New Sony PXW-FS5

[Mikko Topponen]

Originally Posted by Terence Morris

It's 50 Mb/s, NOT MB/s - a very big difference! :-)


----I don't pretend understand much of that in any depth (but promise to educate myself accordingly) - and with respect, I do trust what my eyes tell me; whenever I have seen footage 4:2:2 vs. 4:2:0 even crushed somewhat on Vimeo, it positively pokes me in the eye.

Everything on Vimeo / Youtube / TV / Everywhere is 4:2:0. So even if someone shoots 4:4:4, by the time it reaches Vimeo, it will look exactly as anything shot in 4:2:0.

4:2:2 only really affects color edges and outlines (especially in animation).

[Mike Watson]

Vimeo / YouTube / TV is < 8 mbps. So by that logic, we should abandon these 50 and 100 mbps codecs in favor of something that's 8 mbps because there's no sense in delivering more than that.

In reality, the latitude of color correcting and editing in a more generous color space is what we're after. Even if you compress it when you're done, you had it when you needed it.

[David Heath]

Originally Posted by Mike Watson

Vimeo / YouTube / TV is < 8 mbps. So by that logic, we should abandon these 50 and 100 mbps codecs in favor of something that's 8 mbps because there's no sense in delivering more than that.

Two things. First, it's wrong to equate colour space with bitrate compression. Colour space is a very defined one thing or another matter. You either have chroma resolution equal to luminance or you don't. And if you don't, it's a defined ratio to it, end of story - bitrate compression is far more complex.

And as far as bitrate goes, then simple numbers don't tell anything like the whole story. What's the codec? Long-GOP or I-frame only? If long-GOP, what's the GOP length? How complex is the encoder, and is it able to do any sort of 2-pass encoding? Take all that into account, and it's PLAUSIBLE that something around 8Mbs COULD rival in quality terms the same material encoded at much higher bitrate. Plausible if the encoding used had access to a high level encoder, and could make use of tricks like 2-pass encoding, or the complexities found in a broadcast level hardware transmission coder.

But such is not likely to be found in a camcorder capable of working in near real time with power consumption limits, and yes, in principle you want acquisition to be capable of allowing for concatenation through the post and transmission process.

Originally Posted by Mike Watson

In reality, the latitude of color correcting and editing in a more generous color space is what we're after. Even if you compress it when you're done, you had it when you needed it.

Again, I agree with the broad sentiment, but where do you draw the line? Ideally, you don't want anything less than 4:4:4 uncompressed, if we take the argument to the limit? In practice, that's normally not feasible, and the datarate has to be drastically reduced.

So we're into the land of compromise. And if we decide to accept a degree of subsampling, then why 4:2:2? For a fixed datarate 4:2:0 will mean less samples - so a lower degree of data compression. And - arguably - the lower compression will make more difference than a higher colour space. Maybe. It comes back to the original point. 4:2:2 colour subsampling came about because of interlace TV, when it's usage is highly sensible. For progressive systems it's an anachronism. For most (progressive) use 4:2:0 is the best compromise, when the very best is needed it makes sense to go to 4:4:4. What's the point of accepting horizontal subsampling, but not vertical!?

Terence - to give you a little something else to think about, then the early digital component systems were 4:2:2, but it's interesting to think what happened when cheaper (digital) formats first came out. And the DV variants are the most obvious. They wanted to reduce the chroma sampling by 50% - but how to do it? Two obvious possibilities, either further reduce it by 50% horizontally, have one chroma sample per four luminance (which is 4:1:1) or halve the vertical chroma resolution (which is 4:2:0).

Initially, the decision was to do the former, and in the NTSC system all the DV formats are 4:1:1. The reason is that this can withstand repeated analogue-digital conversions much better than 4:2:0 (the latter tends to eventually give vertical chroma smearing) - and initially the recorders were seen as digital "islands" in an analogue world.

But that didn't last very long, and in the all-digital world 4:2:0 was rapidly seen as the better compromise. And because PAL equipment always came after NTSC equivalents, this enabled the spec for PAL DV and DVCAM to be set at 4:2:0 - not 4:1:1. (Sometimes it's better to be a little behind...... :-) ) This became even more significant as regarding all the delivery methods (DVD, digital broadcast etc) were specced to be 4:2:0. Start off with DV and 4:1:1, and the final result will be a worst case scenario approximating to "4:1:0". Not good! But in the PAL world, start off with DV, and it's 4:2:0 all the way.

What this meant was in NTSC territories, "4:2:2" was rightly seen as important - but because it gave a final true 4:2:0 and not "4:1:0"! Does that make sense? 4:2:2 was desired not because of what it was/is - but because it's NOT 4:1:1 ! :-)

Yes, that's all in the past, but with 4K, so is the whole concept of interlace, and so too should be 4:2:2. But old habits die hard.

[Andy Wilkinson]

I may have remembered it wrong...but I am sure I saw a video where someone (from Sony, I think) said that the FS5 does two-pass encoding. It got my interest as I thought this unusual in a camera - but if true highly desirable!

[David Heath]

No, I remember something like that as well. I'd like to know the detail behind it, because two-pass encoding normally means having the entire material available to analyse before the actual encode - so bits get allocated to scenes which most need them. Easy enough if you're encoding such as a DVD and all the file is available - but on a camera, working in real time.......?

I can only think it must be using a buffer, and delaying encoding by the duration of the buffer, which would give a chance for limited forward analysis before the encode? But normally, I'd think two pass and real time coding to be exclusive?

[Andy Wilkinson]

David,

Found it! Watch from just after the 6-minute mark in this video, about 6 min 20 secs to be exact.

In this video Juan Martinez, Senior Product Manager of Sony clearly says that the Sony PXW-FS5 "uses dual-pass encoding so the quality is incredibly high...none of our competitors use this same technology...so even though it's still H.264 we can achieve a different level of performance."

I was very excited to hear this when I first watched this video some time ago. :-)

Newsshooter at IBC 2015 - Sony PXW-FS5 briefing on Vimeo

[Mike Watson]

Roll call - to see how many FS5's we have on order in this forum. I have one on order, and as we creep closer to the end of October, I get antsier every day!

[Andy Wilkinson]

Ordered mine the day of the announcement.

[Mike Watson]

Some new video today:

Sony PXW-FS5 very quick test with Ronin M on Vimeo

[Jeroen Wolf]

Nice colors, nice footage! Dynamic range seems to be pretty good in the corridors. Overall a good look.

[Mike Watson]

It sold me as much on the Ronin M as it did the FS5. ;-)

[Jeremy Cole]

Yes, the footage of the Ronin M was very impressive. The wide angle lens, less so, but the quality of the footage was very nice. Camera appears to mate well with the Ronin. It is a very versatile camera for its size and price.

[Tom Gresham]

Originally Posted by Mike Watson

Roll call - to see how many FS5's we have on order in this forum. I have one on order, and as we creep closer to the end of October, I get antsier every day!

I have two on order. Placed the order the first day possible. Wish I had them now -- on a week-long shoot in Arizona, and they sure would come in handy here.

[Brian Rhodes]

Originally Posted by Mike Watson

Roll call - to see how many FS5's we have on order in this forum. I have one on order, and as we creep closer to the end of October, I get antsier every day!

I have one on order going to use it as a B camera to my FS7.

[Danilo Del Tufo]

New FS5 Footage (it starts at 10:28 minutes to the end of the clip, while before, when the guy speaks there are just few seconds of footage) :

SONY pxw-fs5 필드리뷰 (4K Ver.) - YouTube

Part Two:

소니 PXW-FS5 핸즈온 리뷰 (4K Ver.) - YouTube

Best Regards,

Danilo Del Tufo