PDA

View Full Version : Content Formats and encoding.



Dan C
11-03-2004, 01:12 PM
Hi Folks,
I'd like to start a discussion on which is the best format to render content in for Catalyst. I'm hearing that Quicktime DV-Pal is the new prefered format, but I'm getting pixelation issues with this format. My prefered format is Motion Jpeg A as it seems to output the highest quality, but can become rather large if the bitrate is not kept low.

Can we all start a poll on favorite formats, why their favorites, and then standardise all attributes (if any) of this format.

I feel that agreeing on one format would save much confusion for both content creators and content/Catalyst users, and also simplify ordering content from non-Catalyst savy content producers. whew......

Dan

samsc
11-03-2004, 01:54 PM
Its DV-PAL or DV-NTSC

None of the other codecs will give you close to 3 or 4 layers simultaneously - unless you cut back the data-rate

None of the others seem to be optimised for multi-processors.

This may or may not change with changes in Quicktime.
You will have to do your own testing

Dan C
11-03-2004, 08:36 PM
Using Adobe Premier to render out quicktime DV-PAL produces a squares in squares in squares type of artifact and distorts the images signifigantly. Is this just premier or does anyone else see this?

Also.... what is the max data rate per layer for optimal performance over 4 layers on Cat 3? (Using Motion JPEG A as ref. if required....)

samsc
12-03-2004, 02:25 PM
Using Adobe Premier to render out quicktime DV-PAL produces a squares in squares in squares type of artifact and distorts the images signifigantly. Is this just premier or does anyone else see this?

Also.... what is the max data rate per layer for optimal performance over 4 layers on Cat 3? (Using Motion JPEG A as ref. if required....)

please post an image of what you are seeing,
also you can send me a few frames of your movie so that i can check that it is what you think it is.

samsc
15-03-2004, 09:16 PM
DV-PAL or DV-NTSC is three times quicker to decode than photojpeg.

DV-PAL has a compression ratio of 5:1

Spam Butterfly
19-03-2004, 12:12 PM
DV-PAL or DV-NTSC is three times quicker to decode than photojpeg.

DV-PAL has a compression ratio of 5:1

And a data rate of 3.6Mb/s.
The fixed compression rate of DV PAL/NTSC a pretty good option. However, it's YUV, has a colour sampling of 4:2:0 (for PAL) and 4:1:1 (for NTSC). It's also a fixed SD size - 720 x 576 for PAL and 720 x 480 for NTSC.

Photo-Jpeg colour is better than DV and it has variable compression settings. It's a great off-line format, and content created in this codec looks very good if it's at 100% quality. Check out the Digital Juice footage we give out with Catalyst. However, it's pretty CPU intensive to decode it and it has a variable data-rate - certain 'busy' frames can take more cpu effort than others. I've heard of people using this in Medium Quality setting - which I think looks shit...

Motion-Jpeg A medium quality works pretty well with Catalyst... It has similar issues to photo-jpeg in the variable data rate stakes.

peppe
05-04-2004, 06:41 PM
....Motion-Jpeg A medium quality works pretty well with Catalyst... It has similar issues to photo-jpeg in the variable data rate stakes.

Often you have to use MJPEG-A as everybody seems to always need the maximum of layers.With that codec you anyway can make bigger content then 720x576 (pal) when zooming in the clips.

I agree that medium quality is work pretty well, but when doing contents with gradients,you see the actual quality and that is not always what you want.
I prefer to do the "gradient stuff" in Photo-JPEG high quality when using it as max 2 layers and the rest in MJPEG-A medium quality.

samsc
05-04-2004, 10:53 PM
Often you have to use MJPEG-A as everybody seems to always need the maximum of layers.With that codec you anyway can make bigger content then 720x576 (pal) when zooming in the clips.

I agree that medium quality is work pretty well, but when doing contents with gradients,you see the actual quality and that is not always what you want.
I prefer to do the "gradient stuff" in Photo-JPEG high quality when using it as max 2 layers and the rest in MJPEG-A medium quality.

you are right. you have to use mpjeg a or photojpeg when using larger or smaller movies than 720x576 or 720x480 - or odd aspect ratios or images like 1000x230 etc.

photojpeg preserves colour information better - but it is much slower than dv. and you will not be able to playback 4 layers.