Jan_Crittenden
Veteran
However you don't seem to recognize and give credit to the fact that the MPEG2 algorithm does give very feathery edges to the hard line required for the Key. With color on every other line vertically, this could be a problem in cutting the key.
But you do seem to look at only facts, like the EX1 has a 1920 X 1080 image and the HVX is less and you totally diss the effect of the spatial offset, and not the total picture that the facts make up. Research requires the capacity to understand the impact of the numbers and not just the numbers. It requires an in depth understanding of how the Long GOP works and the 8 X 8 blocks that the HDCAM EX and HDV make; have in their contribution on the camera's output. If you were not going to go to the record portion and were only going to look at the HDSDI of the EX1, I would say certainly it would have more resolution and would cut a pretty nice key, but the recording format which is what the 4:2:0 and 4:2:2 reference, means that in 4:2:0 you only have color on every other line, not every line like 4:2:2.
This has only impact on the generating of the signal to the quantizing process but has little to do with the recording process. You need to study up on the recording process. 4:2:0 means in its ration that every other line has a color sample, 4:2:2 means that every line has a color sample. If the compression scheme has to deal with a lot of motion, this is where it is going to take the greatest hit.
And having every 15th frame as the only frame that is there in its entirety with all of the others just being the changes is how comforting? What if there are more changes than the resolution can handle. I mean jeez, on their 355 camera they did a prefilter to 1440 X 1080 to optimize the XDCAM codec, but with the XDCAM EX now there is 40% more information to make it into the same 35Mbs bit bucket. Seems like more compression to me. If you can use the facts to prove me wrong, I stand here willing to listen how a Log GOP of 15 frames is going to produce pictures that are better than an I-Frame Codec. And keep in mind during your argument, please tell me how that 4:2:0 every other line of color is going to make it all better than a 4:2:2 codec which has color definition on every line.
And before you waste all of that typing skill, I suggest that you just go try the cameras, because all of this number mumbo-jumbo does not make you money, pictures do. Never having to say you are sorry that you camera cannot take that shot as blocked, counts.
I am not arguing the resolution of the two cameras. The EX1 has more as long as the camera is standing still and you are not recording to the codec on the camera. If you are moving the camera and recording the codec, well it changes and you have no control over it. In fact you don't even get to see what it is doing unless you review each shot.
Here you are just simply misinformed. You need to look at the fact that HD starts at 994Mbs in uncompressed HD and then you have to look at which signal has less information when all is said and done. The DVCPRO HD codec does a prefilter to reduce the amount of compression. The EX, well it doesn't, so more to compress, more compression artifacts. And remember once you have given away the information, ie. in a long GOP scheme, you do not get it back, even if you choose to edit in an intermediate codec.
Long GOP only gets better when the VBR gets down to the Block level like in AVCHD. In the newer MPEG4 codes where the blocks are variable in size, and the codec doesn't have to make the same repeated compression on the same color in every location that is where the improvements have really come. MPEG2 is MPEG2, and just because you think it is better that I-Frame only codecs doesn’t make it true and even AVCHD is not better than an I-Frame Codec. Just not true, more efficient, sure but I am not a member of that club that says before distribution I should give away the details that may prove to be important. And I also want to be the one deciding when to give that away. I don't want my camera to be making those choices; I want to think about that compression after my production is finished. But back to the AVCHD, go read the Sony white paper on the AVCHD codec and let us know what you think. In a nutshell it says that the 9Mbs of AVCHD is roughly equivalent to the HDV codec. MPEG2 is old, old old.
How about pans on vertical buildings cause skew, or compositing a moving image where the geography of the image is changed due to the cmos artifact. And you don't always have control about some things that happen, in verite or documentary work.
And frankly that only reinforces my entire argument, but we aren't talking about a Red here. We are talking about cameras under $7000.
If wishes were horses. Film cameras have a sync input signal. Show me the sync input signal on the EX1. And yes I have read where people have used them with some sort of reference but that is when the sync is known to happen, what about the out of your control flashes. And because it happens on film cameras does that mean that it is a desirable thing. I don’t think so.
That said, how about the fact that the CMOS is more sensitive to Infra-Red, making black different shades of purple, brown or green dependent how much infrared is part of the light source. This is a fact. Or how long does it take to go from record to play back or vice versa on the EX1. Well that would be 14 seconds from PB to REC, and 11 Seconds to PB from record. In P2, is less than 2 seconds from PB to REC and lest than 1 sec. the PB. And these are facts. A camera purchase is important and should be about more that a 1920 X108 CMOS imager and the ability to wrangle the numbers to make it sound viable. It is not what you make but it is about what you keep and what you are able to show your customer at the end of the shooting day.
They are not small quirks when it comes to the user interface and the overall system. We haven't even started to discuss the metadata workflow which adds an entire layer of power to the P2 system. Point is, as a camera that has resolution, the EX1 has resolution, but it is not a stic point it is a moving point and it can vary depending on what you are shooting.
A CCD is a more consistant output, do that research, even the manufacturers of both, Dalsa, say this is true. Ask Juan Matiniez who stated in his article about the differences. He basically said if you are at the high end of digital cinematography, CCD is the choice for you, but at the low end, well CMOS is good enough. I don't happen to agree, but then I have CCDs at the low end to offer.
No imager is a system all unto itself, the rest of the system must be taken into consideration.
Best regards,
Jan
