|July 14th, 2009, 12:33 PM||#1|
Join Date: Mar 2008
Location: Richmond, Canada
HDV to SD-DVD without ProRes
Okay before you all jump at me, I have done plenty of research of the topic both here and out. The consensus definitely seems to say to render the timeline in ProRes 422, but I only have FCS1 or FCP 5.1.4, so ProRes unfortunately isn't an option.
I've tried some other suggestions that people say would give poor results, but I haven't been able to produce any better so I can't tell if it's poor or not. Of course, with big-name movies, we see excellent compression in SD-DVD's and the quality is top-notch. My HDV videos are shorter and doesn't quite look like its HDV counterpart when made into SD.
So now here are my questions:
1) If I export the HDV sequence as a reference file (self-contained or not), are there any custom settings in Compressor that may produce better quality than "Best quality for DVD x minutes"?
2) The other way I've been converting to SD is to create a SD sequence (choosing DV-NTSC in easy setup), then copying and pasting the HDV edited timeline into the sequence. I'll have to render it for quite a while but it seems to work better (though maybe not necessarily better quality) than the above method. Then of course I export as ref file, drop it in compressor then author in DVDSP.
BUT the image now looks soft and nowhere near as detailed as HDV. Sothe way I made it look "better" and more HDV-like was to add a sharpen filter in FCP and set it to around 50. This reveals more of the details that I could see in HDV. I've never seen this solution anywhere. Does anyone approve of this idea?
EDIT: I thought I'd share another problem when I went through method 2 in my last video. When the first clip came up AFTER compressing to DVD in Compressor, it would show some REALLY BAD artifacts or glitchyness, I have no idea what you call it. However, over a few seconds, it would gradually go away and go to normal, all in the same clip. It didn't happen anywhere else in the video, just that one time. It also doesn't happen in the FCP timeline or when watching the reference file, before compressing. Anyone know the deal with this?
|July 14th, 2009, 11:04 PM||#2|
Join Date: Jun 2007
Location: Hamilton Ontario
Unfortunately, you're working with HDV, which is a different animal...
I'll give you what little knowledge i already have on the matter, and is totally software agnostic...
HDV downconverts, require a doublerate deinterlace (bob), a resize, and a re weave...This is the proper way of taking 1080i sources, and resizing down to 480i..PERIOD. So in essence, each frame is split into fields, deinterlaced, resized, and weaved back.
You simply can't take an interlaced frame, and resize it without destroying information. I believe any NLE will simply try to do some framebased resize (therefore giving you a soft look).
I don't know how else this could be done, except through VirtualDub or Avisynth (both free)..
Incorrect video processing will yield inferior results.
Just to give you a visual, here's the process i use to do proper downconverts...Since i use Avisynth, the drawing should help explain my process...
Of course, the desired deinterlacer is all subjective. I just happened to learn about this one, and gives me great results.
|Thread Tools||Search this Thread|