Experinece with deinterlace clip of FX1?

[Gabriele Turchi]

Hi guys

Anywone have tried to deinterlace with sofisticated methods(Magic Bullet ecc....) and great result the clip of Kaku?

The clip " DawgHDV60i60.m2t " is perfect for deinterlace and make a comparison with the clip " DawgHDV24f60CT.m2t " (that was shooted by Kaku in cineframe24mode).

I have tryed with DVFilm maker ,but it accets only avi format and Cineform HD tryal version is not ready at the momemnt to convert the Sony Clip,i don't knwow a good way to create a good avi from sony clip.


Any Experience?


Best regards


Gabreiele

[Graeme Nattress]

I tested that very clip with my Film Effects software for FCP and it worked fine. A standards conversion to 24p fared less well due to the diabolocal 4:2:0 used, which I'll write a new version of the converter to deal with better. With a little fiddling though, it worked great, just as pristine 1080i from better sources converts great. But that's because I was working in a 4:2:2 codec, rather than staying native.

The grain in the images gave the entire thing a very close 16mm look, which was quite pleasing. It will certainly be interesting to see the camera for real and have a good play with it, once more people get the chance to see one. Also, it will be good to see how the pro version handles.

Graeme

[Jeff Kilgroe]

I've worked with some of Kaku's clips and I am getting very good results with converting to 1080p24. I can play them back at full 1080p resolution on my 1920x1200 display and they look awesome.

As Graeme found, the converted video does have a very 16mm look. If the CCDs were of higher resolution, it could probably surpass 16mm, but the green-shifted 960 pixel array just can't pull it off. While these images look wonderful at full resolution, they are still soft and have about half of the perceptable resolution of a HD cam supporting native 1920x1080. Those people saying it compares favorably to a Vericam need to have their eyes checked or they're watching the video on a display smallter than 1920x1080.

The best advice I can give to anyone thinking about editing HDV is convert to a different format before doing anything else. HDV/MPEG2 is too "lossy" and doesn't hold up well to much of anything and the video starts to soften and show increased artifacting. There's more to it than the 4:2:0 sampling and converting to a format with 4:2:2 does help. I got the best results for 24p by transferring the M2T files to 1080p uncompressed 4:2:2 (better plan on upgrading my video array to 2TB of space).

Anyway, I'm using Vegas 4 for most of this as it handles M2T streams natively and HD resolutions. I'll have to upgrade to the latest version of Vegas when I buy an HDV camera (assuming I will) as Vegas 4 can only edit M2T, but can't capture from the new HDV cameras.

The high compression necessitated by the 25Mbps data rate of HDV and the lower resolution CCDs are what hold the FX1 back from competing with the much more expensive pro HD camcorders, but these clips surpass anything I have shot with anything else in the under $10K price range. The FX1 so far is surpassing my DVX100 in terms of detail and color continuity by a huge margin. ...The detail alone justifies me upgrading to an FX1. Oh, and I converted a few of Kaku's M2T clips to 24p 720x480 anamorphic for DVD and they look amazing. A good multi-pass encoder like the one from Mainconcept or even the new one from Adobe give excellent results.

I'll be waiting for the "pro" version to come around this spring (and hopefully some competing cameras) before I make my purchase decision, but the FX1 just looks better and better the more I play with the available footage.

[Carlos E. Martinez]

<<<-- Originally posted by Jeff Kilgroe : If the CCDs were of higher resolution, it could probably surpass 16mm, but the green-shifted 960 pixel array just can't pull it off. While these images look wonderful at full resolution, they are still soft and have about half of the perceptable resolution of a HD cam supporting native 1920x1080. Those people saying it compares favorably to a Vericam need to have their eyes checked or they're watching the video on a display smallter than 1920x1080. -->>>

Isn't the green shift native to NTSC? Perhaps the PAL version has a red shift and will look better?

<<<-- The best advice I can give to anyone thinking about editing HDV is convert to a different format before doing anything else. HDV/MPEG2 is too "lossy" and doesn't hold up well to much of anything and the video starts to soften and show increased artifacting. There's more to it than the 4:2:0 sampling and converting to a format with 4:2:2 does help. I got the best results for 24p by transferring the M2T files to 1080p uncompressed 4:2:2 (better plan on upgrading my video array to 2TB of space). -->>>

What do you use to convert to a different format and what do you recommend? Any PC or Mac program will cut it for that conversion?


Carlos

[Jeff Kilgroe]

<<<-- Originally posted by Carlos E. Martinez :
Isn't the green shift native to NTSC? Perhaps the PAL version has a red shift and will look better? -->>>

The green shift I was talking about is how the green layer CCD is offset by 1/2 pixel. This is a trick that effectively doubles the luminance data for each addressible pixel within the CCD array. By doing this, Sony is able to create the perception of 1440 horizontal pixels, when in fact they do not have that native resolution. The reason for using the green layer is that in an RGB system, green carries the most variance for luminosity and brightness, blue carries the least. If they were to shift the red CCD layer instead of green, the perceivable resolution would still be doubled, but it would have nowhere near the luminosity and contrast range gained by shifting the green layer. The PAL camera version still uses the same CCDs and this green pixel layer shift has nothing to do with NTSC vs. PAL. When shooting HD resolutions, there is also no such thing as NTSC vs. PAL. HD is governed by the ATSC. Sony has links to what resolutions and rates are supported by the FX1 as well as a full ATSC mode chart available on their site.

<<<-- What do you use to convert to a different format and what do you recommend? Any PC or Mac program will cut it for that conversion? -->>>

Any decent editing application like Vegas or FCP-HD will do these conversions. HDV out of the FX1 comes in at a horribly over-compressed, 25Mbps rate M2T stream at 4:2:0. The best thing that can be done at this point (if you're going to do any serious editing or post work with this) is to convert it to a format that isn't going to lose any more information as you work with it. Ideally, resampling the M2T stream into an uncompressed 4:2:2 video file would be best, but not everyone has the disc space or CPU power to handle working with such a beast after the conversion. A compromise may have to be made such as going to a lossless or nearly lossless compression. Every time you re-encode an MPEG2 file (as in the M2T file pulled straight out of the HDV camera) you lose more detail and gain more artifacting and noise. MPEG2 is very intolerant of re-encoding and doesn't hold up anywhere near as well as what we're used to with DV. If you're going to be doing operations that require re-encoding the video more than just once at the time of final output, then you will want to get it into a format that will stand up to what you will be doing to it.

Jeff