Sony FDR-AX100

[Anthony Lelli]

Originally Posted by Alister Chapman

Owning a great camera doesn't make you a great camera operator. A poor camera in the right hands will always result in better images than a brilliant camera used incorrectly.

Sadly Anthony thinks he knows it all and is not prepared to accept that frame rate, resolution and sharpness play a massive part in image judder. He is determined to find fault with the camera rather than perhaps his technique. His comments make it plain to everyone that he does not understand how cameras and human vision works. Does he think that Peter Jackson and James Cameron shoot at 48fps just for kicks?

look , of course twice the frames are better , but that doesn't explain why a building becomes curved for a while when panning. Guessing may be an explanation for that. since it's not optical, must be something else that more frames can't explain.
more frames per second are mostly for slow motion. Don't get smart on me now. Let's try to explain the building curved on pans. regardless of the number of the frames involved.

[Alister Chapman]

Curved objects = rolling shutter.

EX1 sensor = fewer pixels + designed for video = faster read out = less rolling shutter.
AX100 sensor, designed for photos + more pixels = slower read out = more rolling shutter.

EX1 $8K
AX100 $2K

You get what you pay for.

[Alister Chapman]

My AX100 review is here:

Sony AX100 4K camcorder review. - YouTube

[Cliff Totten]

Sony, can you PLEASE add a firmware update that allows us to assign the top handle button (the "photo" button) for focus expansion?

The button you have today for that is in the absolute WORST place on the camera. It's absolutely impossible to reach with your finger. And, I'm absolutely certain that no Sony engineer, designer and camera tester can reach it with their fingers either. I'm quite surprised that an important function like this was allowed to exist in such and knowingly inaccessible place.

You did this for the NX70, please for it for the AX100. It's a bit of a embarrassing joke right now that Sony can easily fix today.

I was hoping Alister would mention this out loud in his review. (They might actually listen to him if he shined the light on it...and I'm sure he must agree with me on this.)

CT

[Noa Put]

I would be surprised if Sony would make a change true a firmware update, on the nex-ea50 there where many features asked by users to improve upon or to add of which the possibility to change the iso with the dial on the side (like you can with the shutter) instead of having 3 fixed selectable iso values with a switch which was my top one on the list. Sony did release one firmwareupdate but that included options nobody asked for.

Improvements are usually made with new models, like when people ask for the possibility to have 4K on the rx10 which I understand the sensor is capable off yet I"m sure that won't happen, unless they bring out a rx10 4K model.

Originally Posted by Cliff Totten

I was hoping Alister would mention this out loud in his review. (They might actually listen to him if he shined the light on it...and I'm sure he must agree with me on this.)

As much as Alister is a respected filmmaker, I don't think Sony would listen to his reviews and make changes based on his findings, at least not on existing products, by the time Alister has a review ready Sony already has another camera in the pipeline with all focus on that one :)

[Darren Levine]

Your 'curved' building problem is called rolling shutter. When a sensor reads the lines of pixels from top to bottom instead of all at once. That's what causes your curved buildings and by far the main factors of its severity are resolution and read speed.

[Cliff Totten]

BTW Alister, great job on your three AX100 videos. (Focus, overview and exposure)

I would love to see you do a "technical evaluation" too. Maybe dynamic range and resolution chart tests? (Sony AX100 vs Sony AX1 or Z100?)

What is your opinion on the 60mbps codec limit? I have noticed with a bitrate analyser that the codec will actually do up to 72Mbp/s spikes at certain times.

I find myself always expecting much worse results than I actually get with this codec. It's surprisingly "OK" (but I still have a hard time accepting it) I have noticed however, that this codec breaks quickly when trying to grade something. If you shoot perfectly in the field than you will be OK but making any significant adjustments in post makes compression artifacts "pop" out quickly.

What do you think Alister?

CT

[Alister Chapman]

Cliff, your assessment of the codec is pretty good. As you say it works very well considering it's only 60Mbps. There are a lot of artefacts in the recorded image but these don't tend to be too noticeable unless you grade the image. I don't think I would like to pass the 4K though a broadcast chain, although it does hold up really well when going to YouTube etc. What really surprises me is that even very rapid major changes to a frame are handled very well, for example fast pans or lightning bolts won't break the codec. It does soften (as expected) but does not break down as many other H264 based codecs can.

I would eyeball the dynamic range at about 10 stops. Onset of over exposure is a little sudden, I'd like to see a bit more roll-off, but for a camera in this price bracket it's pretty good. I don't have any 4K resolution charts, I really need to get some and my Imatest licence has expired. It's very obviously significantly higher resolution than HD.

I do need to do some tests on the image sharpness. It is very sharp and I would like to ascertain how much of that is just pure resolution and over-sampling and how much is in camera sharpening. I also want to look at what happens when you add a touch of diffusion to take the edge off the image. It's just this stuff takes time to do right and I've been really busy lately.

For me this camera is very much like the HC1 that came out in the early days of HD. The HC1 (and A1) was never quite as good as the Z1 etc, but for the money it produced a very useful image and the AX100 is similar. I prefer the AX100 over the Z100, but it's not an F5 or FS700 with raw recorder.

[Pete Bauer]

MODERATOR NOTE:
As those posting to this thread over the past couple of days will notice, it has been heavily pruned to remove the worst of the rudeness and flaming. Be nice.

[Anthony Lelli]

Originally Posted by Darren Levine

Your 'curved' building problem is called rolling shutter. When a sensor reads the lines of pixels from top to bottom instead of all at once. That's what causes your curved buildings and by far the main factors of its severity are resolution and read speed.

It's the readout of the sensor, correct?
too slow?

So how about the framerate now? nuthing to do with THAT, correct? And now the fact that with the stabilization off it gets better may be due to the other fact that the stabilization takes processing power away from the readout. It all makes sense now.

So it wasn't the 30p thingy. Well put Darren!

[Noa Put]

Originally Posted by Alister Chapman

I do need to do some tests on the image sharpness. It is very sharp and I would like to ascertain how much of that is just pure resolution and over-sampling and how much is in camera sharpening. I also want to look at what happens when you add a touch of diffusion to take the edge off the image. It's just this stuff takes time to do right and I've been really busy lately.

Where do you post these test or review clips Alister? When I went to your blog I couldn't find the ax100 review clip back and it appears you have published some more videos about the camera? I"m probably looking in the wrong place but would be interested in following your findings if I know where to look.

[Alister Chapman]

Judder and rolling shutter are two different things and neither has anything to do with processing.

Judder is related to frame rate, resolution and shutter speed. Bending verticals is rolling shutter caused by the slow scan rate of the sensor. More pixels take longer to read so unless you have a very fast sensor the readout rate slows down. This is very typical of sensors designed for stills rather than video.

Faster shutter speeds will make rolling shutter more obvious as there is less motion blur to disguise the artifact.

The videos are only on my YouTube channel at the moment. A written review will go online some time next week along with links to those videos. More in depth tests will follow when I have time between shoots.

[Dave Blackhurst]

Originally Posted by Wacharapong Chiowanich

Alister,

There 's an interesting test of the active steadyshot on the Sony CX900 which also engages the Clear Image Zoom. Though the test was done at 1080p resolution as the the CX900 doesn't shoot 4K I feel this is quite similar to what I got when zooming in near the tele end when in active steadyshot mode in 4K. At the wide end or close to it the difference in resolution is not really noticeable in practice. Over at the dpreview.com site they also found something similar when testing the Sony RX100 Mk3 with the test chart when active steadyshot was used in the video mode as well.

- Stabiliser test: Sony HDR-CX900 stabilizer - YouTube

I haven't yet shot anything at 1080p with my AX100 but given the two cameras have more or less the same hardware the results should be the same at that resolution.

This sort of "test" is exactly the sort that drives one nuts - OF COURSE, when you FULL ZOOM with CIZ on, it's a digital doubler, but if you compared 12x and 12x and took note of the line in the zoom indicator, the two will be comparable... maybe it's the "Queens' English" vs. US English, but I'd give that "test" a big thumbs down as the narration is misleading and inaccurate... indicating to me the "tester" does not know how these things work....

Most likely the CX900 and AX100 1080 modes will be identical... but of course at 4K, detail is an entirely different ballgame.

Yes, a small crop is required for "active" stabilization. Pretty much like saying you need eggs to make an omelet... and if you use egg whites it'll be a different color...

[Noa Put]

Originally Posted by Alister Chapman

The videos are only on my YouTube channel at the moment.

Thx Alister, will check them out.

[Anthony Lelli]

Originally Posted by Alister Chapman

Judder and rolling shutter are two different things and neither has anything to do with processing.

Judder is related to frame rate, resolution and shutter speed. Bending verticals is rolling shutter caused by the slow scan rate of the sensor. More pixels take longer to read so unless you have a very fast sensor the readout rate slows down. This is very typical of sensors designed for stills rather than video.

Faster shutter speeds will make rolling shutter more obvious as there is less motion blur to disguise the artifact.

The videos are only on my YouTube channel at the moment. A written review will go online some time next week along with links to those videos. More in depth tests will follow when I have time between shoots.

The bad panning of the AX100 has everything to do with the processing. The shutter speed of my choice for players rtunning at night is 1/125. I'd rather have more distractions in the back than a blurred player in front of the viewer. I know.. I tried yesterday. The framerate in that regard makes a little or no difference regardless of the internet myth. Like I said. Then a cameramen must make choices considering what's the best at the moment and I did.