1080p and 720p playback

[samsc]

Can get full speed with 4 drive raptor array using a variety of codecs, Pixlet at 86, DVCPRO50 comps in AE at 1280x720 best quality, even photo Jpeg.

4 drives make a huge difference.

CC

what data rate did you see on your pixlet 86 movie?
And your photo jpeg files?

R

[litemover]

what data rate did you see on your pixlet 86 movie?
And your photo jpeg files?

R

Variable, I have many of them now, they range from 2.6 to 6meg/sec. Most of the ones that I have xtracted with the Heuris extractor from my JVC jyhd10u and encoded with quicktime using pixlet 86 range from 5.5 to 6, some of the HD artbeats clips that I have that were encoded with cleaner are much less 2meg to 3meg. For some strange reason, I cannot use cleaner to convert the MPG movies that Heuris extracts from the M2 files in the camera, all for HD but the results are like watching an I-max Movie when spread across 3 large screens and xy'ed to look as one seemless image.

Some play full speed all the time some fluctuate between full speed down to 28 to 21 to 18 to 31. The Data rates are variable, should I clean them to be non variable?

Here is an image of 1 movie from my JVC HD cam spread across 3 screens to give the illusion of 1 screen
Christian

[samsc]

nice image christian

[litemover]

nice image christian

Thanks You.

CC

[samsc]

Variable, I have many of them now, they range from 2.6 to 6meg/sec. Most of the ones that I have xtracted with the Heuris extractor from my JVC jyhd10u and encoded with quicktime using pixlet 86 range from 5.5 to 6,

try encoding with photojpeg.
its at least twice as fast as pixlet to decode.
and i cant spot the difference.
you will need to try encoding settings between 50% and 75%.
you will get more stable playback speeds- i think.

the stability of the playback speed with these codecs is well away from something i can control. its buried deep in quicktime.

[litemover]

try encoding with photojpeg.
its at least twice as fast as pixlet to decode.
and i cant spot the difference.
you will need to try encoding settings between 50% and 75%.
you will get more stable playback speeds- i think.

I would have though DANNY has told me not to and that photojpeg is not optimized for dual proccessors. He tols me that it would be slower. I'm going to start using photo Jpeg now, it looks alot nicer and I can work with it in RT while editing.

How bout the blackmagic codec Richard? I'd love to start using that codec as it is extremely fast to encode and great looking?

Best,
Christian Choi

[samsc]

I would have though DANNY has told me not to and that photojpeg is not optimized for dual proccessors. He tols me that it would be slower. I'm going to start using photo Jpeg now, it looks alot nicer and I can work with it in RT while editing.

How bout the blackmagic codec Richard? I'd love to start using that codec as it is extremely fast to encode and great looking?

Best,
Christian Choi

Photojpeg is slower than dv codecs at the same resolution and compression rate - by about twice-
DV doesnt work at any other resolution than 720x576 or 720x480. So you cannot use dv at 720p or 1080p.
You need to compare codecs that will do this....

this doesnt matter nearly so much when you are only playing one or 2 layers.
( you cannot playback 4 layers of photojpeg content. - it simply takes too long to decompress the frames.
If you need 4 layers to playback at 25/30fps you have NO choice but to use DV-NTSC (30fps) - there are no other options. Nothing else works. )

-

but at 1080p resolution photojpeg is 2 1/2 times faster than pixlet.
It takes 20milliseconds to decode one photojpeg 50% frame at 1080p res - 50ms for Pixlet 50%.
And at 720p resolution is takes 10ms for a photojpeg frame at 50%, and 20ms for a pixlet frame.

This chart shows single layer decompress times;
http://chaldee1.gotadsl.co.uk/~richa...ttachmentid=22

-

Blackmagic is just an uncompressed codec (YUV encoded ), nothing special- means huge files. And doing any kind of uncompressed work is really not noticeable in this environment- and its 60GB/hr - so you dont get much .

So this is the file to use if you have an 8 disc scsi array and need to do 1080p uncompressed. Data rate ~100-150MB/s!

-

Whether or not photojpeg is optimised for multiprocessors- is speculation from me based on reading between the lines of quicktime developer's documentation.
In this documentation - they talk about specific changes to the DV codec to optimise it for dual processors- they dont talk about other codecs like this.
DV -is the core codec for final cut and editing.
So it would make sense to use development resources to make this work as well as possible.

This document is "Whats new in QT 6.5". I dont think this information is in here - i think its in an earlier revision:
http://developer.apple.com/documenta...6_5/index.html
This is the form of the smoke signals one receives from apple.

[litemover]

but at 1080p resolution photojpeg is 2 1/2 times faster than pixlet.
It takes 20milliseconds to decode one photojpeg 50% frame at 1080p res - 50ms for Pixlet 50%.
And at 720p resolution is takes 10ms for a photojpeg frame at 50%, and 20ms for a pixlet frame.

Looks like I need to rerender and/or recapture and re-edit all my 720p content. Pixlet looks grainy to me so it will probably remain grainy even if I run it though cleaner. damn!

That bytes.
CC

[samsc]

Looks like I need to rerender and/or recapture and re-edit all my 720p content. Pixlet looks grainy to me so it will probably remain grainy even if I run it though cleaner. damn!

That bytes.
CC

Im going to try and do some testing in 5% increments from photo-jpeg 50% to 100% to see how the decode time changes.

You will have to do your own testing to determine the subjective factors.
Whether something looks more or less 'grainy' - is a subjective problem.

You will have to determine the point at which something is or is not acceptable for the usage you intend.

My testing of conversion tools seems to indicate that Compressor is a more reliable to way convert between codecs than Cleaner.
Cleaner seems to fuck with things more.
I saw no colour differences between any of the images when i rendered with 68 codec types using Compressor.

[samsc]

Looks like I need to rerender and/or recapture and re-edit all my 720p content. Pixlet looks grainy to me so it will probably remain grainy even if I run it though cleaner. damn!

That bytes.
CC

The thing is christian --- you might not see any visible difference if you stretch 720x480 up to 720p.

You should test this too.

There are quite a few mirages hidden away in all this stuff.
And the closer you look the more problematic it looks.
Same with all types of film-making.