But since there are the CCD's, you have acutally 1.2 Megapixels total. With a little "pixel shift" like Canon uses, you might be able to get close enough to 720p.

DVX100 does 480p in DV mode, but in thin mode, I'm guess it gets more rez - the question is how much and how close to 720p?

That is a good point. I am going to keep that in mind when I write the software.

I beleive 720p uses some quantizing lower than 4:1:1 anyway right?

Juan:

I thought 720p uses regular HD 4:2:2 color sampling, but depending on how you capture the signal, you could use 4:4:4 12-bit and the after post, output standard 720p.

Here's a interesting article:

http://videosystems.com/ar/video_ccd_counting_needed/

<<<-- Originally posted by Stephen van Vuuren : Juan:

Based on the number people dissassembling and altering their camera with home-grown mini35's, I would a pretty reasonable amount.

-->>>


Just a question, what exactly is mini35, i hear a lot of talk about it on this forum.

Cheers
Ben Gurvich

Ben:

Check the dedicated mini35 forum here.

Juan, I only red yr first post...I wish you lots of succes, but I doubt if you are going to see any difference apart from somewhat more compression artifacts in the DV compressed mode for very specific test pictures. Did you know why the first digitalisation needs 12 (up to 16!) bits? Simply because WB, gamma correction,electronic masking, knee processing... all "eat" bits equivalents. If you have much luck (and knowledge in imager signal processing) you will end up with a 8 bit 4:4:4 datastream with somewhat more colordepth and spatial color resolution than the compressed 4:1:1 stream, and this at the expense of a lower S/N. Cam designers allways try to get the "best" picture out of a given CCD structure. Only the DV compression sets some limits which, in the final results are pretty well hidden.

Andre:

I agree that it's not going to magical, but I'm more optimistic. First, it's likely it's the same CCD block that produces PAL signal, so there are more lines available than encoded NTSC DV wise. Secondly, using the thin mode in progressive really strains the DV codec.

I think it's worthwile to see what signal is available.

And, after all, Juan has been generous enough to destroy his camera for the cause! That's like Jonas Salk injecting himself with an experimental polio vaccine. This is a man who must be cheered-on!

Haha Ken! :)

Actually, the procedure is completely non-destructive. The DVX is easy to dissasemble, and the board of interest is the first one that comes out. All probes clip-on, and it doesn't have much effect on the rest of the operation as long as impedance is high enough.

I agree with stephen...i think the difference will be extremely noticeable...i'm not saying it's going to be ~magical~ and generate stuff that's not there, but it IS going to be better. In the simplest of cases, if DV where such a good algorithm, nobody would bother buying expensive DVCPRO50 or digibeta equipment. If the difference between 4:1:1 (5:1) and 4:2:2 (3.3:1) is so noticeable, well...

Agree Juan, 4:2:2/3.3:1 has better spatial color resolution and lesser compression artifacts ( both important for post processing/editing) but 4:2:2 at a 3.3:1 also looks better better because it goes together with larger CCDs, better lenses and is defacto being showed/evaluated in component form. It's also chosen by pro's for the more robust tape format. But plse go on Juan..I am really interested in yr results.

I am so surprise to see Juan said around 80% of data will be lost in the DV compression! I'm keen to see the results!

It's even much more than 80%. If you compare to the already crippled 4:1:1 or 4:2:0(PAL) at 8 bits/sample, then indeed the DV compressor only keeps 20%.(1:5 compression) If you start from Juan's 4:4:4 @ 12bits/sample, allmost 99.5% (~4 Gbits/sec vs 25 Mbits/sec for DV) of the picture content (in term of datarate!) is "thrown away" in the DV concept.

Juan,
Very excited about your undertaking as I sure would like the 4:4:4 12 bit for compositing and possible HD applications. If it cannot get to the full size of HD I'm sure a tool like S-Spline can take it the rest of the way.

Great work.

Juan,

What is your idea of an output? Would you use the Y/C or do you think the DVX could be slightlly modified to house an SDI or RGB (BNC type) connector(s)? I am very interested in this process. Right now I am shooting a lot of content that would be much better suited to do with 4:2:2 or better color. I am running numbers and am thinking of taking the leap to a 4:2:2 system. I am in no rush but your developments could change drastically the financial hole I would need to jump into if I can just modify and use my DVX rather than get into a 4:2:2 camera. Keep on truckin!!

Will Boggs

I know for a fact that a firewire interface can be devised(along with a driver), which would basically allow you to record the signal into any computer.

I am not aware of how the SDI standard works, all i know is that it is unpacketed. However, I would think that yes, it is possible to adapt it to a single or dual SDI interface. However, since the output is most likely not going to comply with any known standard(non-standard frame size, 4:4:4, no compression), i would doubt the SDI output would be of any use because no tape deck would understand/handle it.

So, this is why my goal is a firewire interface and a custom software driver. This way, you can record directly to hard disk and then edit it as you wish. I am also planning to include some option to include decimation in this interface, though probably not compression in the early stages. This way, you can select to record 4:2:2 or 4:1:1 to save space.

I'm waiting on a part right now, but will post updates here.

Juan