FX1 and simple FX in Cineframe24. My experiences

[Anhar Miah]

Like Sony said before its either you're with them or us (them being progressive, and us being interlace)

[Davi Dortas]

Well in Japan, interlace is very popular for dramatic movies, so maybe Sony located in Japan is a reason they choose one thousand eighty i?

[Steven White]

>>how big are your huffYUV compressed files

The data rate is between 40 and 50 MB per second. The files are 23.976 fps files with a 1440x1080 resolution. The final renders of the movie (which I haven't done yet, as I have loads of colour correction to do) will be 1440x818.

A good estimate for the final version of the flick (at about 6 minutes) will be 20 GB. Easily backed up on DVD-R.

In terms of transfer to DVHS anything that's not the plan. The flick will be released on the web first as a Sorenson3 QuickTime file (probably at 720x306) and maybe as a WM9 file in HD. I will also be downsampling it to a 24p DVD. Whenever HD-DVD or Blu-Ray come out and are cheap enough I'll remaster to that... but since I don't have an HDTV, there's no point in worrying about it at this point.

-Steve

[Ignacio Rodriguez]

> Big question still is: why no progressive?

The cam's already low sensitivity to light would be even lower for progressive scan at the same chip size, right?

[Barry Green]

<<<-- Originally posted by Davi Dortas : Well in Japan, interlace is very popular for dramatic movies, so maybe Sony located in Japan is a reason they choose one thousand eighty i? -->>>

Which is probably one of a few very good reasons. For one, I think the Japanese market apparently doesn't necessarily value 24P/30P the way the American (and other) markets do. From my extremely limited understanding, pretty much all dramatic production in Japan is done in 60i, isn't it?

Second, keep in mind that Sony is engaged in a death battle with Panasonic over the future of HDTV in Europe. The EBU is going to choose 720p OR 1080i, unlike the American ATSC which endorsed both standards. Sony is championing 1080i, Panasonic is spearheading the 720p charge. From the reports I've heard, the EBU is leaning strongly (if not unanimously) towards 720p. If Sony had produced progressive chips and made a progressive camera, it might have compromised their position regarding the sincerity of their belief that 1080i is the "right" choice for the entire continent of Europe.

Third, on paper, 1080i sure sounds a lot higher-resolution than 720p, doesn't it? In reality they're just about the same; once you get done with row pair summation, low-pass filtering and the Kell factor, HDV 1080i yields a real-world pixel count of about 775 scan lines of resolution, 720p yields 720 lines. So from a marketing standpoint it makes the Sony sound like it's much higher resolution, even though when it's all said and done they're pretty much the same (1440 x 775 vs. 1280 x 720, they're within about 18% of each other, vs. the "on paper" spec which would show 1080i as having twice as many pixels per "frame"). And 720p is capable of pumping out about 1.6x as many pixels per second, since it does a full progressive frame with full height 60 times per second vs. a half-resolution field 60 times per second for 1080i.

Finally, as Chris said, there's no provision for 1080p in the HDV specification. JVC's already announced their intention to ignore that and go ahead and produce a progressive 1080 camcorder, recording like the DVX and XL2 do, writing the progressive image out in interlaced 1080i form. But as per the pure HDV spec, there is no 1080p in HDV.

So it all really does come back to what Sony said in the introduction: now consumers have a choice -- they can have 1080, OR they can have progressive, but as far as Sony's concerned, not both.

[Dylan Pank]

<<<-- Originally posted by Barry Green : Finally, as Chris said, there's no provision for 1080p in the HDV specification. JVC's already announced their intention to ignore that and go ahead and produce a progressive 1080 camcorder, recording like the DVX and XL2 do, writing the progressive image out in interlaced 1080i form. But as per the pure HDV spec, there is no 1080p in HDV. -->>>

Barry, what would be the practical difference between 1080 progressive scan recorded to 1080 interlace compared to 1080p scan recorded to 1080p on tape, assuming a no-pulldown frame rate (i.e. 30p to 60i, or 25p to 50i - lets ignore 24 to 60i with 2:3 pulldown for the minute.)

I'm kind of at a loss to work out what the difference really would be, other than at the actual playback stage.

[Ignacio Rodriguez]

Good question Dylan,

I wonder myself. Would it not be great to piggyback 1080p30 on 1080i in the same way as the DVX100 and XL2 can piggyback 480p30 on 480i? Would this not be better spatial resolution than 720p30? Sort of like CF30 but with real proscan. That might be the next step for Sony or JVC.

[Sam Edwards]

Hi Steven,
I'm a compositor at ILM and I think your work is just great. Keep it up and keep me posted on your work. I think the cineframe is very useful for your application.
cheers,
Sam

[Steven White]

Woah! Thanks Sam. That made my day : )

-Steve

[Toke Lahti]

<<<-- Originally posted by Barry Green : Second, keep in mind that Sony is engaged in a death battle with Panasonic over the future of HDTV in Europe. The EBU is going to choose 720p OR 1080i, unlike the American ATSC which endorsed both standards. Sony is championing 1080i, Panasonic is spearheading the 720p charge. From the reports I've heard, the EBU is leaning strongly (if not unanimously) towards 720p. If Sony had produced progressive chips and made a progressive camera, it might have compromised their position regarding the sincerity of their belief that 1080i is the "right" choice for the entire continent of Europe. -->>>

It may also be that the path will be 720p first and 1080p then.
Here's couple of good documents about the issue:
http://www.ebu.ch/trev_300-wood.pdf
http://www.ebu.ch/CMSimages/en/tec_d...tcm6-17109.pdf
There's lots of good reasons to forget interlaced picture.
1) ccd/cmos are basically progressive
2) all displays will be progressive
3) progressive picture can be easily shown on interlaced display
4) interlaced picture isn't so easy to show on progressive display
5) in intraframe level progressive picture compresses a lot better
6) distributing eg. 25p material over 50p feed does not need any additional bandwidth
So, as stupid as it may sound, Sony might try to keep up with interlaced just for authority reasons...

<<<-- Third, on paper, 1080i sure sounds a lot higher-resolution than 720p, doesn't it? In reality they're just about the same; once you get done with row pair summation, low-pass filtering and the Kell factor, HDV 1080i yields a real-world pixel count of about 775 scan lines of resolution, 720p yields 720 lines. So from a marketing standpoint it makes the Sony sound like it's much higher resolution, even though when it's all said and done they're pretty much the same (1440 x 775 vs. 1280 x 720, they're within about 18% of each other, vs. the "on paper" spec which would show 1080i as having twice as many pixels per "frame"). And 720p is capable of pumping out about 1.6x as many pixels per second, since it does a full progressive frame with full height 60 times per second vs. a half-resolution field 60 times per second for 1080i. -->>>

Absolutely right!
I do a lot of work for multidistribution (net, dvd, vhs, television) and there is lot of additional work to change interlaced to progressive to some formats. Originally progressive material would show up nicely on everything.

<<<-- Finally, as Chris said, there's no provision for 1080p in the HDV specification. JVC's already announced their intention to ignore that and go ahead and produce a progressive 1080 camcorder, recording like the DVX and XL2 do, writing the progressive image out in interlaced 1080i form. But as per the pure HDV spec, there is no 1080p in HDV. -->>>

But is there a need for 1080p spec?
If you capture progressive frames from camera, but record them as separated fields (PsF), I don't see any violation of the hdv specs.

[Dennis Hingsberg]

I couldn't resist playing around with Steve's Star Wars footage a bit. The clips were color corrected and downsampled to 720x480.

http://www.starcentral.ca/trailers/starwars.mov 6.2MB

[Heath McKnight]

I'm going to be taking another look at the CineFrame 24. We shot my short film yesterday (Sunday) in CF 24 and it looked great. I'm going to capture it, etc., then import it into Cinema Tools in Apple and do a reverse telecine and go from there. Should work, but I'll post my results ASAP.

Also, Jon Fordham said, barring testing of course, that maybe shooting the Z1 in PAL mode with the CineFrame 25 on may work best of all, much like how Frame Mode was used on PAL XL-1 cameras while shooting 28 Days Later... Of course, testing needs to be done as soon as the Z1 comes out, so maybe we'll shoot another short in February that way and see what happens.

So, I'm starting to really think I was a bit wrong in my first thoughts on CF 24.

heath

[Barry Green]

I just did an extraction of a resolution chart in CF24, to figure out what kind of res it's actually delivering. It's tough to tell truly, because besides the uneven motion sampling, it also "pulses" and "jitters" even on still shots. So you can't just pull a still shot out and judge the resolution, as the resolution is moving! I extracted a four-second portion of the resolution chart and made an uncompressed .AVI of it, and posted it at:

http://www.icexpo.com/FX1-CF24-Res.avi

It's a 10-megabyte file, please right-click and "save as..." so it doesn't eat up all my bandwidth.

In examining the individual frames, it groups the frames into groups of five (using a standard 3:2 pulldown scheme). Within those groups of five, it looks like there are two frames at about 425 lines of res, then two frames at about 475, and then one frame at 575 lines. Which means for vertical resolution it's lower than an XL2 or DVX, although still having higher horizontal resolution; and for vertical resolution it's not stable.

CF24 is unusable. It would be much better to shoot CF25 and slow down 4%, or shoot 60i and use something like DVFilm Maker or Graeme's filters.

[Heath McKnight]

Barry,

Try doing a reverse telecine with the CineFrame 24, which is something I'm going to do with some clips, if not the whole film we did yesterday, if it looks good and works well in my tests.

heath

[Barry Green]

I posted a lengthy post in a thread here not too long ago (last couple of weeks) about reverse telecine on CF24. The results are 24 "frames" that feature uneven motion sampling -- a constantly moving object will move approximately 1/3 the distance on even frames, and the remaining 2/3 the distance on odd frames (or some similar ratio, don't remember exactly).

But the thread's here on DVInfo...