1080 --> 720 Unsharp mask settings?

[kanintesova]

Hmm, been thinking about it. I know there are no universal fixes.

But do some of you have "general" guidelines? I know that when I resize RAW's or jpegs's there is a loss of crispness/sharpness when i change/convert to different sizes. Is there a general rule to video as well?

I mean, with all the codecs. Let's say I render a 1080p HD no compression, then feed it through Sorenson for 720 output - does the conversion take in mind for sharpness? (never mind settings for bitrate, it's the "best" but just for the sake of 1080 to 720 downscale in pixels).

Or do I have to make this myself?

I understand there is no universal setting, but what if it was sharp at 1080 (original footage), do I still have to do it manually (again) at 720p (rendered) with the unsharp mask?

Would this be the best way to go? Not just letting the encoder do it? Does it really do it?

Edit: Yes I know there're probably some sharpness settings as "filters" etc. But I have not so much control I feel.

I speak Swedish, english not my mothertounge... Sorry

Reason I speak this topic is I see my sharp 1080 transfer to soft 720

But it was rendered 1080 and then converted in Sorenson (user preset) best for youtuve/vimeo.... No real funky artifacts, real good quality - but SOFT...

I guess my question is: If i want 720p broadcast coming from 1080p source; should I render to 720p and THEN do unsharp mask? Rather than render a 1080p project and encode for output at 720p?


Thanks!

 

[Demistate]

I would use Avisynth to do a lancos4 resize if you want to maintain maximum sharpness, or attempt to sharpen the footage after converting it to 720P.

Any NLE you use will default to a bilinear scale which will produce a soft result at best.

 

[kanintesova]

Thanks Demistate!

Maybe I should check out AVISynth a bit more. Seems like a real good d-encoder!

Do I get an on-screen-Display, preview with it, for tweaking?

But still; So, maybe, as you suggest - it's best to do the offline at 1080p and then - render 1080p uncompressed to 720p and then color-correcting/grading/sharpening ---> to good 720p for broadcast (web)?

What i mean is: If I am going to have final output with good result i should do post-work with same output... For sharpness? But would it also be true for, like say, 2K or above - when you have to downscale... (pixelwise)?

I understand about chroma sub sampling and that it's not the same... Just pixel/resoulution.