Standard definition image quality
 

At the top of page 102 of the NX5 operations manual it state that the SD image resolution is 640 x 360 dots/16:9 and 640 x 480 dots/4:3.

Then I saw near the bottom of the page that the SD resolution is 720 x 480/16:9, 4:3.

So is the 640 x 360 a missprint or is there some meaning I'm missing in the term "dots"?

I'd missed this in the manual, but then I've yet to imagine myself using the SD capability.

720x480 will be the actual resolution of the video files produced. 640x360/640x480 will be the effective resolution from which the 720x480 is derived.

...so if you're really after a nice-looking SD image from one of these cameras, my inclination would be to shoot in HD progressive and scale down in post. I shoot almost exclusively in 720p60 because, in addition to looking great on Blu-ray, Vimeo, etc., it makes the jump to 480i60 very cleanly.

Best,
Aaron

My PAL NX5 gives SD as 720 x 576 @ 9mbps as I would expect. But what's the point now that dropouts are a thing of the past? Like Aaron says, simply shoot at the highest res going and down-convert at leisure.

BTW Aaron, I'm shooting 1920x1080 50i - are you saying that SD downconverts look better from material originated at 720 50p?

tom.

Not Aaron, but yes, it is a bit better because you've got a full progressive frame to work with when downconverting to an SD interlaced field.

I tried this out when shooting trial runs of a rehearsal for a dance recital. The 720/60p was smoother and clearer than the 1080/60i when downconverted and printed to DVD. Not a dramatic difference, but an improvement when I looked closely.

I shot the actual performance itself in 1080/60i because that is what all my other cams shoot.

It looks fair bit cleaner by my eye, but I've never objected to the quality of an SD downconversion from, e.g., my CX12 (which is 1080i60 only).

Conceptually, converting a progressive frame to interlaced is just a way simpler process. Converting between interlaced formats is something one would expect most modern NLEs to do a good job of, but when you start to think about the differences in field order between source and target, preservation of the vertical displacement between the two fields, etc., it's easy to think of ways the process could produce a suboptimal result.

EDIT: (Of course, in this specific case, the 720p also gives you quite a few more lines--a third more--to deal with when scaling to the target field size, so that obviously helps!)

Best,
Aaron