Shake And Hdv

[user_deleted]

QUOTE(Joojaa @ 01/05/06, 04:19 AM)
Again like i said for shake theres no difference wethere or not the source is a sequence or qt. The gui might think so but there is not, if anything ist a oversight of the developpers.
[snapback]224032[/snapback]

Actually, the documentation claims that there are in fact explicit differences between Shake's capabilities with QT movies and sequences. (Though the manual is obviously oriented towards the GUI.) Evidently the developers have been at work deprecating this part of the manual, and I'm glad to hear it!

(I'm uncomfortable with sequences, not because I think they're somehow different in use, but because I've had bad experiences with huge directories and because there are many non-Shake tools that don't understand sequences. On the other hand, it is pretty simple to change a bad frame, etc, with a sequence and network renders obviously can't do anything but sequences -- at least as the intermediate result.)

P.S. The Quad G5 and Shake have arrived, and I hope to have them out of IT's hands by the end of the day! The HDV cam/deck should be here tomorrow. I'll be able to close my (demo) Shake session and move on up to the save-it and reuse-it world!

[joojaa]

QUOTE(Wayne_ @ 01/05/06, 06:38 PM)
Actually, the documentation claims that there are in fact explicit differences between Shake's capabilities with QT movies and sequences.
[snapback]224060[/snapback]

And that would mean theres somekind of difference in how pixels are presented to shake! Wich sounds like some pretty stupid design.

Imena frame 1 is ftrame 1 and looks like frame one no metter how its fetched. And lets face it fore sake to work right it must be able to fetch both at will.

[user_deleted]

QUOTE(Joojaa @ 01/05/06, 12:42 PM)
Imena frame 1 is ftrame 1 and looks like frame one no metter how its fetched. And lets face it fore sake to work right it must be able to fetch both at will.
[snapback]224079[/snapback]

True in theory. (And I'd have designed it that way, but I tend to over-engineer.) But remember that Shake is much older than MacOS with its convenient Quicktime system and there are evidently still assumptions from Shake 1.0 stuck in at least one location in the GUI, if not elsewhere. The good news is that from this conversation it appears that the limitations spelled out in the manual are only applicable in one location in the GUI and can otherwise be worked around.

In terms of a-frame-is-a-frame, the fetched frame will be the same. But the fetching process for a movie can be MUCH more involved and time-consuming than for a sequence, since movies can feature both intra-frame compression (JPEG, some TIFF, etc) AND inter-frame compression (MPEG-2, etc) while sequences will only ever have intra-frame compression. So retrieving a single HDV frame could involve reading 3 or 4 frames (I think) to reconstruct the one you want.

[joojaa]

QUOTE(Wayne_ @ 01/06/06, 08:15 PM)
True in theory.
[snapback]224158[/snapback]

It must be otherwise shake couldnt cope with the qts internaly, and the whole guis result computation would not work at all (sice you can go back and forth in the gui its clear shake can acces individual frames of the qt, wich is enough for thiskind of stuff). Whatever apple ads ontop is a nother thing.

Anyway about the compression thing thats what i said inpost13, of this thread.

[user_deleted]

I`ve just finished two jobs in shake.
On music video in native HDV and a presentastion movie i Apple intermediat codec.
I had big problems with the sync in the native HDV format, when I reencoded it as Apple Intermediat, everything worked fine.
An din the postpro We did a 25fps to 24fps and filout to apple intermediat 25fps (yeah thats right back to 25fps). And converted from interlaced to progressiv. Just to get a more filmfeel. OK the render took 2 days on our dual 2,7 G5, but it made the little difference we wanted.

The worst thing with the HDV format (SONY HDVFX) is the anamorphic resolution 1440*1080 scooped til 1920*1080. On the Presentation movie the non square pixels was not fun to work with.
Og course we used th DVblur FX (downloaded frome her) to make the keying a little easier.
but we had to make a macro to divide every nonsquare pixel. Just to have enough pixels on the edges og the alpha. We used Shake for composting, keying and rotoshaping. And used the multiplane node to compost som keyed video over a 2d image (tracked in PFtrack) to get a moore 3d look. but the multiplane is a little buggy.

OK a little story here.
But
I recommend tha apple intermediat codec instead of the HDV. OK it takes a little more space, but Is space really a probleme today?

Trond-Atle Johansen, Norway

link to the music video: http://www.myspace.com/palekids (compressed by youtube)

[user_deleted]

QUOTE(trondajo @ 01/10/06, 02:13 PM)
I recommend tha apple intermediat codec instead of the HDV. OK it takes a little more space, but Is space really a probleme today?

i dont know, but i find the AIC to be pretty crappy.. it's goot for a quick and dirty edit in iMovie but i wouldnt want to use it in shake, specially not on anything that goes out to film in the end.. how was your final project presented?

i'm also not quite sure why (and how) you did a 25fps->24fps->25fps conversion.. if you just reassigned the frames differently, there wont be any difference at all, and if you retimed the clips, you lost quite a bit of quality.

but two things that are related to your post:

does anybody know how to get out the NATIVE 1440x1080 of a HDV clip? (or the 1280x1080 of a DVCPROHD for that matter?)
the best i could find so far is squeeze them back in shake in the filein, but there must be a way to tell quicktime (or some other app) to read them as square pixels in the first place, no?

also, i found that the 4XX enhancer works great on DV footage (better than the DVblur imo), but not so well for HDV. anybody found any presets for that (and DVCPROHD)?

bests
++ chris

[user_deleted]

Mayby AIC is crap for film, but it look great for my project. The presentation will be viewed on a 42" HD plasma, from a mac. And on a website i 1080p, 720p and some other formates.
The musicvideo will be showed on norwegian telivison in ordinary PALformat.
This is my first to projects in HD. So mayby the AIC codec is bad, My customers was very satisfied with the result (and if my customer is, then I am )

And now for the wierd 25->24->25 converison.
I used the retime fuction i the Filein Node, input to 25 and output to film 24 (and interlaced to progressive). And actually by mistake the Fileout node were 25fps and AIC. After eksporting I actually tought it were 24fps, but viewed it in QT. and it were 25fps. I opened the original AIC output from FCP and viewed them simultaniossly, and there is a little differens. The frames are not identically. ut when I view it in realtime it got that little somthing you might call the X-factor.
The quality has not changed. I

[user_deleted]

QUOTE(trondajo @ 01/10/06, 06:21 PM)
Mayby AIC is crap for film, but it look great for my project. The presentation will be viewed on a 42" HD plasma,  from a mac. And on a website i 1080p, 720p and some other formates.
The musicvideo will be showed on norwegian telivison in ordinary PALformat.

well, since you have to compress it anyway to play it of a single disk (not to speak of the web... and PAL is much lower res anyway), AIC is probably not such a bad idea as a final intermediate format.
however, for best results, i'd convert all the HDV clips to an lossless image sequence (png or iff), then do all your effects and compositing, then render out as an lossless image sequence again as your master. if you're tight on storage, consider jpeg compression. (you could olso choose QT files with png or jpeg compression if you dont like image sequences).

to play it from the mac, i'd compress it to a QT file with jpeg or mpeg4 compression, for web transmission either mpeg4 or h264 (the later uses a lot of processor power though). for the PAL transmission, render it out to an uncompressed 4:2:2 QT file and take it to a post house to play it out to digibeta.

QUOTE
And now for the wierd 25->24->25 converison.
I used the retime fuction i the Filein Node, input to 25 and output to film 24 (and interlaced to progressive). And actually by mistake the Fileout node were 25fps and AIC. After eksporting I actually tought it were 24fps, but viewed it in QT. and it were 25fps. I opened the original AIC output from FCP and viewed them simultaniossly, and there is a little differens. The frames are not identically. ut when I view it in realtime it got that little somthing you might call the X-factor.
The quality has not changed. I