Compressor output losing quality? QTX Deinterlacing?

[Zak Ray]

Anyone have any thoughts on this? Is there any reason not to convert my material to square-pixels if it’s never going out to broadcast, and likely just the web>

[Michael Gissing]

Square pixels for the web? yes. Deinterlace for the web? yes

I work in Pal so I make H264 files as 1024 x 576. If I need a smaller pixel size, I work in multiples of that ratio (usually exactly 1/2 so 512 x 288)

Will making a highly compressed resized version affect resolution? yes.

[Zak Ray]

I’m talking about the master here. Assuming I have a digital master that will never be broadcast, is there any reason not to convert the master to square pixels?

That’s all I’m doing– I’m not touching the interlacing or anything, since that would hobble the quality if I ever wanted to do something with it again. But square pixels seems like a reasonable thing to do to ensure uniformity and prevent misinterpretation by other apps. And I can always change it back, because it’s just stretching the PAR.

[Keith Pratt]

I’d keep it ProRes 720×480 anamorphic 16:9 through the edit and mastering, and only change it to 853×480 when making an H.264 for the web. 720×480 really is the standard for SD, and it’s what apps will by default expect. But for the web, you’re right, square pixels is the way to go.

Two things:
1: what is your source (DV camcorder, 16mm film scan, etc.) and frame rate?
2: check that Compressor has correctly identified the field dominance in the A/V Attributes tab.

[Zak Ray]

Good point.

Nearly all of it was acquired from NTSC DV, so I’ve been just telling Compressor to do lower field rather than even giving it a chance to interpret.

[Michael Gissing]

The master should remain anamorphic as this is a standard that all editing systems work best with and is the broadcast delivery format. Do the anamorphic to square pixel conversion when transcoding and deinterlacing for the web.