Sony FDR-AX100

[Dave Blackhurst]

@ Peter -
Part of the challenge of any zoom lens is that not all parts of the range will be equally sharp (sometimes the best parts of the range of zoom and iris adjustments are described as a "sweet spot"), but the optics should always cover the same image circle (sensor surface).

CIZ is supposed to do some fancy math to retain sharpness while "doubling" the zoom range, basically analyzing the optical pixels and intelligently guessing at what they would look like - generally seems to be fairly effective. Past that point (AKA digital zoom), you'll start to see increasing amounts of degradation as the fancy math can only do so much to recreate the pixels, and things start to get soft and fuzzy...

That's not to say that there aren't times when you "might" use that high zoom, but generally the degraded image will be sufficiently bad that you won't want to... better to get the proper distance to the subject...

There ARE cameras that use the approach of using a smaller center portion of a large megapixel sensor as a "zoom" (Panasonic uses this), but AFAIK, Sony is either optical or creative math (digital, post sensor).

[Steve Mullen]

Dave, do you get the sense that there is a portion of the zoom range after which digital starts? Some cameras have a marker that shows the dividing point between optical and digital. Or, is Clear Zoom always digital?

I've been watching other folks samples and it seems clear there is a point at which the image magnifies motion -- or your own jitter -- and rolling shutter becomes a problem. I would love to be able to set a zoom limit.

Looking at others work on the AX100 and BMPCC -- it seems the key to avoiding rolling shutter is never pan or zoom, keep it on a tripod, and let the only motion be within the frame. Now this in standard for cinema, but I watched some docs this weekend and panning and zooming need to be used to cover in a run-gun situation. It will require discipline to keep camera movement under control. No fast movements. Welcome to the world of CMOS.

Cars speeding by in the background are a nightmare because judder causes strobing. The key is to shoot at an oblique angle. I'm thinking of using 30p to help. But, there is nothing new about this -- video camcorders used 60i to prevent strobing.

I won't comment on aliasing yet. A friend looked at my footage and commented "Helen Keller could see it." Yet, I see much less with other's videos. So I'm going to go out and shoot some more footage at 30p. I can't believe I got a bad unit.

[Adriano Moroni]

About SD

Hi, I have to buy some SD for Sony AX100.
I don't know if I will buy SanDisk Extreme SDXS 064G-X46:
SanDisk SDSDXS-064G-X46 Extreme SDXC 64GB 80Mb/sec Classe 10, Nero/Rosso/Bianco: Amazon.it: Informatica

Or cheaper SDs

SanDisk Extreme SDXC 64 GB:
SanDisk Extreme SDXC 64 GB 45 MB/s Classe 10: Amazon.it: Informatica

Are there differences on tranfer datas only from SD to PC or are there differences also in recording with AX100?
Is it worth to spend more money for buying SanDisk Extreme SDXS 064G-X46
Thanks for your suggestions.

I have seen also a black SD Sandisk extreme:
http://www.ebay.it/itm/SANDISK-Sched...item2c805e9e28

Also this with UHS-I:
http://www.ebay.it/itm/SanDisk-64GB-...item4ad00da17a

What is better?

[Peter Siamidis]

Originally Posted by Dave Blackhurst

There ARE cameras that use the approach of using a smaller center portion of a large megapixel sensor as a "zoom" (Panasonic uses this), but AFAIK, Sony is either optical or creative math (digital, post sensor).

Ah ok, so I guess it does this even if digital zoom is turned off? I have that set to off hence why I thought it was just doing sensor crop for zoom but perhaps not, maybe it's more complex here. I'd attach some pics to show what I mean as far as softness goes but the forum keeps saying "upload failed" when I try uploading the attachment, even though the jpg falls within forum limits :(

[Ken Ross]

Originally Posted by Steve Mullen

Dave, do you get the sense that there is a portion of the zoom range after which digital starts? Some cameras have a marker that shows the dividing point between optical and digital. Or, is Clear Zoom always digital?

Someone has done an assessment on the clear zoom implementation on the AX100. It varies from 5% magnification vs Standard OIS at full wide to about 55% at full zoom. Even so, the degradation of the image at full zoom is not great since the same resolution at full wide is not necessary at full zoom. So the impact on resolution was very minimal. I can see some degradation at full zoom vs Standard OIS, but it's really not that significant.

Originally Posted by Steve Mullen

I've been watching other folks samples and it seems clear there is a point at which the image magnifies motion -- or your own jitter -- and rolling shutter becomes a problem. I would love to be able to set a zoom limit.

Looking at others work on the AX100 and BMPCC -- it seems the key to avoiding rolling shutter is never pan or zoom, keep it on a tripod, and let the only motion be within the frame. Now this in standard for cinema, but I watched some docs this weekend and panning and zooming need to be used to cover in a run-gun situation. It will require discipline to keep camera movement under control. No fast movements. Welcome to the world of CMOS.

I don't understand this at all. There is no evidence of RS in my videos or virtually any videos I've seen posted by people who know what they're doing. If you're inclined to shoot rapidly moving railroad trains, perpendicular to the action, from 6' away while zooming, yes, you'll see RS. For most reasonable work you simply will not see RS and virtually all owners confirm this. The tests that some of these guys perform by violently shaking the camera right, left, up and down, is truly hilarious. Is that how they shoot?

Originally Posted by Steve Mullen

I won't comment on aliasing yet. A friend looked at my footage and commented "Helen Keller could see it." Yet, I see much less with other's videos. So I'm going to go out and shoot some more footage at 30p. I can't believe I got a bad unit.

Aliasing? Where??? There isn't a trace of it in any of the videos I've shot or seen by other owners. Obviously if someone has edited a video and has incorrect settings for a given NLE, artifacts can creep in easily. But even edited videos from those that know what they're doing, show no evidence of aliasing.

[Ken Ross]

Originally Posted by Peter Siamidis

Ah ok, so I guess it does this even if digital zoom is turned off? I have that set to off hence why I thought it was just doing sensor crop for zoom but perhaps not, maybe it's more complex here. I'd attach some pics to show what I mean as far as softness goes but the forum keeps saying "upload failed" when I try uploading the attachment, even though the jpg falls within forum limits :(

Pete, never select 'digital zoom' in the AX100 or any camera for that matter. However using Clear Zoom is a far different animal. Whatever algorithm it is that Sony uses, is very good. You'll probably only catch some degradation at the extreme end of the zoom range. Even then, if you don't perform an A/B with and without it, you might not even be aware of it.

[Dave Blackhurst]

Based on other Sony cameras, there are a couple things going on internally - there's "optical stabilization", which is mechanical. The "clicks" that have been reported are likely internal gyros and servos that provide this stabilization powering up and down when the camera is turned on and off - they are not as advanced as the PJ/CX7xx series BOSS (magic eyeball!), since that sort of system would be quite large comparatively (sensor and lens are significantly larger in the AX100. In theory, there shouldn't be any degradation at all from that, as it's just a system to keep the image relatively stable against some movement - I'm guessing it's about as effective as earlier OIS systems, and a lot less impressive than the BOSS system. This is simply because it would be somewhat restricted to vertical/horizontal moves and perhaps roll (some earlier OS systems didn't adjust for roll, and later ones did). The magic eyeball gimbal moved the whole the lens block, so compensated for more motion, more effectively.

THEN, you have digital image stabilization, which compensates for motion POST sensor, typically by using the "edge" pixels to basically move the image around to compensate for image movements larger than the OIS can deal with. This comes under the "fancy math" title - where the internal SOFTWARE adjusts to try to correct the framing against any movement. Since the software will be using "part" of the image from the sensor to move the image within the framing, you will lose a little resolution (hope that made sense, it operates similarly to image stabilizers used in post/edit - you lose some pixels at the edge to achieve a more stable image digitally "derived and enhanced" from the central portion of the image.



RE Zoom:

The OPTICAL portion of the zoom should be strictly MECHANICAL, meaning lens elements are moving forwards and backwards within the lens to adjust the OPTICAL zoom range - BUT, lenses tend to have portions of the zoom range that perform better than others, and the iris (f/stop) also affects sharpness.

That's a "quick and dirty" description of how lenses "work", and it's why "primes" are popular - a lens designed for a SINGLE focal length can be optimized for that focal length for optimal sharpness, whereas a zoom lens has those internal elements moving around, and has to be "sort of optimized" for a wide range of focal lengths, and thus may be compromised to greater or lesser degrees, typically at the "ends" of the zoom range. It's pretty common to have complaints about a zoom lens at certain apertures and or parts of the zoom range... physics can be fickle.

The way all the "other" Sonys work is that there's a little white bar part way through the zoom range (when CIZ or digital zoom is active) that designates where the OPTICAL zoom ends - then there should be a "C" and the multiplier above the bar when in CIZ range, and if Digital zoom is enabled, it will switch to a "D", again with the multiplier displayed above the zoom bar. I'm guessing the AX100 sticks with this design?


I am seeing some judder or stutter in some samples with faster motion (people running through a frame), and I'm not sure if it's due to shooting in 24p, and if 30p will be THAT much better - it is giving me a bit of "pause"... hoping that it can be worked around, as can RS. Somehow I suspect "camera technique" is going to be more critical than it has been in some recent cameras, sort of have to "step back" a bit to get best results. I guess I'll be keeping a PJ760 around for a while for when I don't want to be as focused on technique!


@Adriano -
I'd suggest double checking the manual for minimum memory specs (IIRC it only calls for "SDXC class 10"? Once you meet the minimum, you "should" be OK, the camera just has to be able to write data fast enough, faster cards "may" mean you can download/read a bit faster. I'm reviewing memory I've already got, hoping to use what I've got initially, but not sure myself what all the differences are... is a class 10 UHS1 SDHC (got some of those!) card the same as a similar spec SDXC card? SO many nuances to memory cards!

[Dave Blackhurst]

Just a quick clarification - OPTICAL is the range from "wide" to "tele" created by the moving glass (or plastic?) internal elements. CIZ "takes over" once you are at the TELE end of the optics, creating the full frame image from the output of the sensor (or a progressive portion thereof, you could say), basically a digital doubler with minimal image degradation.

"Digital" is further digital zooming and interpolation of the data read off smaller and smaller portions of the sensor, and at some point you simply can't create what isn't there, but "may" get some usable footage for SOME purposes (if you think about it, you can zoom in a long way into a 4K image and still have crappy SD level output!). I've actually shot some stuff (stills, no way I can hold the cam steady enough for VIDEO!) at the extreme end of the digital range of the RX10 that was "not bad"!

I mentioned the Panasonic method of sampling the center portion of a larger megapixel still sensor - they do this on interchangeable lens cameras, effectively "doubling" the optical lens range of whatever is mounted while maintaining full pixel resolution. It's a bit different approach from what I describe above... but wanted to clarify that there are different "tricks" to achieving "zoom"...

[Peter Siamidis]

Originally Posted by Ken Ross

Pete, never select 'digital zoom' in the AX100 or any camera for that matter. However using Clear Zoom is a far different animal. Whatever algorithm it is that Sony uses, is very good. You'll probably only catch some degradation at the extreme end of the zoom range. Even then, if you don't perform an A/B with and without it, you might not even be aware of it.

I did actually notice it at the long end even without doing an a/b, this was during a quick test I did where I framed a plant to fill the screen roughly the same way each time but at different focal lengths. I attached some pics of it as seems like attachments are working now. At 6 feet the lens is at max width, at 15 feet is roughly halfway zoom and at 30 feet is full zoom. I never use zoom normally and I always leave digital zoom disabled, but I was thinking of trying the AX100 as a replacement briefly for my now sold VG900 (while I wait for a replacement full frame camera) so I was messing with the AX100 to see how much shallow dof I can extract from it and how clear it looks. That's where I noticed how on the long end the image no longer seems like 4k, but I wasn't sure if that was due to just the optics of the situation or because of how the camera handles zoom.

[Dave Blackhurst]

@Peter -

I know it's minor, but there's about 10% difference in framing between the first shot and the last - that could account for the apparently lesser detail? It's not "much", but I suspect it's "just enough". If it's a concern, reshoot and adjust the framing so the basket is as close to identical as possible in size in the frame and see if that does the trick or not...

It is possible that the lens softens a bit at the end of the optical range (as mentioned above), but honestly, all the shots look "pretty good" (I need a 4k system... now!), considering the framing differences.

[Ken Ross]

Originally Posted by Dave Blackhurst

Just a quick clarification - OPTICAL is the range from "wide" to "tele" created by the moving glass (or plastic?) internal elements. CIZ "takes over" once you are at the TELE end of the optics, creating the full frame image from the output of the sensor (or a progressive portion thereof, you could say), basically a digital doubler with minimal image degradation.

Not sure that's the way it works in the AX100, Dave. This is the first Sony I've had where CIZ doesn't work like others. As soon as you engage Active Stabilization, CIZ is automatically engaged. Doing so, even at full wide, shows a very slightly cropped area. As someone had measured with a resolution chart, there is a 5% cropping at full wide. So I don't think this is taking place in a purely optical environment, but who knows?

[Ken Ross]

Originally Posted by Peter Siamidis

I did actually notice it at the long end even without doing an a/b, this was during a quick test I did where I framed a plant to fill the screen roughly the same way each time but at different focal lengths. I attached some pics of it as seems like attachments are working now. At 6 feet the lens is at max width, at 15 feet is roughly halfway zoom and at 30 feet is full zoom. I never use zoom normally and I always leave digital zoom disabled, but I was thinking of trying the AX100 as a replacement briefly for my now sold VG900 (while I wait for a replacement full frame camera) so I was messing with the AX100 to see how much shallow dof I can extract from it and how clear it looks. That's where I noticed how on the long end the image no longer seems like 4k, but I wasn't sure if that was due to just the optics of the situation or because of how the camera handles zoom.

Pete, I think your tests nicely show what I had mentioned. The image degradation at full zoom with Active IS, although there, is not that great.

Most people will never think the fully zoomed, 4K Active IS, is not a 'full' 4K.

[Ken Ross]

For those curious about the low light capability of the AX100 as well as the implementation of Active Stabilization, part of this video might help.

We were invited to a wedding in the AC area. So a quick edit shows some shots from the AC boardwalk area as well as some interior shots of Caesars. The low light shots were taken in the reception hall, which was very dark. For some shots I had to use the camera fully zoomed with Active IS...not recommended in such a low light area, particularly for a lens that's not constant aperture.

Since we were guests, obviously no tripod was used.

You can download the original video if you're a Vimeo member. Obviously this will give you better quality than the online Vimeo version.

AC & Wedding on Vimeo

[Andy Wilkinson]

Thanks for posting Ken. Insanely sharp!

[Ken Ross]

Thanks Andy!