![]() |
Tom is correct -- for both film (after 2-3 pulldown has been sensed) and interlace video -- deinterlace should use "weave" on a static pixels. On moving pixels, "bob" is the simplest/cheapest, but worst to use. A 2D FIR is much better than line-doubling, but can only be done in hardware, which is why you don't find it in software players.
The entire subject is summarized at: http://digitalcontentproducer.com/hd...work_01082007/ |
Quote:
The article you linked is excellent. When you say Full 1080 HD prosumer camcorders are likely for next year, does this mean 1080p60 and full resolution native sensors? What media will these record to? |
Quote:
Next Monday HDVatWORK will cover the new Full HD JVC. Not 1080p. |
oil paint or watercolor effect not an illusion
[edit]
Alledgedly,[/edit] the encoder in the V1 (that covers all models) isn't robust enough to handle the extra bandwidth required for progressive. The encoder, therefore, runs inefficiently causing a good number of macro blocks. The NR tries to smooth out these macro blocks, but in the process fine detail is compromised, producing this so-called oil paint/watercolor effect. Obviously, certain scenes are more noticeable than others. NTSC seems to handle it better than Pal, and, although not perfect, is adequate for the V1U. [edit] [/edit] My honest opinion...properly used and considering the price and what you get... The V1 is still the best under $5k camera to date. For me... since it is so close to NAB, I'm waiting to see what rolls out. Of course, I absolutely hate the HDV compression method. Peace all |
Steve very nice articles on deinterlacing HDTV's.
Not that this is related at all but the weave method you talk about is a prime example why the 24F mode on the Canon series of cameras can work so well. Not that thats what Canon is using but just the fact that a 1080p display can turn 1080i video into something that comes very close to 1080p. Many people question how an interlaced chip can create a progressive image and whatever Canon does seems to get results similar to what a good quality 1080p HDTV can do with 1080i video. I think the reason why the F modes seem to have about 10% less resolution is because of something similar to the filtering you talked about with one of the methods of deinterlacing that would give 750 lines. This is only slightly smaller then the limit of 810 lines with interlace filtering on 1080i material. Anyway back to the 1080p displays. I wonder how many people can even tell the difference. I mean in order to find the sets that fail you have to feed an image that alternates black and white lines. 99% of video and even graphics will never do this sort of thing because most designers know not to create graphics like this from their old SD interlaced days. Considering a lot of people still think a 480p DVD is HD perhaps this concern is a little over the top (even for me). Of course all the HDTV's I have are 720p anyways and it all looks the same to me. Perhaps the bobbing up and down can be a distraction on the 1080p displays. Very interesting articles though. |
Quote:
It's the DSP not encoder. [edit] [/edit] Steve Mullen has made a huge effort to deflect criticism of the V1 to deinterlacing schemes. Well done! But wrong. All V1s a affected and it has NOTHING to do with deinterlacing. TT |
Guys, this subject has been beaten to death, had 3 threads merged, and constantly requiring edits due to personal attacks, rumors, claims of "insider information," and conspiracies.
It's time to move on. If you have new information on the topic that hasn't been hashed out in this one, please begin a new thread. Brett, thank you for posting new footage for folks to play with. |
All times are GMT -6. The time now is 12:27 PM. |
DV Info Net -- Real Names, Real People, Real Info!
1998-2025 The Digital Video Information Network