32 bit floating point and blu ray

[Bill LiPera]

Recently made a blu ray disc from the vegas timeline with 32 bit floating point and there were multiple areas of jitter, however when I remade the blu ray disc with the 8 bit floating point setting there was no jitter. Any ideas? When should 32 bit floating point be used? Thanks.

[D.J. Ammons]

Interesting. I am curious as to what responses you will get. I thought I had read that you should do your edit in 8 bit but render in 32 bit for best quality.

What exactly do you mean when you say you "made" the video with the two different bit qualities? Do you mean you rendered it to Blu-ray with both settings and the 8 bit render was better?

[Phil Lee]

Hi

Originally Posted by Bill LiPera

Recently made a blu ray disc from the vegas timeline with 32 bit floating point and there were multiple areas of jitter, however when I remade the blu ray disc with the 8 bit floating point setting there was no jitter. Any ideas? When should 32 bit floating point be used? Thanks.

Sounds like a bug in Vegas. Blu-ray is always 8 bit video files, as is most everything else. The 32 bit floating point option relates to how Vegas processes the files internally. For simple edits with no or very few filters and effects you don't see much advantage, but if you do a lot of colour grading and have lots of effects then 32 bit mode prevents rounding errors creeping in, the output is still 8 bit.

Regards

Phil

[Bill LiPera]

Does that mean one should not render out to blu ray or dvd with the 32 bit setting?

[Rob Wood]

if you're just rendering an mpeg file, but don't want to lose the benefits of 32-bit

1) render a master using 32-bit; lossless preferably
2) then create a new project at 8-bit > import the lossless master > render to blu-ray

step one guarantees your filters, transitions, etc are rendered as cleanly as possible.
step two doesn't require greater than 8-bit because it's just a raw conversion: 8-bit lossless to 8-bit lossy... there's no filters, no adjustments, nothing that requires the headroom 32-bit provides.

[Phil Lee]

Hi

There is no way to render 32bit lossless, that codec/option doesn't exist. The best you could do is render 10bit.

The 32bit setting is purely for the processing internally on the time line so:

8bit timeline
Compressed 8bit video decompressed to 8bit uncompressed video
Effects/filters applied
8bit uncompressed video compressed to your chosen output format

32bit timeline
Compressed 8bit video decompressed to 8bit uncompressed
8bit uncompressed video converted into a 32bit floating point values
Effects/filters applied using 32bit floating maths
32bit uncompressed video converted back to 8bit uncompressed
8bit uncompressed video compressed to your chosen output format

The extra conversion between 8bit and 32bit and back again can introduce errors and increases render time, so only worth doing if you have a lot of colour/luminance adjustments and effects on clips that can be processed with less clipping and losses using the wider range of samples in 32bit processing mode.

Also note 32bit can cause level shift issues. See this link for info Color spaces and levels in Sony Vegas 9 and 10

32bit doesn't change the output in anyway, it is purely an internal change to how video is processed, the actual "bit" part of the output is unchanged.

Regards

Phil

[Tom Roper]

I saw the jitters in the 32bit internal render that Bill noted in the first post. No jitter in the 8 bit render.