Sony FDR-AX100

[Ken Ross]

Originally Posted by Ron Evans

As you all know I want 60P and was looking at proDAD reSpeeder as a way of interpolating a 60P stream from the 30P from the AX100. My application is a fixed camera of a stage event so no camera movement once set and I thought this may work. So I started to look into how much better in low light the AX100 would be compared to my AX1. These calculation are VERY rough. The sensor of the AX100 is about 9 times larger in area than the AX1. It has 14Mpixel compared to 8Mpixels for the AX1 so each pixel will gather about 5 times the light of the pixels of the AX1. One would think this would give a large advantage but there is a catch. At full wide the lens on the AX1 is F1.6 and on the AX100, which is wider, is F2.8. So for the same field of view the AX100 would have to zoom in a little to compare, ramping the lens to more than F2.8. The difference could be about 4 times more light going into the AX1 as into the AX100 !!! So the real world performance may only be 1/3 to a stop faster than the AX1. Your comments on this simple calculation are welcome. Have I got it all wrong? A big sensor needs a big fast lens which means size and cost. I think the AX100 will be a real winner for Sony but wonder if expectations are too high for performance.

It will be interesting to see the real tests when they appear.

Ron Evans

Ron, I think it's pretty clear the AX100 isn't for you based on all your posts here.

With that said, and from the footage I've seen thus far, I think you're totally wrong about expectations being too high for performance. The footage looks great and frankly, better than what I've seen from some more expensive cams.

But inevitably, one has to ask in response to your statement, whose expectations? For those expecting professional results from a Handicam, yes, they'll be disappointed. That's not realistic. For those like me who are expecting excellent 4K video from a small, (which the AX1 certainly is not) form factor and possibly even better performance in some respects with its larger sensor, the AX100 won't be a disappointment.

If one approaches this realistically, just as I did with the RX10, I'm sure most will be pleasantly surprised, not disappointed in performance. I saw the same kind of doubt regarding the RX10's performance prior to its release, and we now know how wrong that was. :)

[Ken Ross]

Originally Posted by Joey Atilano

Watching this video many times just on my normal 1080p led TV some of the scenes are super sharp and overall the picture quality looks really awesome. I noticed a noticeable amount of shallow DOF.

Couldn't agree more, Joey. We almost have no right to expect this kind of performance in something so small and relatively inexpensive.

The much higher peak bitrates already eliminate some of the 'bitrate naysayers'. There was absolutely no issue with the codec & bitrate keeping up with all the very fine moving detail.

BTW, being as impatient as I am, I'll probably wind up picking it up at the Sony store on the day of release. That's what I did with the RX10.

[Joey Atilano]

The Shallow DOF is an added bonus. I wish I could pick it up locally but it would cost an additional $180 in Tax. I'll have to live through your videos until I get mine Ken lol. I'm sure I'll check it out at the Sony store while it is shipping to me.

[Ron Evans]

Originally Posted by Ken Ross

The much higher peak bitrates already eliminate some of the 'bitrate naysayers'. There was absolutely no issue with the codec & bitrate keeping up with all the very fine moving detail.
.

The video is a 1920x1080 downscale edited in FCPX so the choice of data rates used is with this edit/encode .mov file not the original XAVC-S file. But yes as I have found 30P XAVC-S at 60Mbps if set with 1/60 shutter is just fine with lots of detail and downscales wonderfully. Default auto on my AX1 for 30P sets the shutter sometimes at 1/ 30 and then of course there is lots of image blur especially with any camera movement. I am sure the output of the FDR-AX100 is just like my AX1 in this regard and the codec is certainly not the issue. I really cannot tell much difference between 30P at 60 or 100Mbps. I too can't wait to see the FDR-AX100 when it comes out.

Ron Evans

[Darren Levine]

still taking any online compressed videos with a grain of salt, im not quite impressed with some of those motion shots, but it may not be present in the raw footage. i like my rx10, but it artifacts just like any other avccam in high motion, i certainly hope the ax100 is a step in the right direction.

im a bit behind on the current info. has there been anything about the lens controls being fly by wire, or directly controllable?

[Phil Lee]

Originally Posted by Ken Ross

The much higher peak bitrates already eliminate some of the 'bitrate naysayers'. There was absolutely no issue with the codec & bitrate keeping up with all the very fine moving detail.

Hi

Higher bit-rates are from the video buffer, it's how these things work. Blu-ray can peak to 50 or 60Mbits/sec given a favourable wind but it's still just 40Mbits/sec, and so the Sony 60Mbits/sec is still the maximum read/write media data-rate. The maximum peaks are generally relative to the data-rate so a camera recording at 100Mbits/sec will have even higher peaks and so on. Even DVD has a buffer and can peak above 10Mbits/sec, but we see it more on modern implementations of codecs as memory is cheaper and so buffers can be set larger.

It is diminishing returns of course as no such thing as a free lunch. For every peak above the maximum reading/writing set bit rate, bits then have to be starved by the same amount to avoid buffer under-runs, so this peak is only for a very short time in favourable conditions, typically on a scene change peaks are seen to allow coping with closing a GOP early and inserting extra I frames.

Regards

Phil

[Ken Ross]

Originally Posted by Darren Levine

still taking any online compressed videos with a grain of salt, im not quite impressed with some of those motion shots, but it may not be present in the raw footage. i like my rx10, but it artifacts just like any other avccam in high motion, i certainly hope the ax100 is a step in the right direction.

The downloaded video from Vimeo shows no motion artifacts that I saw Darren. The complex nature of the branches in front of the pine tree or the fine detail of the cat's fur, exhibited excellent motion detail. The YouTube version, not so much.

I'm actually very surprised and pleased how well the detail is holding up with motion.

[Ken Ross]

Originally Posted by Phil Lee

Hi

Higher bit-rates are from the video buffer, it's how these things work. Blu-ray can peak to 50 or 60Mbits/sec given a favourable wind but it's still just 40Mbits/sec, and so the Sony 60Mbits/sec is still the maximum read/write media data-rate. The maximum peaks are generally relative to the data-rate so a camera recording at 100Mbits/sec will have even higher peaks and so on. Even DVD has a buffer and can peak above 10Mbits/sec, but we see it more on modern implementations of codecs as memory is cheaper and so buffers can be set larger.
Regards

Phil

I'm not so sure Phil. I payed very careful attention to the bitrate as the scene complexity and detail changed in the downloaded video. The more complex the scene, the higher the observed bitrate. The most complex scenes (tree limbs swaying in the wind, the cat's fur in motion) showed bitrates in excess of 90Mbps. That's over 30Mbps over the stated 60. To me it looks like a dynamic response to detail as opposed to a buffer.

I would offer further support for this argument based on the fact that I saw no mush or detail degradation at all. In fact, better than my RX10 under the same conditions, and this at a much higher resolution.

The bottom line? Who cares if the detail holds up much better than expected and many here predicted? :)

[Ken Ross]

Ron, you're right, given this was an edit, we really don't know what the native file bitrate looked like.

But as I said to Phil, whatever it is or was, there were no observed motion artifacts and that's all that counts. :)

[Darren Levine]

Originally Posted by Ken Ross

The downloaded video from Vimeo shows no motion artifacts that I saw Darren. The complex nature of the branches in front of the pine tree or the fine detail of the cat's fur, exhibited excellent motion detail. The YouTube version, not so much.

I'm actually very surprised and pleased how well the detail is holding up with motion.

Did you miss 2:26? i think the motion detail is pretty good considering the bitrate, but at that point in the video the breakdown is pretty clear. i've certainly seen worse, but this camera is far from free of it, and i wouldn't expect it to be considering the bitrate

point it at moving water, that's about the easiest way to see if it turns mushy

[Ken Ross]

Nope, not on the downloaded version Darren. There is absolutely no mush at 2:26 in the downloaded version. None. Perfectly clear and perfectly sharp with the branches swaying in the breeze.

I saw mush clearly in the YouTube version precisely at that spot, but it's totally gone when you download the .mov file he posted. The 2:26 on YouTube is due solely to YouTube compression. :)

On AVS someone posted a frame grab of that spot on the YouTube version and thought that the codec had reached its limit. He then downloaded the video from Vimeo as I did and posted later that there was no mush in the downloaded version and it had considerably more detail. He correctly attributed the mush to YouTube compression. The difference is dramatic.

[Darren Levine]

maybe we're talking about different things... because yes i downloaded it, and have viewed it blown up, and yes there is motion artifacting, as one would expect from high compression. it's not bad, but yes it's there. it appears to be a good performance, but you cannot call it artifact free.

[Ken Ross]

Darren, perhaps I missed it since I watched it full screen only without pixel peeping. I never 'pixel peep' videos or zoom into them. To me I don't see the point since that's not how I watch videos. A few videographers I know watched it and also reported seeing no artifacts with typical full screen viewing.

Perhaps if you typically enlarge your videos it will show up, but I'm sure most video cameras subjected to that kind of scrutiny would also show flaws.

The other thing here is that we're dealing with a new codec that we're not entirely familiar with. So using our experience founded from compression schemes like AVCHD, may not be an accurate predictor in guessing the degree of artifacting. Thus the assumption 'as one would expect from high compression' might not apply to the same degree as what we experienced in the past. I think this is why many that have viewed these videos have been rather surprised by the lack of artifacts.

The videos I'm seeing are at least as clean or cleaner than my RX10. That's good enough for me. :)

[Noa Put]

Not seeing any artifacts either but I also viewed it in a way anyone else would. Rolling otoh shutter seems quite severe though.

[Paul Rickford]

Found a link here for the AX100 manual

http://support.sony-asia.com.edgesui...4534651111.pdf

This is really starting to look very good, really pleased to see that level of quality down sampled to HD in FCPX as that will be my workflow.