Sony FDR-AX100

[Bruce Dempsey]

nice piece Dave

[Jack Zhang]

Originally Posted by Cliff Totten

Sony seems to have put on hold their traditional practice of "camera crippling" on these two models.

I may beg to differ as the Zeiss lenses on these cameras may have terrible cases of barrel distortion, but the processor is doing it's best to correct it. I do notice on the edges of the picture in the sample video some smearing possibly due to barrel distortion correction. Wonder why the price is so low? It's because they put a crap lens in front of a high-performance sensor and processor combo. It's a more extreme case than the C.A.C. in the EX1.

[Bruce Dempsey]

I think I paid 2000 for a HC1 in late 2005. Using an inflation calculator I googled that's a little over 2300.00 today. Sony may well be utilizing the economy of scale algorithm (lol) and plan on selling these in the kazillions so manufacturing costs although high at the outset will dwindle whilst sales skyrocket. (certainly does not seem to be a cheapo lens...that's for sure)

So what about the pan handle zoom controller? Is it old school lanc, cx series which uses the a/v port or HX series utilizing the micro usb port which btw I have found to be problematic in that mains a/c and pan handle functions which use the same port cannot function concurrently

[Glen Vandermolen]

Originally Posted by Jack Zhang

I may beg to differ as the Zeiss lenses on these cameras may have terrible cases of barrel distortion, but the processor is doing it's best to correct it. I do notice on the edges of the picture in the sample video some smearing possibly due to barrel distortion correction. Wonder why the price is so low? It's because they put a crap lens in front of a high-performance sensor and processor combo. It's a more extreme case than the C.A.C. in the EX1.

"Price is so low?" It's their 3rd highest priced consumer camera, behind the VG series.
And I don't see much distortion, although I don't have a 4K monitor to view the footage. The footage looks outstanding to me.
Sony probably could have put a better lens on it, but not for $2,000.

[Dave Blackhurst]

The VG series actually aren't marketed in the same "consumer" class either, meaning the AX100 is the new "flagship" consumer (they actually refer to it as "prosumer" on the web page) camera.

It's still a $2K camcorder, expensive, but not unheard of in "consumer" cams - the CX900 of course "fills in" the second highest slot at a more "consumer-y" price closer to the retiring 7xx series top of the line 1/2.88" sensor cams. The next 4K price point is what, $4500, and I'm not sold on the performance from a small sensor, vs. what the AX100 brings to the table.

We have to have hands on to know how good or bad the camera will be in practical use, but it sure looks like the bang/$ ratio is in line. 4K, even if only 30p, plus what should otherwise be a pretty good 1080 60p camcorder with full manual and a nice combo of lens/sensor/processor... seems pretty reasonable to me even if there are a few quibbles here or there.

It's definitely going to give the "competition" some fits (oh wait... what competition... when's that GH going to actually see market again?). Sony didn't put their foot in the door, they've kicked it down...

[Cliff Totten]

I agree, assuming that video really was shot with an AX100, it's WAY better than any $2000 camcorder that I know of today. The lens looks fine to me.

The fact that Sony didn't pull the typical "Handycam & NX30" trick of allowing you to control gain or shutter speed but NOT both at the same time....shocks me! It seems they didn't block anything.

Anybody know a better $2,000 camera today? (GH4 might be a challange soon though)

CT

[Jack Zhang]

Originally Posted by Glen Vandermolen

"Price is so low?" It's their 3rd highest priced consumer camera, behind the VG series.
And I don't see much distortion, although I don't have a 4K monitor to view the footage. The footage looks outstanding to me.
Sony probably could have put a better lens on it, but not for $2,000.

I did view it on a Seiki 39'' monitor and the corners look more interpolated than the center of the picture, especially in the boat scene from the sample video.

What would be great is if Sony entered the MFT market with XAVC MFT cameras competing directly against the AF100, but I think we all know they're loyal to E-mount, which leans more towards a APS-C sensor rather than MFT.

[Glen Vandermolen]

That's the peculiar part: Sony seems to have put the better, bigger sensor in their consumer 4K cam, compared to their prosumer 4K cams. I definitely prefer a 1" sensor over a 1/2" sensor. I'll gladly sacrifice zoom range to accommodate the bigger sensor.

[Ron Evans]

Originally Posted by Glen Vandermolen

That's the peculiar part: Sony seems to have put the better, bigger sensor in their consumer 4K cam, compared to their prosumer 4K cams. I definitely prefer a 1" sensor over a 1/2" sensor. I'll gladly sacrifice zoom range to accommodate the bigger sensor.

I would buy one too if I could have 60P. Until then my FDR-AX1 is OK. I think the zoom range is the limiting factor. I want to see how the clear zoom looks too as 18x may be good enough. Certainly the 24x with AVCHD would be great and that is at 60P so does represent an improvement over my NX30U with better control too and may be a good reason to sell my NX5U !!! .

Ron Evans

[Dave Blackhurst]

@ Jack -

Sony ALREADY supports A and E series mounts/lenses in APS-c and now FF, why on earth would you think they would EVER jump on MFT?! They have a potent new combo with the RX/AX sensor and Bionz X processor, they aren't going to go reinvent the wheel and add yet another format in a shrinking market...


@ Glenn -

Sony is trying to (successfully by all appearances) open up some new markets for people who don't want to carry a bag of lenses and accessories, but still want better quality images and convenient to carry cameras. In the process, just as when they released the old TRV900 and HC1, they are creating some tools that border on "professional grade", and might even outperform on some levels, and meet the needs many pros might have for "personal" cameras that don't deliver lousy results!

I'd also expect that perhaps there will be some pro cameras with this sensor/processor once they vett it - the sensor and processor look to have a lot of potential.

[David Heath]

Originally Posted by Cliff Totten

I'm no expert on the math behind "pixel binning". However, it's my basic understanding that binning is simply taking a cluster of pixels, capturing each output values (like "voltage" readings for lack of a better word) and averaging those clusters into one single value. (one value per cluster)

In essence, yes. But always keep in mind the Bayer structure - you can't simply take adjacent values as they will represent different colours. So it's define a group, then take binned values for each of red, green, blue within that group. Since the "values" are electric charge (no of electrons) then it follows each photosite can only be used towards a single "cluster".

Originally Posted by Cliff Totten

The idea or the goal is to get 1920x1080 single (pixel) values from each their nearby clusters.

You have to consider the geometry, and this is where it all starts to get messy (and more difficult to generalise).

Let's assume a sensor of dimensions 4608x2592 in 16:9 video mode - simple maths shows it's virtually 12 million photosites, so consider it typical for a sensor for a primary still camera.

If you were to read to get 1920x1080 values directly, then how would you do the grouping? Divide 4608 by 1920 and you get 2.4 horizontally, same number vertically. But you need to be thinking of whole numbers - "binning" implies reading the charge from a photosite and putting it in a "bin" along with charges from it's neighbours. Hence each photosite can only contribute towards a single value. And in the past it's been essential to simplify, to speed up the readout, to achieve frame rates.

Thinking of Bayer geometry, the basic pattern is a 2x2 block - 1 each of red, blue photosites, and 2 green. In practice, it's been necessary to think about doing the grouping in terms of 4 such blocks - so a 4x4 grouping of 16 photosites (so 4 R, 4 B, 8 G) So each of these becomes a "basic resolution element".

And it therefore follows that the resolution will be the sensor dimensions divided by the block dimensions, so division by 4 in such a case, which here would predict 1152x648 - something slightly less than 720p standard. For output, it would be scaled up to 720 or 1080 format for recording, though obviously the actual resolution could never be better than 648lpph in that particular case.

And measure the results from such "primarily stills" camera sensors, and that's exactly what they do show - in ball park figures anyway, the exact figures will vary depending on the exact dimensions. And that's why such as DSLRs have tended to give "roughly 720p performance", even in 1080p rec mode. But well, 720p isn't bad is it? And to get it from a still camera sensor......?

But times move on.

Originally Posted by Cliff Totten

4608x2592Now correct me if I'm wrong but isn't that the same (roughly) technique that Photoshop and other apps would do when scaling a 20mp picture down to a 2.1mp (HD size) image?

If dis guarding high amounts of image data is the goal, wouldn't a pixel averaging scheme (after be Bayer) be the best mathematical way to do it?

No. As a generalisation, such will use much more complex algorithms, and not just simple averaging, same with true deBayering. Without going into detail, such will take full account of relative positions of pixels as well as their value. (And any single original pixel can mathematically contribute towards more than one final pixel - that's not the case with binning.)

Originally Posted by Cliff Totten

I imagine these pixel binning calculations are extremely complex.

Pixel binning is relatively simple to do, and at sensor level - that's the beauty of it.

But as with so many things, the more effort that gets put in, the more you can get out. In the example above, the 4x4 binning approach gives you about 648lpph. Do a full deBayer, and you could expect luminance resolution of around 75-80% of the sensor dimensions - so around 2000lpph. In other words, something like true 4K.

It's worth mentioning again that you can get the best of both worlds (simple and easy processing, with good performance) if you use 2x2 blocks, and make the sensor dimensions twice the desired resolution - 3840x2160, if you want 1080p. That should give full HD resolution via simple processing - and is exactly what cameras such as the Canon C300 do. But such a sensor is 8 megapixel and not nowadays regarded as good enough for still photography.

Now, has my computer finished rendering? Back to work! :-)

[Jack Zhang]

Originally Posted by Dave Blackhurst

Sony ALREADY supports A and E series mounts/lenses in APS-c and now FF, why on earth would you think they would EVER jump on MFT?! They have a potent new combo with the RX/AX sensor and Bionz X processor, they aren't going to go reinvent the wheel and add yet another format in a shrinking market...

Thing is, this RX/AX combo is going to always be paired with lenses with worser than normal barrel distortion issues. While software can make that somewhat invisible, it's still garbage in, sifted garbage out.

It's way more likely we'll see an E-mount (as I've repeated many times in this thread) EA50 successor. NAB could bring the 4K back to the prosumer en masse.

[Ron Evans]

Originally Posted by Jack Zhang

It's way more likely we'll see an E-mount (as I've repeated many times in this thread) EA50 successor. NAB could bring the 4K back to the prosumer en masse.

I agree. This model would have the space in the body for XQD cards and fan too. Would be perfect to push 4K into the wedding marketplace !!! The VG series could be the move up from the FDR-AX100 !!! FS100 update would complete the picture . Whole family of 4K/AVCHD camcorders fixed and interchangeable lens !!

Ron Evans

[Ken Ross]

Originally Posted by Glen Vandermolen

"Price is so low?" It's their 3rd highest priced consumer camera, behind the VG series.
And I don't see much distortion, although I don't have a 4K monitor to view the footage. The footage looks outstanding to me.
Sony probably could have put a better lens on it, but not for $2,000.

I'd agree with this. I too don't see much distortion at all. Looks great to me! If there's distortion in the lens, but the internals are correcting for this, who cares?

[Darren Levine]

"$2000 camera" and "consumer" don't make much sense to this fella. I guess the true definition of consumer is whether or not you can buy it in a best buy or pc richard or whatnot

What consumer items that are considered luxuries, not necessities, bridge the $2000 mark?

homes, cars, some high end washer/dryers, all are i would gander the majority of the time: necessities

TVs would be the obvious comparison, but most are way below the $2000 mark. and if people are spending say an average of $1000 on their TV, why would they spend twice as much on a camera, something that is used far far less than a daily use TV...

I think they should have embraced the prosumer nomenclature, or ditched even more features and make it closer to the $1000 mark. this is the first THREE ND filter consumer camera...