Sony launches PXW-Z100 4K Handheld XDCAM

[David Heath]

Originally Posted by Cliff Totten

Anybody care to speculate how many lines of resolution this two cameras would resolve?

Depends on the lens, but with good normal deBayering 80% of the sensor resolution is a decent first estimate: hence 80% of 2160 lpph - so 1728 (probably more realistically around 1700)

Note that's for luminance. Chrominance will be somewhat lower.

With Bayer and Nyquist issues calculated, would 2000 lines be out of the question?

I'd be amazed if anyone got that....... :-)

[Matt Davis]

With due respect, if we're into counting the number of photons dancing on the head of a pin, don't forget that Sony do a very nice line in 4K Cine cameras - F5/55. I fear that the lens may be the limiting factor here.

Remember (oh sheesh, I'm going to sound old now) when the Z1 came out? HDV? I distinctly remember the moment: DSR570 shooting DVCAM, Z1 shooting HDV, downsampled to DV50, then pushed through a really early Colorista. The DSR was actually better, but in 80% of our shoots we couldn't afford a DSR, and with the Z1 i) the shoots would happen and ii) we could do Colour Correction on them. I could see the difference, I could argue for the DSR570, but would the client?

Fast Forward to the Z100. If you're shooting narrative on a microbudget, Black Magic's got you covered. But here's the beef: would you prefer the BMCC or the Z100 if I got you to cover the Notting Hill Carnival by yourself?

What will the camera resolve? Probably not as much as some cameras. But if you downsample its 4K to HD, a lot more than others! LOL

[David Heath]

Originally Posted by Matt Davis

With due respect, if we're into counting the number of photons dancing on the head of a pin, don't forget that Sony do a very nice line in 4K Cine cameras - F5/55.

Forgive me if I misunderstand, but it sounds a little like you think the F5/55 should be superior in predicted resolution, at least in terms of the sensor alone?

If so, it's not true. Neither is it true for any Bayer 3840x2160 sensor - the best you're likely to get out of them in terms of luminance resolution is about 1700 lpph. This is inherent in the working of the Bayer sensor and the deBayering process. Figures vary, but comments like "halfway between true sensor resolution and 50% of it" and "about 80%" seem the norm and generally accepted. One things for sure - you WON'T get 2160 lpph out of such a sensor, be it an F5/55, an PXW-Z100, or one from anybody else of those dimensions.

What I'm NOT saying is that it must therefore follow that the an PXW-Z100 is therefore just as good as an F5/55. That would be to ignore lens issues, but most importantly all sorts of other factors such as diffraction limiting and dynamic range to name just two. What I would say is that because of the sensor dimensions, the MAXIMUM resolution EITHER camera could be expected to give is around 1700 lpph.

Practically, then I don't disagree with the practical points you make with reference to Z1 and DSRs. I was just replying to a straightforward "what resolution can we expect" question. And 1700lpph needs to be seen as a maximum figure for any 3840x2160 sensor - a camera may meet the figure in centre frame, but not at the edge, due to lens issues, for example. What it can never do is exceed (roughly) that figure - and that's as true for an F5/55 as for a PXW-Z100.

[Alister Chapman]

If it's using the normal Sony Cinematone gammas and 709 with knee it should be around 10 stops which is is pretty good from the approx 1 to 1.5 micron pixels (given the lenses focal length the active area of the sensor may be about 1/3").
The EX1/PMW200/PMW/300 pixels are roughly 3 microns, F5/F55/FS700 roughly 5.6 microns.
As there is a direct correlation between pixel size, dynamic range, sensitivity and signal to noise ratio, even though this is a back illuminated sensor (which will help make up for the light lost in the bayer filter) I think the numbers above will tell you where the Z100 and AX1 will sit. Noise reduction has improved in recent years so noise can be better masked although this comes at the expense of smear and some image softening. But noise reduction becomes less and less effective at higher gain levels.

Can't help but observe that the specs for the NX5 lens and Z100 lens appear identical: NX5U f = 4.1 to 82.0 mm (equivalent to f = 29.5 to 590 mm at 16:9 mode, f = 36.1 to 722 mm at 4.3 mode on 35 mm lens)
Z100 f = (equivalent to 29.5mm to 590mm)

But Sony are not publishing the actual focal length for the Z100. Is this because it would tell us the actual active area of the sensor????

[Cliff Totten]

I have never scaled down 4k video for delivery to HD before. Now, I know this is a complex question and I'm not quite sure how to word this;

When shooting in 4k, the image "noise" or "grit" or "dirt" should be very fine and detailed. Now, in theory, when it is scaled down to HD, the noise should become "smaller" and less noticeable?

Or....

Will the noise be a certain "size" based on the surface area of the entire sensor and not the resolution or density of it?

That's a strange way to put it but I'm trying to think of any advantages of deliberately shooting 4k for the purpose of HD delivery. (noise "perception" is reduced??)

Does that make any sense? lol

[Gints Klimanis]

Overall, downsampling averages and throws away detail, be it desired (image data) or undesired (noise), for which the target image size doesn't have the resolution to represent.

It's impossible to do a perfect downsampling without any spacial artifacts. The sharpness advantage of 4K-downsampled-to-HD over native HD you may see is largely due to the optical anti-aliasing filter fixed to deal with the highest spatial sampling of 4K.

[Ron Evans]

I did this cropping when I first got my FX1 HDV to SD. It worked very well and expect to do the same with hopefully the FDR-AX1 when I get one !! The noise will stay the same size relative to the pixels. If you just crop a HD sized image from the 4k image you will be taking roughly a 2k image( image and noise ) from the 8k source. In viewing terms the pixels will look 4 times bigger and so will the noise compared to the 4K image on a 4K TV. On a HD TV I don't think you will see the difference. The more you zoom in of course the more significant the effects noise may become as you will now be using a scaling function and how you do this will govern the output. When I did this with HDV to SD DV I fixed the scale as a crop and just panned around or choose a different position in an attempt to eliminate any scaling issues.

Ron Evans

[Alister Chapman]

It depends on how you do the down conversion.If you do a simple down scale then the noise will have a similar level but with larger effective grain size. A good quality anti aliased downscale should result in a very slight reduction in noise as adjacent 4k pixels will be averaged together so any random noise ends up getting reduced very slightly.

The sharpness or resolution of a good quality video camera is governed by the optical low pass filter. These filters are always a compromise as the cut-off is not instant. So an OLPF designed for an HD camera will start reducing contrast quite some way before 1080 and then still allow some detail beyond 1080 to leak through which manifests itself as a small amount of aliasing. As a result you will get a gradual fall off in contrast as you approach the cameras maximum resolution and this has a noticeable effect on the perceived sharpness of the image. If you start with a nice 4k Bayer sensor the you should have a resolution of around 3.2k which you can down convert digitally to HD. A digital low pass filter can be far sharper and more precise than an optical filter so you can get better contrast at the limits of the down converted resolution.

But all of this depends on just how well you do the down conversion and it may take a bit of work to find the optimum settings, plus some decent computing power to do the conversion quickly. Get the down conversion wrong and you will get aliasing artefacts that may spoil the image, possibly not obviously so at the first generation but something that may appear after subsequent encoding passes.

[Matt Davis]

Originally Posted by Alister Chapman

Can't help but observe that the specs for the NX5 lens and Z100 lens appear identical: NX5U f = 4.1 to 82.0 mm (equivalent to f = 29.5 to 590 mm at 16:9 mode, f = 36.1 to 722 mm at 4.3 mode on 35 mm lens)
Z100 f = (equivalent to 29.5mm to 590mm)

But Sony are not publishing the actual focal length for the Z100. Is this because it would tell us the actual active area of the sensor????

I've asked, but answer came there none. Hmmm.

So... I sort of did an unscientific test by comparing the EX1 and Z100 on top of each other and observing the difference in angle of view - the EX1 is a little bit wider. 6-8% or so?

If the EX1 lens 'sort of' starts at the equivalent of a FF35 32mm lens, we might say that the Z100 kinda looks like a 35mm lens? But my figures don't tally with yours probably because I forgot to factor in the 16:9-ness. I don't have an NX5 to compare with.

But I do feel many videographers should be factoring the cost of a 'Wangle adaptor' - I wonder if the NX5 kit (with wide lens shade and French flag) would suit?

As for the 709 thing - here we get into a whole new ball of wax. Please do correct me if I'm wrong, but one of the side benefits of 4K was a bigger colour gamut - bigger than Rec709, probably the DCI P3 thing, plus the shift to 10 bits per channel too. Hence, 422 10 bit HD looks very nice upsampled to HD, whereas 420 8 bit stuff doesn't, therefore a good idea to shift to 10 bit. I am clinging on to this info by my fingernails and liable to oversimplify!

PS - @David, no, I was just wanting to distribute the point that the Z100 wasn't going to be a 'cheap F5'. LOL

[Ron Evans]

It remains to be found out the differences between PXW-Z100 and the FDR-AX1. This is the AX1 page on the Sony site Sony 4K/60p Ultra HD Camcorder | 4K Camcorder Review | FDR-AX1 | Sony USA Lens specs look identical to my NX5U. The FD-AX1 sensor looks just like my Cybershot HX30V, same size, same sensor count 18.9MP. The HX30VB takes stills at 4,896×2,752 and will do so while shooting video at 60i with OIS and Active EIS !! If you set it for 60P it cannot takes stills at the same time . Probably drain the battery and get too hot to hold !!! Does the PXW-Z100 have a different sensor to the FDR-AX1 ?

Ron Evans

[Glen Vandermolen]

Originally Posted by Ron Evans

Does the PXW-Z100 have a different sensor to the FDR-AX1 ?

Ron Evans

I'm betting the sensor is the same.

As far as the Z100 being a cheap F5, or even a cheap F55 - yes! For me and my budget, anyway. I can't justify getting an F5, much less an F55, but the Z100 should be within my reach - barely. Besides the no raw video option, doesn't the Z100 have the same XAVC codec as the F55?
For $6,500 I don't expect an F55. I'll settle for a 1/2.3" 4K camera.

And I'd pick the Z100 and AX1 over the cheaper BM4K, just because of the ease of use. No lenses, no rigs, no external battery source required. Just unpack, add a battery and media card, and off I go.

[Ron Evans]

I think the sensor is the same too. Sony marketing giving differentiation for a year before installing the LongGOP firmware. If that was available at the introduction the PXW-Z100 would be my choice too. However for my theatre show stuff the FDR-AX1 is a nice replacement for my NX5U even at 1920x1080 doing 60P a nice match with my CX700 and NX30U with 4K as a bonus. May indicate that the HDMI output from the FDR-AX1 may be 10bit 4:2:2 at some point with the HDMI 2.0 firmware update !!!

Ron Evans

[Jack Zhang]

As it stands right now, I still consider this the Z1U of the 4K generation. Give it another 2 years and the imperfections like the minor CA and the bayer artifacting and interpolation will be greatly lessened in a newer model that would be the EX1 of the 4K generation. (which would include manual controls instead of servo)

I also wonder if we'll see the return of Letus adapters if the DSLRs don't catch up to 4K in time, which I wouldn't doubt within the next year.

[Ron Evans]

I think its more appropriate to consider the PXW-Z100 the NX5U of the 4K generation and the FDR-AX1 the AX2000 of the 4K generation since they look to have used a lot of these models in the construction and retained AVCHD recording too. I don't think I will see much difference operating one of these cameras to my NX5U !!! Same batteries, Lanc controller, mic etc. The AX2000 had a smaller battery, no mic, less menu controls, no HD-SDi, timecode input and did not have the ability to use the FMU128 compared to the NX5U. Looks like the FDR-AX1 does not come with a mic, no HD-SDI .........

Ron Evans

[Jack Zhang]

I'm more referring to the image quality rather than the functionality. I personally want to stick to my BP-U workflow and semi-manual Fujinon lens. It would be very bad if the BP-U line of batteries is retired after the PMW-300. Only with the BP-U lineup can you get a 96Wh battery.