Sony FDR-AX100

[Wacharapong Chiowanich]

Originally Posted by Dave Blackhurst

Not sure why CIZ and AS are linked together,.....................

Dave,

I own quite a few Sony handycams with the Active Stabilization but none of the new models that have the advertised Clear image Zoom and here is what I've observed:

- As soon as the Active OIS mode is engaged, the cropping and zooming in of the image to fill the 1920x1080 pixel frame is activated right from the wide end. For instance, on my HDR-PJ760VE (a model with the BOSS tech) and HDR-CX700E (non-BOSS), the 26mm (35mm AoV equivalent) at the widest end becomes close to 29mm or narrower. Once I zoom in to the tele end of the range, the cropping and zooming degree increases progressively until I reach the end of the zoom. The AoV of the image at the tele end of the zoom corresponds to that of the focal length of roughly about 442mm, or 17x (17 x 26mm) as claimed by Sony, and not 10x (260mm) as indicated by the mechanical zoom ratio alone.

- In the old days Sony didn't call this Clear Image Zoom but mentioned this behavior in the manual which was true.

- The numbers aside, what I've seen from the max zooming and engaging the Active OIS is that the sharpness DISTINCTLY deteriorates on my calibrated monitors. There was no doubt by anyone who saw the image. It was plainly softer at the tele end apart from the very noticeable CAs in contrasty scenes but that's another issue.

- Looking at the specs, the above two models share more or less the same 1/2.88" sensor, the design of which dates back to Sony's first back-illuminated CMOS sensor used in the XR500/520 hard disk series. Sony specs claim the sensor's pixel count is 6M+ gross and 4M+ effective for video. This may seem more than enough to be de-bayered and reserved for the purpose of digital stabilization etc. as you suggest but the truth I and everyone else saw on the monitors suggested things may not have worked as we thought they should.

- There is no question the Active OIS in the Sony's small sensor Handycam models is very effective in terms of stabilizing the image and that the system is superior to most of the IS systems of the competition out there. However I believe this (Active and NOT Standard or optical-only IS) is achieved at the cost of effective video resolution. Though I don't have a solid proof for it I think the baseline pixel count of the image pre-digital stabilization processing in the 1080p modes starts somewhere around the theoretical 1920x1080 or about 2M pixels and then decreases, progressively, from there once you engage the Active OIS.

- Looking at it another way, all these marketing names of digital zooming may be mostly just the by-product of the working of the Active OIS system while in the Standard OIS mode it is just a similar cropping and zooming process but maybe with some different interpolation algorithm.

This could be a reason why some members on this forum such as Peter Siamidis etc. have noticed the softening of the image at full zoom also on the AX100.

[Steve Mullen]

Originally Posted by Dave Blackhurst

@Steve -

I'm not seeing objectionable "artifacts" for the most part, maybe tiny, minor ones here or there in a "worst case" scenario, but certainly not something to condemn a $2K camera. I don't think we doubt that YOU are seeing things, but as a practical matter, others ARE NOT, at least not to the degree that you seem to be. The video looks eye popping-ly GOOD overall, even YOURS is not bad!

I don't remember saying the amount was "objectionable" or "significant." I only needed to establish the existence of aliasing in the face of the claim there was "no aliasing." I do not want report something that didn't exist. Now, I now there are multiples types of aliasing. They are visible on a computer's screen, on a 63" HDTV connected via HDMI, and totally independently via an Apple TV streaming from YouTube. And, of course, they are also visible when the original file right from the camera is viewed.

I'm pretty sure one could keep them to a minimum by simply shooting with care.

I have just run additional tests to see if 24p verses 30p makes a difference. It doesn't. Also tests to see if manual verses auto exposure makes a difference. It doesn't. These tests look at line-twitter.

When a camera sensor's frame moves vertically, very thin horizontal lines or details in the scene will be captured by a row—then fall between this row and the next, and then be captured again. Fine detail will appear and disappear as long as there is any motion. The fine lines/details will be seen to "twitter" if there is the slightest vertical camera shake -- or a vertical pan or a zoom. A progressive camera can reduce twitter by passing the video signal through a low-pass filter. While this cleans the signal, it also slightly reduces vertical resolution. Some cameras have a Sharpness control that enables you to adjust the amount of filtering employed.

We have seen these "line twitter" artifacts before.

REVIEW OF SONY V1E:
"Playing back 1080p (Progressive) footage on the Sony looked a little different. The colour saturation and contrast were about the same and it handles highlights superbly, only there is some 'line-twitter' evident in the images, especially on horizontal edges. I filmed some books stacked horizontally and there was slight line-twitter along the book edges in places. However, I was able to fix this by changing the Sharpness setting in the menu. By default the sharpness is set to +7, by dropping it to a more realistic +3 the line-twitter vanishes.

Nobody in their right mind would really want the sharpness set so high; +7 is a bit overkill. I can only assume that Sony set it so high from the factory because they assume most purchasers of the camera will be hobbyists who think this look is really great. Personally, whenever I look at an image that has sharpness and detail levels cranked up, I don't see a sharp stunning picture, I see a picture full of junk and garbage that should not be there so I always lower these settings; usually quite drastically. Now if you lower the sharpness settings you could be forgiven for thinking that the image looks soft, it doesn't, it simply looks natural. Try it for a few weeks, then crank the sharpness up again and you'll see just how horrible too much sharpness actually is." ©2006 Nigel Cooper

Line-twitter 24p test - YouTube

Line-twitter 30p test - YouTube

[Steve Mullen]

"The numbers aside, what I've seen from the max zooming and engaging the Active OIS is that the sharpness DISTINCTLY deteriorrates on my calibrated monitors. There was no doubt by anyone who saw the image. It was plainly softer at the tele apart from the very noticeable CAs in contrasty scenes but that's another issue."

To minimize rolling shutter I've found it best keep from zooming-in too far. The question is that when it is engaged at wide -- does resolution degreed. And, how far one can zoom before resolution degrades?

It seems Sony first downscales the sensor image to UHD. When Active is engages a smaller window within this image is used -- with the buffer areas being used for EIS. Then, since UHD is recorded, the smaller image is up-scaled back to UHD. This two stage process can't help image quality.

But, why is this system part of a digital zoom system? Why isn't the optical zoom good enough? Or, did Sony simply decide that the logic that downscales and then upscales could be used to provide a 50% extra zoom range. I can't see any need for a 15X zoom, but maybe marketing did.

Anyway, how do the optical and digital zoom work together? Is the optical zoom used until it reaches its maximum and then the digital zoom begins and adds to the range?

I'm going back to Optical to limit the zoom range and see if jitter increases too much. In fact, what led folks to use Active?

Here's the Blood Moon using the 160X digital zoom. So bright I used ND.

AX100 Blood Moon - YouTube

[Dave Blackhurst]

I've seen PLENTY of artifacts on broadcast TV, and some in movies, all most likely shot with cameras FAR more expensive than the AX100...

Thus my qualifier of "objectionable" - of course some of this comes under the "once seen you can't un-see" category. But if you're going around looking for flaws, you'll find some, pretty much guaranteed. I'll admit that the stutter that sometimes shows up in some videos gives me a bit of pause, but most videos don't seem to have it on a level that's worth quibbling over, especially at this price point.



And about the active stabilization and digital zoom... shot the bloodmoon last night with the trusty old PJ760, and the active stabilization and digital zoom were interlinked - setting active meant that digital zoom was also engaged... so I guess this is a "feature"... and it was a real pain having to have stabilization on when on tripod and needing SOME digital zoom on top of a 2X tele lens!

[Steve Mullen]

"I've seen PLENTY of artifacts on broadcast TV, and some in movies, all most likely shot with cameras FAR more expensive than the AX100..."

I'll admit that the stutter that sometimes shows up in some videos gives me a bit of pause, but most videos don't seem to have it on a level that's worth quibbling over, especially at this price point."

I mostly see aliasing on SD to HD and interlace converted to progressive -- fox, espn, etc. I don't see any issues with film or CineAlta shot programs.

I really wonder what these very rare pauses are. I have assumed it is not the AX100 codec and just the load of data with UHD bogging the computer down.

I know morie can appear on the BMPCC, but I see it very rarely and I've watched a lot of samples. The brick building is the only time I've seen it from the AX100 -- although a bit occurs on the bird cage at the very end. So it seems not a real problem on either camera. Likewise, both can show RS, but if folks are careful with movement it shouldn't be a problem with either. Neither show aliasing on diagonals -- so also not a problem with either.

Bottom-line, noting a camera's artifacts -- doesn't make the artifacts automatically a "problem" for an owner. It's called getting to know your camera.

The AX100, however, has two artifacts that are serious problems for me: motion judder and line-twitter. Motion judder at 24/30 afflicts almost all "video" camcorders. I never see it on "cinema" cameras. Line-twitter never slows-up on "cinema" cameras either. The reason for the difference is that makers design for two different demands: a very clean image or a very sharp image. If one is in the "once seen you can't un-see" category -- then clean is an absolute must have.

There in lies a question. The GH4 with lens is a bit more than the AX100 and -- if you like DSLRs -- is clearly a pro level camera. And, then there is the question of 4K. It is very slow to work with -- and a problem for streamed videos (which is the way most folks share media in 2014) and there are no optical discs -- so I've come to question my need for 4K at this time. That opens a huge choice of cameras. The Sony 7s, the Sony RX100 -- that has the same sensor as the AX100 and who's video looks amazing like that from the AX100, and the BMPCC. Both of the later are cheaper. Waiting may pay rewards.

[Wacharapong Chiowanich]

Originally Posted by Steve Mullen

It seems Sony first downscales the sensor image to UHD. When Active is engages a smaller window within this image is used -- with the buffer areas being used for EIS. Then, since UHD is recorded, the smaller image is up-scaled back to UHD. This two stage process can't help image quality.

I believe this is what actually happens in my Active OIS-capable Sonys when the Active Mode is engaged. Just replace the word UHD with 1080p, of course.

On newer models with the separate Clear Image Zoom function such as the AX100, I have no idea if there is a distinct threshold at the tele end that indicates the extra digital zooming begins or if, in case of the Standard OIS mode, it means only digital zooming effect is applied after the tele end of the optical zoom range without any digital stabilization.

You ask why some use the Active mode and the answer is simple, they need to lessen the shake as much as they could either knowing or not knowing the mode degrades the resolution.

[Ken Ross]

Originally Posted by Wacharapong Chiowanich

Dave,

This could be a reason why some members on this forum such as Peter Siamidis etc. have noticed the softening of the image at full zoom also on the AX100.

That's true, but the big difference between the AX100 and the models you mentioned, is the AX100 has so many more pixels to deal play with. As a result, the softening will be far less than regular HD camcorders.

[Ken Ross]

Originally Posted by Steve Mullen

I don't remember saying the amount was "objectionable" or "significant." I only needed to establish the existence of aliasing in the face of the claim there was "no aliasing." I do not want report something that didn't exist. Now, I now there are multiples types of aliasing. They are visible on a computer's screen, on a 63" HDTV connected via HDMI, and totally independently via an Apple TV streaming from YouTube. And, of course, they are also visible when the original file right from the camera is viewed.

I'm sorry Gentlemen, this is just getting more than a bit silly, it really is. This frenzied attempt to find artifacts (and in one case that's precisely what it is), ANY artifacts in a $2,000 Handicam camera is utterly absurd. If this kind of scrutiny were applied to megabuck cameras, you could find something.

Then to go on say that you could keep artifacts to a minimum by 'shooting with care' is again just a silly comment and more hyperbole to make readers think the artifacts are worse than they are. The video that I shot and posted was not shot with 'care' to avoid artifacts. If the camera had these artifacts that I was trying to avoid, why would I have shot a large expanse of a brick wall?? Most people that have viewed this video do not see artifacts. In fact, had I shot that same scene with far more expensive cameras, it probably would have been crawling with artifacts. No scrutiny required.

For the life of me, I don't know why one guy is on, for lack of a better description, an 'AX100 witch hunt'.

Again, hold up any camera to this kind of scrutiny and you WILL find something. In fact, with most cameras costing multiple times that of the AX100, you'll find FAR more artifacts than what exists in the AX100.

I'm sorry, it certainly appears that Steve is on some kind of a mission motivated by who knows what? To prove that under intense scrutiny he can find something? Big whoop!

Excuse me while I get back to shooting some amazing video with this little thing.

[Ken Ross]

Originally Posted by Steve Mullen

And, then there is the question of 4K. It is very slow to work with -- and a problem for streamed videos (which is the way most folks share media in 2014) and there are no optical discs -- so I've come to question my need for 4K at this time.

And therein lies the problem. If one is predisposed in feeling you have no need for 4K at this time, it puts some serious question into the objectivity of the person performing 'reviews' of 4K equipment. It's human nature, it just is. But this goes a long way in explaining some of the things I've read.

Secondly, some of us have actually done some editing with 4K footage and know it's not as onerous as some make it out to be. I'm editing on a Toshiba all-in-one computer, hardly a powerhouse, and a 2 minute video can take about 3 minutes 10 seconds to encode. Hardly any different than AVCHD. So let's be fair about that one too.

Originally Posted by Steve Mullen

That opens a huge choice of cameras. ......the Sony RX100 -- that has the same sensor as the AX100 and who's video looks amazing like that from the AX100, and the BMPCC. Both of the later are cheaper. Waiting may pay rewards.

So now an HD camera looks 'amazingly like' that of a 4K camera?!?!?! I've seen lots of footage from these cameras and in terms of clarity of detail and resolving power, there is no comparison. None. Zero. I'm sorry, I'm nearly speechless. There is no objectivity here.

[Mark Rosenzweig]

Steve Mullen said: "the Sony RX100 -- that has the same sensor as the AX100 and who's video looks amazing like that from the AX100."

This statement reveals either bias or ignorance. The RX100 is well-known to be one of the most artifact-laden cameras of all cameras. See Slashcam.de for the details. The AX100 uses a completely different method of taking video from the same sensor, which precisely avoids the problems of the RX100. The RX100 videos are visibly moire-contaminated. It can be seen in test charts (unlike that from the AX100) and in almost all videos.

After seeing post after post about the made-up artifacts of the AX100, it is now abundantly clear that Mullen is not a source of information or views that should be taken seriously.

[Marc Salvatore]

I have a question for AX100 owners. How shallow of a DOF can you get? I realize this is not a big DSLR type chip but can you get more shallow DOF than an EX1?

Thanks, Marc

[Dave Blackhurst]

@Ken -
EXACTLY, you can probably find "artifacts" (or color shifts, or...) if you have a trained eye, no matter WHAT camera is used! A $2K camera that produces video that is pretty much like looking out a window (one that isn't dirty and doesn't have optical distortions in the glass!) is a pretty slick thing - two years from now, it'll probably be "less" slick, surpassed by other cameras, that's tech for ya! For now, it's the only game in town if one wants a 4K "handycam" without breaking the bank...

Sure, if you've got m4/3 investments, a GH4 makes pretty good sense, and it's in the same price range (but w/o lenses, and those things COST $$).


@Steve

Let's sort out a couple things, there is an RX100... decent camera, got plenty of decent results (video was trickier in the first release) out of mine, and an RX100M2 that we believe shares the sensor (but does NOT share the processor) of the AX100. Honestly, if someone doesn't need a lot of zoom, and wants a "pocket" camera, these ARE decent choices, I have one for what it does - shoot "pretty good" images in a pocket cam... better than a P&S, and still discreet and easy to carry.

Then there is the RX10 - we believe it's the same sensor and processor, and frankly is a pretty amazing camera (if it had enabled 4K, I wouldn't even be thinking about the AX100, to be honest). If I had to recommend an "all around" camera for someone who wants great results and doesn't mind a slightly bigger package, again, it's hard to beat. I am happily using mine while waiting for an AX100 to pop up at the right price (after someone reads one too many internet criticisms and sends theirs back <wink>). I got my RX10 at around 30% off list, open box... and not in a "hurry" to replace it whatsoever. It is a beast among cameras!

EACH of the above cameras serve a purpose, has plusses and minuses, and will no doubt be "dated" in a few years. BOTH have been good cameras for my uses. Note the word "uses", not technical nitpicking - images from both are good enough, especially on a $/quality basis - could I get "better" results from a bigger, heavier, more expensive camera?? Maybe... but would I use this (theoretical) bigger, heavier camera or spend the additional money for "incremental" increases in performance? Nope... and I have bigger heavier more expensive cameras (and some less expensive ones too) that are sitting around more than being used since I got the RX100 - now the RX10 is what I grab... and that likely won't change should an AX100 be added!


Where the AX100 "fits" (and I see it as a complement to the RX's) is where one needs a dedicated video camera, and where the additional resolution will provide options in post, or "better" quality than that available from a 1080/60p camera. I want to test the higher bitrate 1080/60p, as I suspect I will use that mode at least a bit, though the 4K will probably be the first choice. The high framerate video is also of interest for certain things.

I'm very aware of the challenges of dealing with 4K, it's a "system" upgrade for me, meaning not just a camera... so budget and workflow are both big "issues"! But the results others are getting are more than enough to be convincing.

In another part of the DVi forum universe, a video was posted by someone giving an address "against" 4K... I just turned it off after he made a statement that "you can't see the difference"... when such a ridiculous blanket statement is made, all credibility is GONE, sorry. If you can't see the difference, get thee to the optometrist!!

Some are comparing 4K to "3D"... 3D gives me and many others headaches... 4K makes me feel more "there" than 3D could ever do, and no headaches. To me and my tired old eyes, the reasons to move in this direction are obvious, and I fully expect it will be far quicker to adoption than HD was.

That said, going to a 4K camera may NOT be for everyone, at least with this first to market device (or any of the other "first to market" devices, soon to be followed by a FLOOD of 4K cameras). Steve, you might be perfectly happy with an RX100M2 (if pocketability is important) or the RX10 (if you want a better lens and more features that can be packed into a pocket device). I see the differences the AX100 provides, but I CAN live with what the RX10 can output, it's not like it horrifies me with lack of quality or anything!

[Steve Mullen]

Fanboys will always defend what they just bought to the bitter end. They are so motivated they will simply not see anything that might cause them to re-think their purchase. And, they will setup straw-men by claiming falsely someone has said X and Y -- which they will then argue against. They will call anybody who doesn't agree with them every name in the book. Google Cognitive Dissonance.

But, outside a forum they have no voice. And, they never will precisely because no one will hire a fanboy as they can't be trusted to be objective.

No matter how loud they scream -- in the world outside of where they post -- no one will ever hear them.

Can't say it has been fun, but it has been useful to read what folks have posted. So here is an image and if you don't see the moiré pattern on the brick -- what can I say. Time to start writing so "over and out."

[Mark Rosenzweig]

Originally Posted by Steve Mullen

Fanboys will always defend what they just bought to the bitter end. They are so motivated they will simply not see anything that might cause them to re-think their purchase. And, they will setup straw-men by claiming falsely someone has said X and Y -- which they will then argue against. They will call anybody who doesn't agree with them every name in the book. Google Cognitive Dissonance.

But, outside a forum they have no voice. And, they never will precisely because no one will hire a fanboy as they can't be trusted to be objective.

No matter how loud they scream -- in the world outside of where they post -- no one will ever hear them.

Can't say it has been fun, but it has been useful to read what folks have posted. So here is an image and if you don't see the moiré pattern on the brick -- what can I say. Time to start writing so "over and out."

Name calling is the last resort of someone trapped in a corner and who is stubborn.

1. You can see the test charts on Slashcam.de. They show that the Ax100 has less moire than the GH4, the AX1 and the Canon 1 D C. It in fact has none, while the others clearly do. Slashcam.de is not a fanboy. Those tests confirm what us fanboys see. They clearly contradict what you think you see, which is based on not understanding what you are doing. Do you have a critique of the Slashcam.de tests? or do you want to call them names?

2. Your picture is what the clip looks like on your computer; it is not a full-resolution frame grab from the original video but is a picture of your computer screen showing a video! LOL. You still do not get that what your computer is showing you is not what comes out of the camera in 4K. And you have clearly demonstrated with this picture and your fumbling with an editor a biased and stubborn quest to show something that patently is not there. And you do not know how to test camera video.

For your edification:

"As you can see , the Panasonic in this league is not the worst and not the sharpest 4K camera but takes a healthy midfield. The Sony FDR AX100 shows the rings in the far cleanest imaging behavior , while light moire / aliasing structures occur at the Panasonic GH4 , the Sony AX1 and also the Canon EOS 1D C in the rings. Obviously Sony uses the full sensor readout more pixels reserves before down scaling. Canon , Panasonic, and Sony's AX -1 , however, put on a similar debayering algorithm that just can not get more out of 4K Bayer pixels. However, what Sony dissolution victory verhagelt again , is the strong sharpening and the high contrast that is not moving back at the Sony AX100 . The possible dynamics of the Panasonic because we like much better.


The original German:

"Die Sony FDR-AX100 zeigt in den Ringen das bislang sauberste Abbildungsverhalten, während bei der Panasonic GH4, der Sony AX1 und auch der Canon EOS 1D C leichte Moires/Aliasing-Strukturen in den Ringen auftreten. Offensichtlich nutzt Sony beim kompletten Sensor-Readout noch mehr Pixelreserven vor der Downskalierung. Canon, Panasonic und die Sonys AX-1 setzen dagegen auf einen ähnlichen Debayering-Algorithmus, der aus 4K Bayerpixeln eben nicht mehr herausholen kann. Was Sony Auflösungs-Sieg jedoch schon wieder verhagelt, ist die starke Nachschärfung und der hohe Kontrast, der sich bei der Sony AX100 nicht zurückfahren lässt. Die mögliche Dynamik der Panasonic gefällt uns da deutlich besser.

I read this as: the AX100 has the highest,cleanest resolution, but there is artificial sharpening and too high contrast.

The link: http://www.slashcam.de/artikel/Test/...-Schaerfe.html

[Dave Blackhurst]

@Steve -

I don't really know what is going on with your system... but the still you posted is SIGNIFICANTLY degraded over my pulling up Ken's Vimeo post, and zooming in up to almost 500% (yeah, who watches video that way?). More than anything, your still looked like mush... where the video viewed on a 1080 screen looks very sharp, even zoomed in a realistic amount...

I did notice that when the video was not full frame, there was a lot of moire... which disappeared almost completely when I hit the full frame button/view (OK, so there was a tiny bit of mosquito noise... not sure it can be avoided at this bitrate) - again, this is an indication not of the camera quality, but of the quality of the interpolation between the two window sizes!

Darrin's tests also exhibit minimal problems in patterned areas that will fall apart with most cameras, as his testing shows...

I don't doubt YOU are seeing something, but I think the rest of us are questioning the methodology to arrive at the conclusion...

Just remember one thing... the internet provides a platform where someone can provide evidence quickly and efficiently of what they claim. So far the footage posted does not support your conclusions, OR you allegations of "fanboy-ism"... I don't have a purchase to "defend", heck, I'm waiting for the first few of these to pop up as "return/open box" at a significant discount, from people like the guy who said it's a worthless brick! SOMEONE "hated" my RX10, thank goodness! I continue to be happy and impressed with it, horrible imperfect thing that it is...

I am reviewing AX100 footage others are posting (yours included), I do see temporal issues that arise from "low" 24/30p frame rates, but no way it's a deal killer, nor am I seeing terrible aliasing/moire that would disqualify the camera from practical use.

I have yet to download my blood moon footage, but I think I had far better results than your sample with my PJ760 (w/ 2x tele lens), and I got decent results with the RX10 at full digital zoom... and yep, I was thinking an AX100 would have been useful, if I had one in hand.



Back to the 4K trail...

Picked up a cheap used 4K TV, only to scratch my head about what I had that could output to 4K... no computers I've got will output 4K at the moment (memo - need to find parts for a cheap 4K build!)...

Finally dug up an HDMI to allow me to hook up the RX10... in stills mode it outputs 4K, and the difference between 1080 and 4K was quite convincing at a short distance - I'm sure it would be less noticeable at greater viewing distances (my eyes don't resolve at distances that well!), but I definitely notice it's "easier" on my eyes the sharper the image, so I'm happy with "step #1". One thing I noticed is a couple photos which looked almost perfectly sharp at 1080 had noticeable, if slight focus issues at 4K! Focus issues are going to be more and more "fun" with these high resolutions!!