PMW300 XAVC Firmware

[Tim Lucas]

Has anyone installed this? I did last night and had a wee play with it today.

Not sure what i was expecting but didn't notice a mind blowing difference. Sure its not 4k but could notice no difference from XDCam HD 4.2.2. 50 Mbs.

It shows up in Premiere Pro CC as AVC Intra 100. I guess that's what it is, so where does the X come in.

Just trying to get my head around this. Sony promotes XAVC as the new 4K format. So this is half the resolution of 4K with 10 bit, 4.2.2 100 Mbs 1920 X 1080 recording.

Where does this codec fit in the market place then?

Tim.

[Jack Zhang]

People don't like Long-GOP due to it's predictive frames, mostly in higher end productions that may have CGI added later. Intra-only (or I-frame only) records complete frames with no predictive frames, which is extremely useful for VFX work and mastering.

[Andrew Smith]

XAVC moves from an mpeg2 based compression to mp4 based.

It also has a greater bit depth that it supports, I believe. Something like 14-bit, but don't quote me on it. This will be handy for future television/video standards where you can see the difference in the end user TV displays ... which would explain why we probably can't currently see the difference.

I've only shot 16 seconds on the format as a bit of a technical test, and I can't see the difference. Mind you, I've only just moved from HDV to XDCAM 4:2:2 50Mbit and I'm still in amazement at how the gates of heaven have opened up over here.

Andrew

[David Heath]

Originally Posted by Jack Zhang

People don't like Long-GOP due to it's predictive frames, mostly in higher end productions that may have CGI added later. Intra-only (or I-frame only) records complete frames with no predictive frames, which is extremely useful for VFX work and mastering.

It's important to distinguish between acquisition formats and "edit-friendly" codecs. It's true I-frame only may be desirable for heavy post, but a long-GOP format can be transcoded to such between shooting and editing.

There is also long-GOP and long-GOP - it's not just sheer bitrate that distinguishes XDCAM from (say) HDV - it's the way the bitrate is allocated, the way it gets split between I-frames and difference frames.

A good codec should be dynamic, according to the amount of change between frames. Little change means lots to the I-frames, and an extreme case of every frame being different may (theoretically) mean as much to the difference frames as the I-frames! That's why, for all else equal and the same bitrate, a long-GOP codec should always give superior results to I-frame only - there are normally many similarities between frames.

It's also worth noting there are long-GOP versions of XAVC, and I-frame only versions of Mpeg2.

The main driver for the arrival of XAVC is that it caters for resolutions and framerates that aren't in the MPEG2 spec. Most notably 4K and such as 1080p/50 recordings.

My understanding is that at 100Mbs it is very similar to AVC-Intra 100, and may well be considered the same from a decoding point of view - they both share the same underlying technology. But being newer, the encoder has more tricks to call upon to do the encoding, so for the same bitrate it should give a better result than "old" AVC Intra 100.

Where does it fit in the marketplace? I've read Sony documents that indicate they see XAVC and XDCAM coexisting for quite a while to come. If your interest is such as 1080i/25 or 1080p/25, and especially if bitrate is important, and/or the need to perform fairly straightforward edits quickly, there's not much to be gained from XAVC. But if you're looking to 4K, or something like 50p, or need a 10 bit codec, that's where XAVC will more and more come into it's own. (Andrew - the main significance of increasing the bit depth to more than 8 bits is less how the original picture looks on a display, but more the added flexibility it may give for post adjustments - if the camera front end is up to it.)

[Andrew Smith]

Hi David,

Thanks for that. I had that latest Dolby extended dynamic range display technology swirling around in my head as I was writing the post. Just lacking the time to double-check things at the moment. (hence the no-quote disclaimers)

So glad to have a comparatively future-proofed camera. The PMW-300 is definitely money well spent.

Andrew

[Andy Wilkinson]

I must say that my PMW-300 is proving to be a rock solid performer for the type of work I do and produces lovely clean images. I will upgrade mine with this new firmware very soon as I have a few days space between big jobs coming up in the near future.

Too bad we don't get 1080p50/60 XAVC capability though. For sure that will hurt some sales of this cam - but it it what it is... and so far I am delighted that I upgraded to it from my EX3.

[Jack Zhang]

I'm willing to bet the PXW-X380 will have 1080p/50p/60p. The PXW-X280 will be the non-interchangeable variant of that.

We won't see 3-chip 4K, only 3-chip 1080. I'm willing to bet later this year a 1/2'' 3 chip system capable of 1080p/60p/50p will be introduced.

[David Heath]

Originally Posted by Jack Zhang

We won't see 3-chip 4K, only 3-chip 1080. I'm willing to bet later this year a 1/2'' 3 chip system capable of 1080p/60p/50p will be introduced.

Maybe - but the industry looks set to bypass 1080p/50, at least as a broadcast format. Moving to 4K at the same time as 50/60p.

Will we get a 1080p/50(60) camera, or will cameras likewise move straight to 4k with the higher framerates.....? And I don't know the answer to that one...... But if the demand is for 1080i(p)/25 (for current HD systems) and for 4K/50(60)p, is there any point in a 1080p/50 camera?

[Roshdi Alkadri]

I experienced banding in the PMW300 at 8bit. Can anyone confirm that the new 10bit firmware fixed this?

[Andrew Smith]

Do you have some demo footage of the 8-bit banding that you can share?

Andrew

[Roshdi Alkadri]

Hi Andrew, i posted about this under the same sub-forum if you scroll down
http://www.dvinfo.net/forum/sony-xav...0-banding.html

[Andrew Smith]

Just looked at the frame. Is it the tone across the front of the fridge? I notice these are blue-ish in colour, and a possible culprit would be the increased difficulties that sensors have with the blue channel. Just a thought.

Andrew

[Roshdi Alkadri]

Not just that scene, even in brightly lit situations i would get that. I ended up returning the camera. Do you have any low light footage with the new firmware?

[Andrew Smith]

Here's some comparably low light footage. I think the sensor gain was 3db, if at all. CU cam is the PMW-300 and the wide cam is an EX-1. Hot spots are from the lighting which was out of my control, as was also the audio which I tool a feed + iso recording from.

Vimeo

Whilst this video is hidden on Vimeo, you can click through from here and download the original 130Mb (10Mbit encoded) uploaded file if you wish.

Andrew

[Roshdi Alkadri]

Thanks very much for the link. It looks good so far, as you know, vimeo compresses the video further. I couldn't access the download, and i don't want to trouble you. An MXF file would be better, i'm just going to rent the cam for a day or two, update it and shoot away under different situations.

The scenes that i shot which banding was evident included white walls and general solid areas.

Again, i appreciate your help Andrew.