Here is the details.
http://homepage.mac.com/kakuito/blogwavestudio/LH20041021153911/LHA20050107210055/index.html

Please do not make direct link to the files in the ftp. If I see this happening, I will have to stop serving the files. It is okay to provide the link above.

I will update that post later to provide you the comparison from digitally up-converted files.

Kaku, i do not understand why the benifit of the 1st method?

Beacuse if it is playing from the tape, then you already have mpeg2 compression, i thought the point of this method was to try and bypass mpeg2 compression or have i missed the point?

Anyway great work Kaku!

P.S is there anyway that the same setup could be achieved from a laptop (thus avoiding having a computer setup in the field?)

1st. method is experimented because software companies are still under development of capturing and converting HDV files at this point.

Especially, QuickTime seems to have more problems handling mpeg2.

I should also post the same footage converted from m2t format to compare. I will do that over the weekend.

And for now, as far as I know, there is no solution on laptop to do uncompressed HD right now. Does anyone know? I know where you are coming from, it would be great to be able to take this outside.

Somebody should start to manufacture a brick that has analog hd component input and fw800 output and embedded cineform's visually losless or sheervideo's math losless compression!

Okay, pulled the footage down and brought it into After Effects and did MAJOR level hauls on the image, checking into all three color channels, looking for indication of MPEG artifacts, which I can find quite easily on any other .m2t footage.

There aren't any. This is without the slightest doubt BEFORE compresson. Thanks Kaku, one major rumor laid to rest!

BTW, for those having trouble looking at the footage;

download, un-zip. Rename the file with .mov extention.
Go to BlackMagic-Design website and download the appropriate win or mac codec. Place the file in your QT

One thing... there is plenty of horizontal fine lines interference visible when mauling the image... I assume this is the result of using some gain. Hopefully not a result of the conversion steps...

Jim Arthurs

Kaku, did you notice if you can switch to CF30 and still see no major picture time delay when live?

Many conflicting theories on the possibility that CF modes involve the MPEG compressor on the camera.

Seeing no extra delay in the live image would suggest that CF happens before MPEG...

Thanks,

Jim Arthurs

First of all thank you Kaku for your time and diligence in posting the footage for us to examine

My interest is in the turtle footage which is direct off the cam head

In concert with Jims opinion I believe there is no m2t compression evidence in the footage, however there is some subtle compression which I am trying to discern where from

For those who dont have broadband I have upped some frames from the turtle footage for others to analyse, they can be found here

http://s17.yousendit.com/d.aspx?id=0PRHO561FA0TU1DGA089KLQNNC


Take any of these bitmap frames and load into photoshop and magnify to 300% and carefully look via the blue channel only

It will be seen that there are some 4x4 pixel blocks which indicate compression - so where are these coming from? - could it be due to the blackmagic lossless jpeg encoder or from somewhere else?

I also saved them as a photoshop jpeg quality 10 to compare and the level of compression howver subtle is certainly better than this

maybe the blackmagic codec isnt truly lossless - realise the turtle file should be about 540meg, but it is only 365meg

I didn't notice anything else than 2x2 blocks. Because of 4:2:0?

Btw, these clips look really nice with 50% size (=deinterlaced).
My work these days is mostly web distributed, so z1 might be quite nice for 960x540 endproduct...

John Jay,

These are 2Vuy codec (which provides more real time editing), so little bit of something could be going on. I will try with other one.

Kaku

do you have the 4:2:2 version or the $:4:4 version (deliberate mistake ):) of the Blackmagic DeckLink HD Pro?

The links seem to be down, any chance of the footage coming back online?

<<<-- Originally posted by Jim Arthurs : Kaku, did you notice if you can switch to CF30 and still see no major picture time delay when live?
-->>>

There is no delay in the picture in any mode on the analog side. That would suggest to me deinterlacing before MPEG 2 compression.

Thanks Mike... I think that disproves the theory of Steve Mullen's that the CF modes are tied into the MPEG chip...

Regards,

Jim Arthurs

Yes, otherwise how would you introduce cineframe in DV mode?

don't you have compression when you export a frame to Bmp or jpeg? isn't that where its coming from? why don't you compare the frame export to a 32bit targa (.tga)?

erm,

there's no compression in a bitmap 24 bit - thats why its 6meg..

is it? ok.

just suggesting.


joel

Been lurking here. Is it positive that the analogue out is only 8-bit, has there many any spec sheet (of the chips and D/A convertor involved) to verify this, or test, could there be 10, 12 or 14 bits colour depth? And has there been any test to verify the resolution, where ever it is the native 960, or processed pixel shifted 1440, 1920, or 2880 at this stage of the pipeline? I say this because who knows what quality is being passed to the custom mpeg2 encoder chip to increase the quality of it's processing? Also has anybody tested the Signal to niose ratio of his stage to see what resolution and bit depth it is limited to?

Thanks

Wayne.

<<<-- Originally posted by Wayne Morellini : Been lurking here. -->>>

Hi Wayne.. that's okay, I've been lurking at your hang-outs... read that whole thread started by Obin in one afternoon... and backwards at that. Big headache. :)

Barry Green states from his tests that the vertical resolution when interlaced is around 800 lines. CF 25 or CF 30 (the in-camera "fake" progressive) he measures to approx 575 lines (more likely 540). Best guest is that CF delivers one single raw fields worth of rez. In interlaced mode, according to Barry, each field is fuzzed down a bit closer to 400 lines to prevent buzzing on interlaced displays.

As to the specs as measured "live" from the component out, I don't think anyone has done any qualitative work here. We're pretty lucky to have had anyone actually try what Kaku has tried. I sure intend to more, when I have my Z1 in hand.

Regards,

Jim Arthurs

The 775/800 in interlaced, and 575/540 in CF30, were measured from frame extractions from tape, not live, so no I don't know if the analog component out might be delivering more. We were monitoring on a large 36" HDTV, live, and I certainly don't remember it being higher, so I think those are probably pretty accurate.

As for horizontal resolution, I posted some res chart extractions here... it looks very clean at 800, and passably clean at 900 although maybe at 900 it crossed the threshold to "unresolved". It did seem like the live analog component looked acceptably clean even at 1000, but examining the tape it looks like 800 is the cleanest resolved resolution. So perhaps it is delivering more horizontal resolution at live analog component out, as opposed to the downsampled-to-1440 version on tape.

<<<-- Originally posted by Jim Arthurs : <<<-- Originally posted by Wayne Morellini : Been lurking here. -->>>

Hi Wayne.. that's okay, I've been lurking at your hang-outs... read that whole thread started by Obin in one afternoon... and backwards at that. Big headache. :) -->

Crazy, that thing is over 500 printed A4 pages ;)

Thanks

Barry, so the measuments of lines, is that actual pixel lines or Test Chart lines? I understasnd the measuments have changed since 4:3.

Is it delivering a 4:2:0, or 4:2:2 (or 4:4:4) through the analogue out, what does the monitor report?

Thanks

Wayne.

john,

is it possible to find somewhere kind of mirror of your previously mentioned link - which is now not working.

http://s17.yousendit.com/d.aspx?id=...TU1DGA089KLQNNC


thank you


filip

Sorry Filip,
Ive recycled it, maybe someone who downloaded it can help