Yes, of course we remember.

But we are a skeptical bunch.

However, today's issue of Post magazine has an interview with Matt Merkovich at Digiscope (Minority Report, Blackhawk Down, Master & Commander) in Santa Monica.

They are testing it for shooting effects plates and elements for compositing. a complete system is about $10K, including a new Mac because the software doesn't run on Windows and they gave up on Macs, because "there's currently not enough processing power."

2 hours of 10-bit HD on a 200GB drive, log color space (or linear or gamma curve).

(For the 10-bit of course you'll need Vegas 7 if you want to do more than straight cuts.)

Uhhhhh...

Really?

He's kidding

if he knew - he couldn't say

if he didn't know - then he could say and did - so there you go.

Not really - though that wouldn't stop him from being right.

Dave

Panasonic have a similar camera, the HVX 200, not quite 4:4:4 at only 4:2:2 and only 8 bit but a bit easier to deal with, sort of.

Bob.

The modded DVX100 does allow a little extra exposure latitude, which is kind of what's missing from low-level cameras. Sort of like a low-budget SD version of the Viper.

You're right in that many people would consider the HVX200 though.

I just hope the DVX200 survives the initial impressions that the whole was less than the sum of its parts.

On paper, it's fabulous. I can't wait for more real-life reviews to hopefully dispel the early mixed feelings about the output.

I did some comparisons of those affordable HD/HDV cameras and lost all hope (bear in mind i still think some great work will come out shot with those cameras as in the end it's not the camera alone that makes great movies). The cameras are too noisy in dark areas and HDV is way too compressed. So unless i was shooting grungy project where noise is something that we are looking for ... I won't be using any of those cameras any time soon. ...ANd those words are coming from a guy who claimed that HVX200 is gonna be a "holly grail of indie filmmaking."

That wasn't the same guy who tried to talk his way around the noise issue by suggesting we think of it as the equivalent of film grain was it?

But here's a question that I've asked elsewhere, no one had an answer for though. What' the advantage of 4:4:4 with a block that uses pixel shifitng?
My understaning is that pixel shifting increases the resolution of the luminace channel as that's the sum of the R,G,B channels, by offsetting the CCDs better res can be achieved but the U,V channels are still the same res as the res of the CCDs, so upping the color sampling will not achieve much, if anything. Going from 4:2:2 from 4:1:1 od HDVs 4:2:0 OK, maybe, but upto 4:2:2?

I've read that on the F950 going from 4:2:2 to 4:4:4 produces no noticable improvement in the film out on the big screen but a big increase in recording complexity.
Bob.

Hi farss. If we are talking about the raw data from a 3 CDD camera then I thought everything should be RGB and there will be no luminance channel at all. How does the pixel shift change this?

Richard Hunter

Not 100% certain.
Consider a 1000x1000 b&w grid. Theory says I need at least 2Kx2K elements to resolve it. Now (and I'm really guessing all of this) instead of having 3 coplanar CCDs each of 2Kx2K elements I could have 1/3rd that number and still resolve those 1Kx1K lines.
However if the lines are red and blue things I assume aren't looking so flash, worse still if the lines are both compound colours we're really in trouble.
The assumption is that as we perceive differences in luminance much more chroma then pixel shifting is justified and I cannot argue with that but when we're talking recording RAW RGB with the intention of better keys and composites etc I'm just not so certain.
Bob.

"F950 going from 4:2:2 to 4:4:4 produces no noticable improvement in the film out on the big screen but a big increase in recording complexity", Bob, you're joking about this, right?

Apples and oranges.

There's a bit more going on here than just the basic color sampling difference.

If everything else was the same, you wouldn't see much difference between 4:4:4 and 4:2:2, because chroma smoothing can do a very decent job with 4:2:2 (much better than 4:1:1 or 4:2:0).

However, there are additional differences in the typical compression levels, and that's where you see a bigger difference, especially for keying.

HDCAM 4:2:2 is DCT-compressed 10-bit (about 2.7:1 if I remember correctly), while the "HDCAM SR 4:4:4" variant most talked about today is 10-bit totally uncompressed RGB recorded to hard disk. Otherwise there's the 4.2:1 compressed HDCAM SR 4:4:4 tape format that uses the advanced MPEG-4 Studio Profile.

There is a lot happening in this field, and the F900 series has had a lot of updates on the camera side in addition to the tape recording side (btw, there are 450 of them in Los Angeles alone!).

Nope, sorry can't remember the title of the movie, it was an underwater horror movie with a BIG budget. F950s in underwater housings 1,000M inside a cave underwater. Feed to recorder on the surface. Connection dual fibre.
The 4:4:4 stream cannot be recorded to tape, only a big disk array and they were nervous about trusting that in the wilds with dodgy power and they went with 4:2:2 to tape as their tests showed no difference in image quality on the big screen.
However as mentioned these beasts are working in 10bit and native 1920x1080, square pixels.
From memory though no compositing was being done, I think the above water scenes that had composites were shot on film.
Bob.