PMW-350 or EX1 with nanoflash

[Tom Roper]

Originally Posted by Tom Roper

Assuming I do not want to do color corrections or grading of the PMW350 footage, am I worse off accepting the artifacts from 1200 mbps (pre-compression) capture to 35 mbps EX 4:2:0 codec ported to Blu-ray disk from SXS without further transcoding?

or

Why would Nanoflash 100 mbps 4:2:2 transcoded to 35mbps 4:2:0 Blu-ray be any better that just going straight EX 35mbps 4:2:0 to Blu-ray, since the latter was internally sourced before compression from 1200 mbps 4:2:2 and doesn't require transcoding, while the former was only sourced from 100 mbps 4:2:2 and does require transcoding?

Yes, side by side of course the Nanoflash 100 mbps 4:2:2 looks better than EX 35 mbps 4:2:0! But when I transcode the Nano to Blu-ray 35mbps, will I not just be adding back into the Blu-ray encode all the motion artifacts I just took out with the Nano?

I'll answer my own question here, it appears that when I transcode the 100 mbps 4:2:2 Nanoflash file to a Blu-ray supported mpeg-2, 35 mbps 4:2:0, I am adding right back in the some of the same motion artifacts I took out with the 100 mbps stream, as compared to the "native" XDCAM-EX 35 mbps 4:2:0 HQ file.

In other words, the native file written to the the SXS card is basically as good as the file that got transcoded from the Nanoflash to be Blu-ray compliant 35 mbps 4:2:0. This goes against what a few of you have observed, therefore I am not sure you were comparing to the "Native" file from the SXS card. I use "Native" in the strictest sense, that the only processing to the SXS file is the stripping away of the mp4 or mxf file headers to reveal the mpeg-2 long gop 35 mbps HQ 4:2:0 stream, or in other words, a 1:1 file copy.

There's no comparison whatsoever to the 100 mbps Nanoflash file, but you can't port that onto to Blu-ray, the bit rate is too high for the format.

What does seem to help, is a two pass re-encode. And there is a substantial improvement for 1080/60i interlaced files, not as much for 1080/p progressive files. Of course if the footage is to be handed off for color correction or grading or broadcast distribution, little doubt the Nanoflash files are far superior.

[Paul Cronin]

Interesting discovery Tom. I will have to try this test myself when time permits.

[Piotr Wozniacki]

Tom,

Your observations is something I was keeping in mind when I first made my decision to purchase the nanoFlash. Even though BD is indeed limited to 40 Mbps 4:2:0, and so are most NLEs we're using (for instance, the highest mxf quality output from Vegas is 4:2:2 50 Mbps, while MPG2 is 4:2:0 of up to 80 Mbps)...

Printing to film and other special uses aside, for us video mortals it's still great to start with 4:2:2 of 100 Mbps or above. Heavy grading, keying etc. is just one reason. Frankly, I don't expect BD to remain the ultimate delivery format for a long future - sooner or later, we'll be given some greater bandwidth one.

In the meantime, even if I'm to use some computer-based players as my "delivery" means, I'm OK with it. I'd therefore expect NLEs to be able to output MPG2 flavours in 4:2:2 and at 100 Mbps at least - this has a higher priority with me than any optical media being more capable than the current BD is (though the two are mutually dependent, of course).

[Tom Roper]

I agree with that. And while I sound like an advocate for native smart rendering, I like my Nanoflash very much, do not feel like a wasted expense at all. It's benefits are wide, whereas smart rendering of native files has more limited application, event, corporate, eng etc.

I also feel smart rendering is not very well understood, that while some NLE's like Vegas can do it, Blu-ray doesn't recognize mxf until that header is stripped away, and the only way some people are doing that is by the usual method of rendering the output to mpeg-2, throwing away all the benefits in that one step, perhaps unknowingly.

Did you know that Vegas will smart render to 100 mbps from the Nanoflash 100 mbps 4:2:2 with no recompression? It will report the output as 50 mbps because Nanoflash changed the file header, but the output is 100 mbps 4:2:2 when smart rendered from the 50 mbps 4:2:2 XDCAM 422 template. You may have to check an option box in the preferences to enable smart rendering. So in other words, you can join Nanoflash clips virtually losslessly, and Vegas will only render a few frames around the transistion points while copying the rest.

As for the future of Blu-ray, it is true I am committed to it because we don't have a better distribution format. I personally feel it will go obsolete about the same time as regular DVD, and not really outlive it. Blu-ray may never achieve mainstream acceptance, but done properly looks stunningly good, so it's clear to me that if it can't look great at 40 mbps h.264 AVC, the problem lies not within the constraints of the format. I don't believe there will be a big push for another high bandwidth mainstream format for a long time, rather the i-pod generation seems more interested in personal sized portability, whereas home theater and perhaps HDTV seem the domain of an aging populace. (Like me, a fossil!) Just my $0.02

[Tom Roper]

Just an addendum...

We already really do have potentially good alternative to Blu-ray from the burgeoning growth of low cost, high speed USB flash drives. The problem for J6P lies in learning about the myriad of codecs, software players and fast processors required. You can hand it to him on the flash, and expect the question, "How do I play it, or it won't play for me, or it's all jumpy." Computer monitors while closing the gap, as a group generally don't have the cinemotion playback features of HDTV monitors, while the latter are expanding in versatility with flash and network ports which is good. The power demands of some high end PCs are getting out of hand, what with 850-1000 watt power supplys and fans, this is not very green unless we are also using it to heat the room.

[Piotr Wozniacki]

Tom,

Speaking of smart rendering mxf in Vegas:

Yes, I'm aware I can get a 100 Mbps file smart-rendering nanoFlash 100 Mbps clip - but it works with VP 8.0 only; with 9.0c (both 32 and 64 bit) all I'm getting is audio... There must be some problem with how Vegas Pro 9.0 handles mxf, as each nano clip starts with a green frame when put on timeline.

Can you confirm that?

I also agree it's sad that authoring apps (like DVDA) do not accept mxf...

[Tom Roper]

Piotr,

I'm sorry. I don't have the version 9. I am still using version 8.0(c) running under Vista. Don't you remember me telling you I'm a fossil?

For me there is no problem with mxf or getting a green frame on the timeline.

Also under 8.0(c), DVDA 5.0 does accept mxf for input, it just doesn't smart render it. There are other authoring apps that will however.

[Piotr Wozniacki]

Originally Posted by Tom Roper

Piotr,

I'm sorry. I don't have the version 9. I am still using version 8.0(c) running under Vista. Don't you remember me telling you I'm a fossil?

That makes two of us (fossils) ;)

Originally Posted by Tom Roper

Also under 8.0(c), DVDA 5.0 does accept mxf for input, it just doesn't smart render it. There are other authoring apps that will however.

Interesting what you are saying about DVDA - yes, mxf is accepted, but requires re-compression even at 35 Mbps 4:2:0 (which is a pain in DVDA not only due of no smart-rendering, but generally DVDA renders slower (using just one CPU), and lower quality (at least with MC MPG2)).

As to the 9.0, while it brings a couple of nice improvements, handling mxf (not just from nanoFlash - also EX native ones) is flawed. Smart-rendering or even simple timeline playback causes memory leaks (the RAM usage goes up continuously at the rate of some 0.1GB/sec, which eats all my 8 GB RAM at no time, then freezes the whole system).

And indeed, the green frame problem is non-existent in VP 8.0c.

[Tom Roper]

Piotr,

If you don't want to recompress mxf in DVDA, there is a workaround. I've done it many times, and it only works on the 35 mbps mxf you get off the SXS due to the bit rate being too high for the ones coming from the Nanoflash.

But what you do, is use Vegas Pro to encode a separate elementary stream for the audio, usually an AC3 5.1 file. Then you use a program to strip away the mxf file headers, leaving you with just an mpeg-2 file with the file extension .m2v. The .m2v file is accepted for input by DVDA. You point DVDA to the 5.1 dolby audio file as well as the .m2v video file. Both of these files will smart render from DVDA without recompression, at 2-3x faster than realtime. The only thing that will slow down the process, are your menus etc. If you delete them, you'll see how fast it really flies.

I'm going to email you the utility for stripping away the .mxf headers. It will separate the .mxf into (3) native essence files, (1) video and (2) audio. Uncheck the boxes for the audio files, you can't use them. Instead use just the .m2v video file it creates, and the ac3 5.1 audio file you created in the earlier step with Vegas.

I guess you will have to use 8.0(c) to do this. I assume you still have it. That's a shame btw, that version 9 is broken like that.

[Tom Roper]

Piotr,
Email sent

[Piotr Wozniacki]

Thanks Tom.

I remember I did use Snell&Wilcox utility over 2 years ago when I first came into contact with mxf files from my EX1; now, I must be a real "fossil" as I all forgot about it since:)

Probably because it never actually installed as an Explorer plug - I used it in stand-alone mode.