Sony FDR-AX100

[Dave Blackhurst]

Anthony -

When you use D90 and "broadcast quality" in the same sentence, I have to scratch my head and laugh... you say my "analysis is limited and childish"... not to worry, duly noted, keep it up and you'll find out where it ends up. Or was that the third swing and whiff?

I think you've already lost ANY AND ALL credibility you may think you have, so you go right ahead and adjust that tinfoil hat, please do yourself a favor and keep the conspiracy theories to yourself, along with the condescension and insults. OK?


We all know there is a difference between 30p and 60p - DUH! What we are discussing is what the practical implications of that are, not some crazy theory that the manufacturers could give us such a camera right now, and aren't doing it out of some whacky evil conspiracy to deprive us of the "perfect" camera of our dreams that they could sell to us for "peanuts"...

Think for a moment, if they COULD release this theoretical "dream" camera with "every" feature, don't you think they'd sell in huge #'s, right NOW, no need to trickle cashflow in over multiple years by selling "deficient" products improved incrementally here and there to try to sell product!?! Sign most of the DVi'ers up for three or four of these "dream cameras"... at least. I'll sell a few of my "crippled" cameras right away! Or NOT... I'm sure we'd all probably buy the "deficient" AX100 that didn't exist 10 years ago over the "deficient" HC1 that did, aside from the simple fact it DIDN'T EXIST... now it does, for the same MSRP... and the HC1 was not bad, in it's day!


30p 4K may turn out to be more usable (or less) than expected... many of us have come to appreciate/prefer 60p, at least as far as 1080 goes, so we have reservations. Until cameras are in hand and tests run, we can't know for sure. There are numerous practical (and factually based) reasons that have been discussed here for the "limitation"... and practical discussions of how to deal with it (or not). You might consider that's what it's about, using the equipment that's available to the greatest extent possible...

[Cliff Totten]

60p is fine. Yes, it's cool for doing slo-mo effects and all that. And, yes, if you are shooting sports were you are using faster shutter speeds, you can see some nice motion with it.

For me? 60p is not that important. For me, 29.97p is the sweet spot. (I dont like 24p)

I shoot allot of indoor events. Many times I'm stuck in low light situations, or situations where I'm not allowed to control the light. Other times, I just want to run with the lowest gain that I possibly can. So,..a 1/30 shutter is my speed of choice. This of course, renders 60p completely useless to me.

Where do you play TRUE 60p? Youtube?, Vimeo?, Blu-ray?, iPad?, Android?, Projector?

Anything run over HDMI 1.4 and lower is likely running 60i. This is where 29.97p shines. (29.97 PSF inside a 60i signal looks great)

Unless you have a specific "motion" need for 60p to use slo-mo effects or you really want to shelf stuff for future use, than I don't see a huge need for it. For today, you are really just going to drop that 60p down to 24p, 30p or interlace it to 60i anyway. (This is a "general" comment and yes, there are some exceptions out there, I know)

CT

[Phil Lee]

Hi

There is some sample footage from https://vimeo.com/groups/226931/videos/87997397 where the original can be downloaded.

Note that it has been via an editor and re-encoded, and is some strange hybrid of frame rates, on checking the file it has original frame-rate at 25fps, and frame-rate being played at is 23.976fps, this coupled with high shutter speeds (no 180 degree shutter rule here) I found it nauseating to watch due to the juddering and strobeing. 4K at slow frame-rates in the hands of someone who doesn't know what they are doing is not pleasant to watch.

There is a very strange jelly like affect over the whole image, a mixture of rolling shutter and hunting focus maybe, it gives the impression that nothing is solid, like it's all printed on a sheet of flat plastic pulled tight but not enough to stop a breeze from causing slight undulations. I'd like to see some footage shot with the camera on a tripod and imagine stabilisation turned off as it could be that causing it.

Lots of compression artifacts over complex scenes and moire, although these may have been added by the re-encoding.

Be interesting to see what others think, however I'm getting tired of watching juddery 4K at 24fps already.

Regards

Phil

[Ken Ross]

Phil, can you provide a timecode for where you're seeing compression artifacts? Many who have seen this video have not seen these artifacts you're mentioning. I believe only Cliff mentioned it too and I also asked him to provide a timecode, but I haven't gotten one from him.

I looked at the new downloaded version and saw no compression artifacts. Even the numerous tree limbs swaying in the breeze in front of the fir tree, showed no breakdown at all...and man, I watched carefully on my 23" HD monitor.

I just took another look, this time specifically looking for moire. I don't see any here either. There's plenty of brickwork where this might show up, but I see none. None in the foreground, none in the background. I'm baffled. I've been plagued with a number of cameras that suffered from moire, but in this footage I just don't see it. Again a time code and reference to an object would help.

So either I've totally missed this or I'm totally insensitive to it even though I've seen breakdown in numerous other cameras on this same monitor and setup.

As far as the jello is concerned, again I see it only at the end. I think it was mentioned somewhere that the shooter hand held it throughout and was not feeling well when he shot it. So perhaps what you're seeing is a shaky hand, but I see no jello until he began to frenetically wave the camera back & forth. Even with his hand shaking, I can't say I see evidence of jello...until the end. We don't even know if he used OIS.

I'll hopefully have the camera a week from today, but I see no signs thus far that I'll be impacted by artifacts. My biggest concern remains the stability of the OIS utilized in the AX100. To me that's the weak point of my RX10.

[Ron Evans]

My reason for using 60P is to get the smooth motion of 60i. I find 30P is better than 24P but at a slow 1/30 has too much motion blur for me. I shoot 60P with most of my cameras even though output is always 60i as this is the same temporal motion just half the vertical resolution.

Ron Evans

[Cliff Totten]

I seem to have a love/hate opinion with this video. At times is very VERY eye popping. while other times it seems to have a very "un natual" motion to it.

So here is what I see:

0:07 - 0:14 Straight vertical lines on the tall building on the left. It has what I could call a "micro-wobble". A very fast "fludder" on the long vertical edge wall.

0:30 - Vertical line in the dead center of the screen when the camera moves.

2:01 - "heat" shimmering affect appears to be amplified in a "synthetic" way by rolling shudder? It just seems very like an odd distortion that is more than a "normal" heat shimmer. (if that makes any sense...lol)

Overall,..the best way I can describe it is "un natural motion" with a "micro wobble" or even "micro skew"

Could this be just the frame rate conversion? Maybe.

I'm not going to judge the camera on this one video. I need to see allot more in true 29.97p first. I need to see it raw off the camera and not converted for the web.

I'm getting one!

CT

[Phil Lee]

Hi

Just watch the first 30 seconds, it is as clear as day the encoding is suffering. Note that watching it downsampled will hide a lot of the issues.

Use VLC Media player, play the download clip, right click on on the playing video and unselect 'Always fit to Window' and it should now be playing at full size.

You can clearly see the encoder struggling and see detail vanishing then reappearing. At around 24 seconds look at the stair case and the detail coming and going with very slight camera movement, look at the balcony edges detail come and go, because the encoder hasn't enough bit-rate to cope with the movement.

The fruit at the 1 minute mark, you can watch the mosquito noise all over the fruit.

At the 1:27 mark look at the railings and bricks, covered in compression artefacts that flicker in and out.

Look at 1:47 with the building with scaffolding and the crane, looks a complete mess.

Look at 2:19 with the trees, a complete mess of compression artefacts. I've attached a 100% crop from 2:19. I could go on.

If this was HD no one would be interested, it is horrendous. Because it is 4K that most people are viewing on small monitors and downscaling everyone is having some sort of mass hysteria for it. It's laughable really to hear people go on about this, no compression artefacts, really?

It is what it is, 4 screens of poor quality 15Mbits/sec HD stitched together and shrunk down to hide all the nastiness.

As for the funny wobbling effect, it's all over the place, you'll see it more at full screen resolution.

Regards

Phil

[Troy Lamont]

I'll repost what I posted earlier because it's still applicable.

I can definitely see a shaky hand, but the user admitted to and apologized for it saying he had a cold. I don't think the OIS was enabled either, my take.

Shaky hand =/= jello effect.

Looking at the cat around the 3:22 mark, that's just shaky cam footage, zero jello. I've actually never seen that effect unless the camera was panned too fast and that isn't happening in any of the footage until the end again. Meh.

I'm downloading the 4K version and I'll take into consideration the possibility of a jacked up encode error for what could be considered a newbie in the 4K arena. I'll watch it on the Samsung F9000 4K set and I'll post my findings.

I can't wait until this thing comes out so we can put all this middle-man stuff and differences to bed. Talk about beating dead horses and there are SOO many factors to take into consideration when judging the footage so religiously. One of the biggest if the software/hardware that's it's being played back on.

Again I'll reiterate that it looks like the OIS wasn't enable for the handheld stuff. I own 3 Sony Cams (TRV-25, FX-1000, HD-70U) and Sony does a great job with OIS on it's handhelds and although I don't have a 4K cam other than my Galaxy Note 3, I can't see Sony missing the mark on OIS this bad. I'll pose that question to the original shooter as well.

[Dave Blackhurst]

The first time I watched the "cat" video (You Tube IIRC) there was some seriously nasty "stuff" starting at around 2:16 - where the trees were moving around in the wind. I've since watched it on Vimeo, and it looks fine in those same spots. I saw similar artifacts one other time, then couldn't replicate 'em.

There are LOTS of points in the "data chain" to goof something up or for something to not work as expected, and 4K is "new". I'm only set up to "properly" display 1080, and even then run into problems sometimes with one program or another "acting up"...

THIS is why a camera that can shoot 4K is a "toe in the water"... which will no doubt be followed ay another "toe" (new computer build to better handle the higher res clips), and another "toe" or two (4K capable screen(s).... another "toe" (big honking HDD's to store clips).... and so on! I remember how the HC1 wasn't an "isolated" purchase, nor were AVCHD cameras... this round, I'm trying to plan ahead for the entire set of "investments" that are no doubt coming!

If the quality turns out to be "good enough" with 30p, it might be a bit of a blessing in disguise, I've been spec'ing out new components for a computer build, and expect 4K to require some serous "muscle", even at 30p - with phones and tablets having reduced computer sales, high end components don't come as cheap as they used to!!

[Phil Lee]

Hi

The facts are we have 4K with each HD quadrant only getting 15Mbits/sec on the AX100, and that is exactly the quality I see, perhaps worse given the encoder can't make as much savings in motion search given it has to encode the equivalent of 4 HD images at the same time.

The main problem is everyone is viewing 4K down the wrong end of a telescope essentially, you can take anything and shrink it down and it appears to look sharper and more detailed. 4K isn't about shrinking down though, it's about a future of much bigger screens and enlarging.

It astonishes me that people are saying there are no compression artefacts in this Sony AX100 footage, but clearly it is riddled with them, again it is looking down the wrong end of the telescope.

Hopefully the original footage is better and the worst of the compression artefacts are due to poor encoding later, but we still can't get away from the facts. To have AVCHD quality but in 4K would require 28Mbits/sec * 4 = 114Mbits/sec encoding, and many would argue that AVCHD quality is already pretty poor. As the encoders are little more than HD encoders with the LSI overclocked (we've not seen new silicon yet, that will arrive with H265 hardware encoding), on 4K they are even less efficient.

So 15Mbits/sec per HD quadrant, it is what it is, and was always going to be a struggle for any camera. Perhaps the GH4 will be a little better with 25Mbits/sec equivalent per quadrant, then again, that still isn't that many bits per second.

Given the Sony AX100 will record HD in 50Mbits/sec at a smooth frame-rate of 60fps, that will probably be a lot more pleasing.

Regards

Phil

[Dave Blackhurst]

Cliff and Phil have some valid points where "something" seems to be happening, hard to say what that "something" is... High zoom (and we don't know whether CIZ or digital is involved) seems to show more "issues" - some of which could be, as Cliff suggests, "atmospheric" distortions you'd never "see" until you start shooting super high resolutions and high zooms over distance.. Things you normally wouldn't "see" with your nekkid eyes will definitely look "un-natural"...

I've wondered a bit about how well say a wedding video will be received when EVERY detail (and flaw) will be glaringly apparent. That's one instance where sharpness might be a liability!

I think that that section with the tree branches in the wind would probably stress a codec quite a bit, sort of a "worst case scenario", which leads me to a different take on all this...

What sorts of things might be interesting enough to shoot that would stress a codec to the breaking point?
Are really "busy" scenes what we will be shooting? Or will "real life" shoots be a bit less stressful and more forgiving? If using a tripod or monopod vs. handheld will help, that should be easy enough to resolve.

I'm trying to think in terms of "real life" shooting situations, and real life image quality expectations - I see FAR worse artifacts and noise all the time - the news is a BIG "offender", despite being a "big city, network" news station in a brand new facility... While WE might see every possible flaw, will anyone we're shooting for (other than ourselves) even notice? At the moment, I'd say if camera work is good, and the content is of value or interest, most people won't be picking the image apart looking for the few "bad" seconds in a clip. It's like having a couple out of focus still shots out of dozens of GOOD ones - doesn't make for a "bad" camera, as long as it's not a big pattern or persistent problem.

[Ken Ross]

Originally Posted by Cliff Totten

I seem to have a love/hate opinion with this video. At times is very VERY eye popping. while other times it seems to have a very "un natual" motion to it.

So here is what I see:

0:07 - 0:14 Straight vertical lines on the tall building on the left. It has what I could call a "micro-wobble". A very fast "fludder" on the long vertical edge wall.

I don't see it. I see just a bit of movement that I'm attributing to hand holding. Nothing that I haven't seen with every camera that's hand held.

Originally Posted by Cliff Totten

0:30 - Vertical line in the dead center of the screen when the camera moves.

Again, to me it looks like nothing more than hand held movement.

Originally Posted by Cliff Totten

2:01 - "heat" shimmering affect appears to be amplified in a "synthetic" way by rolling shudder? It just seems very like an odd distortion that is more than a "normal" heat shimmer. (if that makes any sense...lol)

This one I replayed 4 times before I saw anything resembling what you saw. I saw a trace of it, but the lens looks like it's at full telephoto. I often see effects like that at full telephoto with a number of cameras, though it's usually much worse. I also noted vents on the roof of that building, so it could well have been heat waves.

Originally Posted by Cliff Totten

Overall,..the best way I can describe it is "un natural motion" with a "micro wobble" or even "micro skew"

Could this be just the frame rate conversion? Maybe.

I'm not going to judge the camera on this one video. I need to see allot more in true 29.97p first. I need to see it raw off the camera and not converted for the web.

I'm getting one!

CT

Yeah, man, I don't want to say you guys are truly, really, micro-analyzing this stuff. I mean do you guys actually watch your videos like this??? If so, you'd never see the content you shot! :)

[Ken Ross]

Originally Posted by Troy Lamont

I'll repost what I posted earlier because it's still applicable.

I'm downloading the 4K version and I'll take into consideration the possibility of a jacked up encode error for what could be considered a newbie in the 4K arena. I'll watch it on the Samsung F9000 4K set and I'll post my findings.

I can't wait until this thing comes out so we can put all this middle-man stuff and differences to bed. Talk about beating dead horses and there are SOO many factors to take into consideration when judging the footage so religiously. One of the biggest if the software/hardware that's it's being played back on.

Again I'll reiterate that it looks like the OIS wasn't enable for the handheld stuff. I own 3 Sony Cams (TRV-25, FX-1000, HD-70U) and Sony does a great job with OIS on it's handhelds and although I don't have a 4K cam other than my Galaxy Note 3, I can't see Sony missing the mark on OIS this bad. I'll pose that question to the original shooter as well.

I've spoken to two guys who viewed this on 4K TVs, one on a 65" Sony UHD TV (the other I'm not sure) and both saw none of what Phil is reporting. I know Phil has never been a fan of this cam, that's no secret, but again, I think this is micro-analysis that no camera would stand up to. When watching on a typical large screen UHD TV, we sit at a normal viewing distance of 8-10' for many people. I watch my HDTV at that same distance. At that distance you just don't see this even if it did exist and was of the magnitude that Phil is claiming. I just don't get the concept of pixel peeping with video. Who does that? Not me at least.

And good God, the 'wobbling all over the place' is precisely the effect of hand holding a camera and not doing a good job at it. I see it with every single camera that's hand held and either the OIS is off, there is no OIS or the shooter just has a shaky hand. This is nothing to do with 4K. This has nothing to do with jello and it has nothing to do bitrates. It has everything to do with a steady hand or lack thereof. No mystery in my mind.

This camera is not perfect, no camera is, but as bad as Phil thinks...I don't think so. Phil continually refers to the bitrate per quadrant as if we're using the same codecs we always have. This is a different codec, that's more robust and doesn't need the bitrate of past codecs used in cameras like this. ;)

[Dave Blackhurst]

One more reason for 30p vs. 60p - 30 frames per second should be less data than 60 frames (for instance the RX10 has a top bitrate of 24Mbps for 24p vs. 28Mbps for 60p, the next step is a 17Mbps 24p - not far off from 15...). It may not be a LOT less data, but it should reduce the load somewhat? I think it's safe to say "there's no free lunch" when you start dealing with a lot of data bits. You either have to accept some compression losses, or deal with huge files and data handling issues.

I'm sure this is also part of the design tradeoff that went into this camera, and will of necessity go into every other 4K camera. I'm not sure it's "fair" to judge a camera that's a first of its kind on where it "fails" vs. what it does do? So far what I've seen looks promising, and if it's half as good as the RX10, it'll make a nice dedicated video complement to that camera. I'm sure there will be a few "open box returns" available at discounts when some people discover it doesn't bring them a chilled beer on demand...

[Ken Ross]

That IS true Dave and something I forgot. This IS 30p and requires less of a bitrate to avoid artifacts as would a 60p video.

Coupling that with the fact this a new codec, more robust than AVCHD, more efficient than AVCHD and you can begin to see why it's simply not fair to say it's a fail because it's 15Mbps per quadrant. Apples to oranges.