DVL-Digest 705 - Postings: Index MPEG2 sampling versus DV Uncompressed DV Which 2/3 inch lens is best? - (2) MPEG2 sampling versus DV - Adam Wilt Assorted comments on compositing... > This is where MPEG2's 4:2:2 final recompression > for output to analog should be superior to DV > compositing and then 4:1:1 output to analog. A common misconception. If your source material is 4:1:1 DV (or 4:2:0 DV), you have ALREADY lost the additional chroma detail and recompressing to 4:2:2 MPEG-2 CANNOT make it better. Indeed, the format transcoding will likely cause a slight loss of the quality. As all compositing is done internally at 4:4:4, the clips will be upsampled to 4:4:4 whether from 4:1:1, 4:2:0, or 4:2:2; a DV-originated image at 4:1:1 upsamples to 4:4:4 every bit as well as a DV clip upsampled to MPEG-2 at 4:2:2 and then upsampled to 4:4:4. "4:2:2 for output to analog" is too late to affect compositing quality. > Max quality MPEG-2 with the RT2000 is 25Mb/s the same data > rate as DV. But it is 4:2:2 not 4:1:1. And because of this, you're trying to fit 33% more data into the same compressed space. One of the things that makes DV work well is that the reduced chroma sampling reduces the data rate prior to the compression, resulting in milder compression. MPEG-2 I-frame at 25 Mbps, all else being equal, will have higher chroma resolution at the expense of higher compression. > So if you are working with 4:4:4 > graphics and compositing lots of them together -- MPEG-2 is better. This one has it right! ONLY if your source materials stay at 4:4:4 or 4:2:2 prior to compositing will 25 Mbit MPEG-2 give better results. If your key source has been shot on DV, recompressing to MPEG-2 buys you nothing. Uncompressed DV - Perry Mitchell At the moment direct processing in the DV compressed domain is not possible, probably even for a mainframe computer. Any processing of the video therefore must involve first decoding it, and usually 'up-sampling' to say uncompressed 4:2:2 or R:G:B (which is sort of 4:4:4). This will involve some interpolation which is a posh word for guestimating, and will NOT add any information. Depending exactly on what you want to do with the processing, there are good arguments for and against such up-sampling, so there is no out and out 'best way'. There is NO possible added value for transcoding (to MJPEG or MPEG2) and this will inevitably lead to extra losses. The only justification is to allow use of existing hardware that is dedicated to say MJPEG storage, but I'll repeat that this CANNOT improve the DV quality. It can of course give you access to better edit facilities, and if a sufficiently low MJPEG compression is used then you may consider the drop in quality acceptable. Think of DV as Dutch, and uncompressed as English. If you want to have a technical conversation with another Dutch guy and maybe throw in some uncompressed titles and graphics, then it makes a lot of sense to hold it in English. It doesn't make any sense to translate your Dutch to German first!! Which 2/3 inch lens is best? - "Perry" Everything that man Danny Grizzle says is a pearl of wisdom! About 15 years ago I was asked by a Canon executive what would be my ideal portable lens (2/3in) and I replied 10x5mm. They all laughed and said very difficult, but you can now get these numbers (actually a little better). Shortly after I bought a Canon 8x6mm (not x2) and have very rarely had to put anything longer on, but the extra wide would have often come in handy. Perry Mitchell Video Facilities http://www.perrybits.co.uk/ Which 2/3 inch lens is best? - "Perry" A modern quality portable video lens gives the equivalent on a 35mm camera of about a 600mm telephoto. Look at the price of a professional 600mm still lens and then remember that the video lens is a zoom with both iris and zoom servo systems built in, and the price is not quite as astronomic as some folk have suggested. Perry Mitchell Video Facilities http://www.perrybits.co.uk/ (diese posts stammen von der DV-L Mailingliste - THX to Adam Wilt and Perry Mitchell :-) [up] |