|
|
Posted by Spex on 11/22/06 15:37
Jan Panteltje wrote:
> On a sunny day (Wed, 22 Nov 2006 09:58:25 +0000) it happened Spex
> <No.spam@ta.com> wrote in
> <45641f42$0$8759$ed2619ec@ptn-nntp-reader02.plus.net>:
>
>> Jan Panteltje wrote:
>>> On a sunny day (21 Nov 2006 14:14:07 -0800) it happened "Mike Kujbida"
>>> <kujfam@xplornet.com> wrote in
>>> <1164147247.491244.232250@k70g2000cwa.googlegroups.com>:
>>>
>>>> Jan Panteltje wrote:
>>>>> On a sunny day (21 Nov 2006 13:38:13 -0800) it happened "Mike Kujbida"
>>>>> <kujfam@xplornet.com> wrote in
>>>>> <1164145093.775293.55280@h48g2000cwc.googlegroups.com>:
>>>>>
>>>>>> The downconverted HD will look much better than straight SD because
>>>>>> you're starting with a higher quality to begin with.
>>>>> This is not correct, 'aliasing' will occur.
>>>>> For aliasing, in the simplest form, think how to put 10 dots on a line in 7.
>>>>>
>>>>>
>>>>> 1 2 3 4 5 6 7 8 9 0
>>>>>
>>>>> a b c d e f g
>>>>>
>>>>> Never quite fits now does it?
>>>> Sorry Jan but my own personal experience (as well as that of several
>>>> users on various Vegas forums) tells me otherwise.
>>>> I borrowed a friend's Z1 one day and did some test shoots of water
>>>> flowing down a stream into a pond. I locked the camera on a tripod and
>>>> shot in both SD & HDV. I then brought this footage into Vegas 3 ways,
>>>> SD, HDV & HDV downconverted. I also rendered the HDV footage to
>>>> SD,again in Vegas.
>>>> I then looked closely at all 4 shots on my reference monitor (JVC
>>>> TM-H150CGU - 750 line SD monitor). To my eyes, the downconverted HDV
>>>> (either way) looked better than the straight SD footage.
>>>>
>>>> Mike
>>> Well I cannot argue with [anybodies] experience.
>>> Maybe there are other factors at work, but the theory stands.
>> Your theory is ill thought out.
>>
>> You are overlooking the fact that the people who write NLE editing and
>> compression software know a little bit more about their subject than
>> many give them credit for?
>>
>> The images are not subjected to a rudimentary resize and compression
>> algorithm as your example implies. The images are in fact resized
>> generally with sub-pixel accuracy, anti-aliased, then sent to the
>> compressor.
>
> And that makes these better then the original format?
> But no more fits in the original format!
DV vs HDV down-convert. HDV down convert is slightly better due to
better DV conversion by software than in realtime in camera. The
conversion is likely working to more decimal places figures than
realtime. Not all DV encoders are created equal for example the Canopus
DV codec has always been highly regarded for quality and is probably
better than the codec in the camera. This is where the gains are being
made when down converting HDV to DV.
The big benefit people are finding by working in HDV to later output to
SD DVD is maximum quality is maintained throughout the workflow. A
typical HDV camcorder can record 600 - 700 tvl to tape in 4:2:0. By
finishing the project in HDV or high quality intermediate codec then
resizing to 720x480 the colour resolution is now approaching 4:2:2.
When fed into a MPEG2 DVD compressor there is twice as much colour
information and "real" resolution for the compressor to work with which
produces much better output. This pays dividends when working with CGI
and titles. Anyone who works in DV with graphics will tell you how
badly mashed they get!
If you work with DV footage throughout your workflow you really do shaft
yourself. The 720x480 DV stream has really only about 520x400 (if that)
lines of resolution at best with a colour resolution 4:1:1 for ntsc.
Feed that into your DVD compressor which colour samples at 4:2:0 and it
make mince meat of your footage. Incidentally DV PAL to SD DVD is
better as the use a consistent colour sampling scheme of 4:2:0.
Navigation:
[Reply to this message]
|