It really depends on what you're doing to the footage, and what hardware it comes from. 1080>720 isn't an integer step so there will be pixel-blending on the resize, and with most prosumer kit having chroma subsampling you're not reading complete per-pixel data in the first place.
Even rendering to AVI I doubt anyone could look at an unedited 720p clip on a TV or Web video and tell if it came from a 720p capture or a scaled 1080p capture - in the former you'll have lens blur and chroma SS across pixels, in the latter you'll have fractional resampling giving near-enough the same feel, but if you apply effects that work on a per-pixel basis (such as unsharp masks or keying) to the footage before scaling they can look different. Not necessarily better or worse, just different; and your audience will only see one version so what you notice doesn't really matter!
It certainly makes sense to grab the biggest frames you can on location, as aside from any future use in full-HD projects you get the option to place it at larger than "fit" size, so you have leeway to crop stuff out or stabilise motion.
Prior to Mercury and solid-state drives, working in 1080 over 720 was a really noticeable drag - but now it's more about the codec than the frame size. A "tricky" codec at 480 can be more laggy to scrub than a "simple" codec at 1080, it depends on how much work your CPU has to do extracting the frame data before the MPE can scamper off with it.
14
Replies
AdChoices