PMW 350 Video Overview & Test Footage!!!

[Bob Grant]

Originally Posted by Joakim Sandstrom

If something is better than the other that purely up to taste, this was a scientific theory about accuracy. CCDs seem to be more accurate, perhaps because of the higher fill factor. But the CMOS look may be more artistic, this is what people love with the RED for example. CMOS sees light differently. CMOS has a more film-like rolloff in the highlights. Alister pointed that out in some EX350 test. And the somehow muted color palette is very *in* at the moment. I think the 3-CCD sony look offers something new, never before seen, hyper color definition, even better than super 35, perhaps because of it's accuracy.

As far as I'm aware both CMOS and CCD use the same device to convert photons into electrons. The difference is in how the electrons are handled. The difference in fill factor would affect the size of the actual photodiodes and hence lattitude / sensitivity. I cannot see how there would be any difference between CCD and CMOS that would affect color response as that's detemined by filtration before the detectors.

[Adam Reuter]

I have to pipe in here guys with the CMOS vs. CCD debate. I have to ask how everyone is viewing this footage. On a computer monitor (i.e. sRGB-calibrated) on a "real" HD monitor that's calibrated to 709. Because if your answer is the former than you need to re-evaluate. Because I am still to do this day (because I am not monitoring properly) dealing with issues of saturation/contrast in my video and editing on a computer monitor.

In my case I think it boils down to my software is not converting colors properly to be viewed on an sRGB calibrated monitor...which I use mostly for photo to print editing. Video is secondary to my photography, unfortunately.

The images I get from the EX on real HDTVs look just as good as any CCD HD camera...saturation-wise. I do think the colors pop a bit more now with the T1 filter on, but even looking at my footage pre-Tiffen T1 the colors still pop. I think it's all about the Picture Profile...which Sony leaves at a neutral value PURPOSELY from the factory.

My Nikon D300 CMOS still camera has much better saturation than my Nikon D50 CCD camera. And as a matter of fact I think the chips in Nikons are Sony or Sony-based. EXSPEED or something like that?

[David Hart]

Originally Posted by John Poipie

Well Guys,
CCD or Cmos is a nice discussion, but to tell you the truth it doesn't matter to me. What I want is a camera that performs well, looks well with a reasonable price and meets my technical and creative possibilities. This 350 has a lot of these potentials.
The look is great; it will open doors for you
The picture looks better than the EX3
The low light performance is very good
The available buttons for div functions and manual operation is very professional and useful; power switch is what we need compared to the EX1 and EX3. The ND filters are now what they should be (4 positions)
The VF is good to excellent; no need for extra hood
The In/outs are on the right place imo with the right connection sockets
The weight is normal
The power consumption is good compared to other shoulder babies
The price is right
The use of "cheap" media(SD memory cards) next to the SxS cards to record to, makes it very attractive
The Stock lens is good and with autofocus(!?)
The shooting modes are for all trades; SD/HD Pal/NTSC
Still to consider are the Flash(light) problem with the Cmos(there we go again) and the IR problem if it still exists.
What I like to see in the package is: Battery/Charger/SxS card 32Gb/Quick release plate
To be short: great camera with reasonable price.

Hi John
Totally agree, i have purchased one a few days ago and the test shots are fantastic!

Regards David

[Joakim Sandstrom]

Originally Posted by Bob Grant

As far as I'm aware both CMOS and CCD use the same device to convert photons into electrons. The difference is in how the electrons are handled. The difference in fill factor would affect the size of the actual photodiodes and hence lattitude / sensitivity. I cannot see how there would be any difference between CCD and CMOS that would affect color response as that's detemined by filtration before the detectors.

If theres no difference then how come CMOS sensor cameras have IR problems that CCD cameras do not?

[David Heath]

Originally Posted by Joakim Sandstrom

......how come CMOS sensor cameras have IR problems that CCD cameras do not?

You make it sound like all CMOS cameras have IR problems, all CCD cameras don't, and that's not true.

What mainly determines the cameras level of response to IR is the spectral response of the IR cut-off filter before the sensor. The beamsplitter should ensure that the IR is only directed towards the red chip - but that will give output for any wavelength light (or infra-red) that falls upon it, it's only a "red chip" because only longer wavelength light is directed towards it.

Making filters with very sharp cut-off frequencies is very difficult - it's nearly impossible to block all infra-red without affecting the red end of the spectrum. It's all a question of what designers rate to be the best filter compromises. But CMOS v CCD is irrelevant here - it's all down to the IR cut filter.

[Alister Chapman]

I wonder if it has something to do with the prism design. CCD sensors can only be read one way, so the image arriving on the face of each of the chips must be the same way up, in practice this means that each channels light path through the prism involves 2 bounces for each of the red and blue light paths (mirrored plus mirrored equals right way up) or is straight through (right way up) for green. CMOS sensors can be read pretty much any way, forwards, backwards, upside down etc. So this means the prism can be simpler as it doesn't matter if you only have a single bounce (mirrored) for red and blue and straight through for green. This makes the CMOS prism smaller, lighter, cheaper. The extra bounces in a CCD prism may have the effect of giving a sharper cut off for the blue and red channels. Just speculating.

[David Heath]

An interesting thought, but I doubt it makes a difference. I'd expect the colour separation within the prism to only work around one wavelength at a time - so (allowing for overlapping) shorter wavelengths get sent the blue route, longer ones the red route, with the inbetween going the green route.

So the red/green separation doing just that, rather than band filtering and cutting out IR.

And what about single chip cameras? Some are affected, others aren't.

[Alister Chapman]

You may be correct, but each reflection in the prism is created by a dielectric mirror coating which can act as a band pass or cut-off filter, so with 2 reflections it would be easier to achieve a sharper cutoff than with a single reflection. In most 3 chip CCD prisms the blue is picked off and sent to it's sensor in the first prism block by a high pass mirror dielectric coating on the back of the first prism block, green and red get passed through a band pass filter and in the next bit of glass the red is bounced by a low pass dielectric mirror and green passes through a further band pass. Before getting to the sensor the red gets another bounce via either a Hi/low pass or bandpass coating, so the red goes through 4 possible filtration steps. With a Sony CMOS prism the green is picked off in the first block with a bandpass coating and has a second reflection before reaching the sensor. In the second prism block blue is picked off via a high pass and goes straight to the sensor, red passes straight through the entire assembly directly to the sensor, so red only get 2 stages of filtration, one of which is pretty broad band as it is just picking off the green.
With a bayer filter you have absorption filters in front of each pixel and the quality of these will affect the pixel response. It's very difficult to make an absorption filter with a very sharp cutoff which is one of the reasons why the colour response from bayer cameras is rarely as precise as 3 chip designs, it would also be more difficult to give a sharp IR cutoff. Of course with either there is (or at least should be) some filtration ahead of the sensor block and this is normally where any IR cut takes place. At the end of the day it's probably down to cost rather than what can or can't be done.

Good to see you at BVE.