or just a few frames.

When I use the “Export QuickTime Movie” command, I first change my sequence settings to an uncompressed codec. Then the exported QuickTime file has not seen recompression, or just a tad. The original question remains: is feeding that QuickTime file to Compressor less good than exporting straight from the FCP timeline to Compressor?

[Dennis Couzin]

Gary, we’re not anonymous here, so curiosities about who’s accomplished what are easily satisfied by googling, and needn’t clutter the discussion.

[Dennis Couzin]

Thanks all. I’d never considered the hard drive. I transferred the file to a newer internal drive and the playback became a little better. It’s spec’d at 175 MB/sec, but 100 MB/sec is too much to expect. Yes, must RAID.