[Craig Seeman]

[Rich Rubasch] “The true 4 x 3 equivalent to 720 x 480 is in fact 720 x 540.”

There’s much discussion as to what “true” is when it comes to aspect ratio. I’ll point to this well researched article for those who don’t mind video engineering math.

https://lipas.uwasa.fi/~f76998/video/conversion/
One might look at this chart and the “notes” section too.
https://lipas.uwasa.fi/~f76998/video/conversion/#conversion_table
also this
https://www.lurkertech.com/lg/pixelaspect.html

[Craig Seeman]

Try 512×384 (ideal), 528×396, 544×408 if you can go over a bit (and maybe crop).

Rich pointed at one way to do the math. Here’s another:
720×480 non square is also 640×480 square pixels.
You can think of moving up or down frame size by 32×24.

Because of the way most codecs work when they encode in blocks, keeping things divisible by 16 evenly is a good rule. At the very least the number should be divisible by 4 or 8.

Here’s an example:
4×3 (odd number not good)
8×6 (6 is not evenly divisible by 8 or 4)
16×12 (12 is divisible by 4 but not 8 or 16)
32×24 (it’s 4×3 ratio and both numbers are divisible by 8, that’s good). So this is a pretty good base unit of measure.
64×48 (this is ideal since both numbers are divisible by 16 evenly).

Keeping these ratios in mind, divisible by 16 best, 8 good, 4 ok, you’ll find that:
512×384 is divisible by 16.
544×408 is divisible by 8.
528×396 is divisible by 8.

540×400 is close to 544×408 so the distortion might be too small for your eye to notice. You could go with that and crop a few pixels so it doesn’t distort instead.

BTW, you might see some sites use
240×180 (divisible by 8) for their smallest frame size and others use 256×192 (divisible by 16). 320×240 is divisible by 16.

For those who want to look at this another way. With 720×480 non square which is 640×480 square, you can get target frame size by:
64 by 48 multiplied by n (best frame sizes)
32 by 24 multiplied by n (best and good frame sizes)
16 by 12 multiplied by n (best, good and “passable” frame sizes)

I hope I haven’t muddied the waters too much but I thought the above might explain some ways to figure out frame sizes that don’t distort and work well for encoders.

[Craig Seeman]

[Ben Waggoner] “You’ll want to make sure that you have Motion Estimation set to Best as well.”

Sounds like a Compressor setting.
In Episode, in Frame Rate, Algorithm/Motion Compensated.

If you’re deinterlacing –
Create New Fields (avoid Duplication unless encode speed is critical)
The choices between Interpolation, Blending, Edge Detecting (and Heavy) depend on content.
Blending give smoother motion but less sharp than Interpolation. Edge Detecting Interpolation can give you more distinct diagonals.

[Craig Seeman]

Changing the sequence setting works.
If you’re handed a file that you must convert, you can change the sample rate in QuicktimePro too (export to DV.mov 44.1 instead of DV.mov 48.1) and then export the resultant file to WMV. It’s a bit of a pain but they are working on a fix (I’m beta testing it).

[Craig Seeman]

You can compress WMV in Squeeze (and even Compressor) with Flip4Mac. Squeeze even includes a very basic version (just a few unmodifiable presets) free.

BUT, as Rich says, Episode is major flexible and with the new Compressor plugin you can even use Compressor’s processing filters if you so choose as well as qmaster.

[Craig Seeman]

FCP uses Quicktime as its “base.” If you open an FCP DV ref movie (or self contained) you’ll see the interlacing and the video pixel aspect ratio.

As to what’s happening in Squeeze it may very well depend on what settings you’re using. Squeeze doesn’t have the most flexible processing filters but you’d need to post the details, settings, preprocessing (if any) you’re using. Also keep in mind that DV isn’t the best codec to work in if you’re expecting “crisp” (or even smooth) graphics.

[Craig Seeman]

And keep this nifty thought in mind. If there’s anything you like about Compressor 3, you can use the Episode/Compressor plugin which comes with Episode.

You might try doing a 2 Pass encode (with a much longer encode time) and maybe even scene change detection to force some more keyframes (I-frames) beyond that of the GOP structure.

[Craig Seeman]

What Danny2007 said. If you’re not doing streaming you might want to consider 2 Pass VBR for a more efficient file size giving the codec room to “breath” (increase bit rate) for the fast motion parts. You may also need to look at deinterlace settings if your source is not progressive.

You’ll get “natural” key frames on edit points generally but the farther you get aways from an edit point/natural key frame, with fast motion, the more likely you’ll want/need “forced” key frames. With “long” (subjective) stretches of fast motion you’ll want more “forced” key frames.

[Craig Seeman]

Check this Apple KB article on this issue

https://docs.info.apple.com/article.html?artnum=93234

[Craig Seeman]

After exporting the Reference. You can open in Compressor and make the various codec you need including WMV with Flip4Mac which does work inside Compressor.