Quote:
Originally Posted by nothru22
If you send that TV 1080i then the DVD signal will go 480i -> 480p -> 1080i -> 1080p as it's commonly believed that the signal is deinterlaced first before scaled to 1080i. It's good to send the TV it's native resolution, but i -> p -> i -> p can't be good. |
Obviously the solution in this case would be to go 480i->480p->1080p (missing out the 1080i step)
However as 1080p output isn't available, I think that if you want to convert 480i to 1080i you have to go via progressive for best quality.
If you wish to convert a 480i image to 1080i you have two options AIUI.
1. De-interlace 480i to 480p, scale to 1080p and interlace.
2. Treat the 480i as 240p and the 1080i as 540p and scale the fields as if they were frames. This means that the two fields are treated entirely separately, and is not a recipe for high-quality, as fine static detail won't be interpolated properly?
The former should deliver BETTER quality than the latter, as the scaling in 1. will exploit the full vertical resolution present in the source?
What you can't do is take a 480i FRAME (i.e. both fields without de-interlacing) and scale this to a 1080i FRAME - as the scaling will mix fields between input and output, and you get horrible tearing effects - with wobbly edges on movement etc. (You often see this on poor quality PC non-linear editors being used incorrectly)
Conversion from 480/60i to 480/60p and then back to 480/60i (or 576i to 576p back to 576i) is now quite common within broadcast digital video effects units (the boxes that scale, rotate, zoom, page-peel etc. video in TV shows) as the quality is better if you process frames rather than fields, allowing smoother edges, clearer movement etc. (This is the Frame-based vs Field-based DVE argument for those that know)