Sony FDR-AX100

[Cliff Totten]

F9.3 on tbe long end? Huh?

No no...im sure thats just a typ'o on the pre production lens.

Im sure sombody in the factory was dyslexic and accidentally typed "9.3" instead of the real "3.9".

Im sure its just a simple mistake.

[Glen Vandermolen]

F9.3 could be real. The sensor is very densely packed with pixels to achieve 4K. As a general rule, the more pixels you cram into a sensor's given area, the less sensitive the sensor becomes.

I like the pics of the prototype with the XXXXXX info on the lens. As someone else stated, maybe it's an ideal camera for porn!

edit - I don't think the 9.3 on the lens means f-stop. It might mean the zoom range, maybe in a 35mm equivelant. The actual info on the lens is: 2.3/9.3-111.6.
As a comparison, the info on the RX10's lens is:2.8/8.8-73.3 and we know the RX10 has a constant aperture lens throughout its zoom range.
9.3-111.8 is roughly 12x, as 8.8-73.3 is roughly 8.3x.

[David Heath]

Originally Posted by Glen Vandermolen

F9.3 could be real. The sensor is very densely packed with pixels to achieve 4K. As a general rule, the more pixels you cram into a sensor's given area, the less sensitive the sensor becomes.

The f-stop is solely concerned with lens optics - the ratio of focal length to iris diameter. If a sensor is less sensitive, it will need to be illuminated more - hence may need a faster aperture for a given illumination - but that's a different matter to lens ramping.

Under "Optics/Lens" in this sheet it gives the answer - 4K Camcorder with 1" sensor - FDRAX100/B Review - Sony US - f4.5. So it ramps a little over a stop over the whole zoom range. Pretty much what would normally be expected.

edit - I don't think the 9.3 on the lens means f-stop. It might mean the zoom range, maybe in a 35mm equivelant. The actual info on the lens is: 2.3/9.3-111.6.

The 9.3 and 111.6 do indeed mean the focal lengths at the widest and tightest ends of the zoom, but they are actual focal lengths in mm - not 35mm ff equivalents. (Remember we're talking about quite a big chip size.)

Since the width of the frame is quoted as 13.2mm, and ff 35mm is 36mm, those figures can be used to scale up to 35mm equiv figures:

Wide: 25.4mm (9.3x36/13.2)
Tele: 304mm (111.6x36/13.2) - a 12x lens, as Glen says.

[Dave Blackhurst]

I believe David is correct - 4.5 on the long end is the widest aperture, elsewhere in the specs there was an indication you could further stop down to f/11 if needed. I tried the range from f/2.8 to f/4.5 with the RX100, it's not a big "loss" of light - I know ANY is sort of traumatic once you've had constant f/2.8 <wink>, but I'm guessing that in practical use it won't be a major issue.

[Cliff Totten]

"Sensor readout".... We haven't talked much about this yet.

We are told that the RX10 reads all it's pixels in video mode and does not line skip. (for 1920x1080 scaling) We suspect that the AX100 uses the same sensor and could possibly use the same Bionz X processor. I'm assuming the RX10 does a "pixel binning" process for scale down to HD.

What about the AX100? Is 4K too much data for that kind of readout? How does it get 14Mp (in 16x9) to scale down to 4k? (8mp)

I would imagine at that resolution, if it did line skip, it wouldn't be a big problem?

CT

[David Heath]

Originally Posted by Cliff Totten

"Sensor readout".... We haven't talked much about this yet.

We are told that the RX10 reads all it's pixels in video mode and does not line skip. (for 1920x1080 scaling) .......... I'm assuming the RX10 does a "pixel binning" process for scale down to HD.

What about the AX100? Is 4K too much data for that kind of readout? How does it get 14Mp (in 16x9) to scale down to 4k? (8mp)

You need to see them on a chart to say for certain. In a lot of ways pixel "skipping" or "binning" will give similar results, so no, if the reports about the RX10 are correct, I don't think the RX10 is doing anything like that.

It is hoped that it's not only reading out the whole sensor at frame rate, but that it's deBayering it fully, (as it would for a 14 Mp still) and then doing a full downconversion to HD from that.

As far as the AX100 goes, then once you've got the deBayer of the whole sensor, it shouldn't be much more difficult to downconvert to 4K than HD.

Since still digital cameras started to do decent video, that sort of processing has been too much for the technology without severe power /cost etc issues - hence the pixel skipping and binning that's had to be used with such sensors to get viable HD. (And none of them has managed to deliver much better than 700+ lines of resolution.) But technology moves on, and processing gets more powerful, so it's quite possible we are now seeing a new highly desirable era, and something as described above becoming a reality.

Hopefully. I'm optimistic, but I for one would like to see chart results before being sure - and they need to have input frequencies beyond 4K resolution to really tell what's going on.

[Dave Blackhurst]

Again David is correct from what I've been able to find - the Bionz X is a new generation of processor that can supposedly handle the date rate off the entire sensor (which I've seen described as "5K") at frame rates sufficient for video.

This approach, rather than "tossing" (If I understand the term "binning" correctly) large chunks of information, would have ALL the data available to munch on and use to produce final output, thus should yield much better resolution and overall results. So far that seems to be playing out in the RX10, and I suspect we will see the same in the AX100.

There are always "bottlenecks" in the engineering of such things - one component or another that forces compromises (like I suspect the 4K/30p being a concession to being able to use readily available memory). To the end user, it just looks like they left something out, or crippled functionality, or maybe doesn't work like they thought it would... when the sad truth of the reality is that there are limits to technology!

Ten years on, these cameras that are bleeding edge today will no doubt look rather limited and archaic, as technology marches forward... As fond as we all were of the HC1 "back in the day", and it tickled us to see it re-appear again in the AX100 design DNA, I'd venture that it will look rather "dated" when output and performance is compared!

[Wacharapong Chiowanich]

I've never forgotten how impressive it was to have first seen my new HC1's HDV clips being played back on my 1024x768 computer screen. Back then it was still a big leap visually compared to the best my "state of the DV art" workhorse DSR-450 could do. In about 2-3 months' time I believe the AX100 would likely do the same to my FS100 and EX1R's video like its old cousin did to the DSR's.

[Emmanuel Plakiotis]

David Heath made a small mistake in his calculations, because he didn't take into account the ratio difference. Hence the FF equivalent of 9.3mm 1" sensor is 29 and not 25. It would have been 25, if both formats were identical. But FF is de facto 2/3 and AX100 1" sensor, is 16/9.

[Jack Zhang]

There's no real proof that the AX100 sensor is 16:9. For all we know it is the RX100 II sensor but with a better video processor. That sensor is 3:2.

[Emmanuel Plakiotis]

When I was saying sensor, I meant the video area not the actual sensor. The sensor most probably is 2/3, because the still pixel count is 20MP vs 14MP for the video. The video frame is definitely 16/9 and that's the important factor when calculating the FF equivalent.

[Emmanuel Plakiotis]

That's my #16 post in this thread:
http://www.dvinfo.net/forum/digital-...ml#post1826824

"Digging the pics of the Sony site I found the following:
lens 9,3-111,6 f2,8-4,5 equivalent to 25-300 in FF, according to my calculations for 3/2 aspect ratio. For 16/9 probably around 29-348. Good wide angle, but only for stills.
18X clear image zoom. Probably zooms in to native 4K resolution. Possibly more zoom in for HD.

Usually I do my calculations by myself, but always verify them with a nifty iphone app which is called "Pcam"

[David Heath]

Originally Posted by Emmanuel Plakiotis

David Heath made a small mistake in his calculations, because he didn't take into account the ratio difference. Hence the FF equivalent of 9.3mm 1" sensor is 29 and not 25. It would have been 25, if both formats were identical. But FF is de facto 2/3 and AX100 1" sensor, is 16/9.

The RX10 & AX100 sensor is 13.2x8.8mm - which is 3:2 - that is given in the same link I posted. The derivation of 16:9 must therefore involve some vertical cropping, so I'd expect the effective dimensions for (16:9) video to be 13.2x7.425mm.

"Full frame 35mm" is normally taken to be 36x24mm - also a 3:2 ratio. If such a camera gets used for video, then it's effective area will be 36x20.25mm.

For my calculations I was using the full sensor (3:2) area in each case - 36x24 mm and 13.2x8.8 mm - which yields an equivalent 25mm figure for focal length. Switch both to a 16:9 mode and the same equivalence holds true - 9.3mm on the RX10 sensor in 16:9 mode will still give exactly the same angle of view as 25mm on a full frame sensor in 16:9 mode.

If you define "equivalent focal length" as ALWAYS relative to the diagonal of 3:2 36x24mm (regardless of aspect ratio) then what Emmanuel says is quite true. But it seems more logical to relate on a like for like basis, especially when the RX10 has a 3:2 sensor, which may be used as such or switched to 16:9 mode. Otherwise, use a real FF camera in video mode and it likewise must have an "equivalent focal length" of a little more than it's real focal length.

At the end of the day it's a matter of convention, and maybe (as with calling this a 1" size in the first place) common sense has little to do with established convention.........

[Emmanuel Plakiotis]

When we speak about equivalent FF focal length is an empirical way to understand the angle of view of a uncommon format for a given focal length. Therefore is more appropriate ALWAYS to measure them against the diagonal of the 3:2 FF (36X24), which is something that most people involved with photography or cinematography understand by experience.

"...Otherwise, use a real FF camera in video mode and it likewise must have an "equivalent focal length" of a little more than it's real focal length."
As David mentions above:
25 focal length of FF cropped for 16:9 yields an angle of view that is smaller and more closely reassembles to 29 FF 3:2 (uncropped).

As I had stated in my initial post, which I have previously reposted again, the AX100 has an equivalent focal length 25-300 for 3:2 aspect ratio (still) and 29-346 for 16:9 aspect ratio (video).
Since AX100 is predominantly a video camera, is more helpful to state the 29 as the equivalent focal length and not the 25.
Actually most reviewers of the camera and I think Sony itself, do exactly that.

[Bruce Dempsey]

Ever since just before hdv I've yearned for a camera which would shoot stills and video at the same time and here it is finally at last...Just pick any frame.