This keeps cropping up here and there. The question being asked is why, as it is a new camera, isn’t the PMW-200 just as sensitive as cameras like the PMW-F3 or FS-100. Why didn’t Sony design new sensors for the PMW-200 with the same low noise as an F3?
Let’s take a look at this, first what would happen if you fitted the PMW-200 with the sensor from an F3 but only used the centre 1/2″ portion? Well the resolution would fall short of what is needed for SD, let alone HD, because the cropped area would not have enough pixels to produce an HD image, you need the whole F3 sensor area to get enough pixels. What does that tell us about the F3’s pixels? It tells us that the pixels are much bigger than those in the PMW-200.
Pixel size is the primary thing that determines the signal to noise ratio of the camera. Cameras like the FS100,F3, C300 and GH2 etc have big sensors with big pixels, that’s why they have low noise. That’s why 1/3″ cameras don’t do as well as half inch and half inch doesn’t do as well as 2/3″ and so on. It’s down to the laws of physics. Over the last few years any noise and sensitivity improvements in sensors have been tiny, what we have seen with the large sensor cameras is simply the result of bigger pixels on a bigger sensor. Modern sensors like the ones in the EX have QE’s (Quantum efficiency) approaching 70% where 70% of the photons of light falling on the sensor are converted to electrons. If you want a bigger output and thus a better ratio of signal to noise then you use bigger pixels so that you capture more photons and as a result get more electrons. To do that without sacrificing resolution you need a bigger surface area and thus a bigger sensor.
To expect a significant improvement in sensitivity and noise performance when the sensor size and layout is not changing is not realistic as there have not been any changes to the laws of physics or core sensor technologies in recent years. The PMW-200 is designed to be easy to use, have a good zoom range and be easy to focus. As many have already discovered using cameras with Super35mm sized sensors for run and gun, news and documentaries is often very hard. You can’t get fast compact high ratio zooms without spending a small fortune, the shallow depth of field makes focussing very tricky. There are many productions where a smaller sized sensor is more appropriate and that’s where the PMW-200 comes in.
This is such a great topic. I’m also very interested in the QE of sensors today.
I did my own test tonight between my EX1r and my new CX760. (little brother to the NX30) The end result is very predictable. The EX1R is the superior performer. The EX1r has 3 full 1920×180 1/2 sensors. The CX760 has only a single sensor that is slightly larger than a 1/3rd inch. The CX760 is also afflicted with bayer pattern layout. If that wasn’t enough,…this poor little sensor is cursed with having about 6 million TINY photosites on it. With specs like that, one would rightfully expect horrible performance.
On paper this is a true “David vs. Goliath” comparison. Here are two simple screen caps that I just grabbed.
EX1r – F1.9, 29,97p@1/30 shutter +9db gain
http://dl.dropbox.com/u/4564537/ex1%20vs%20cx760/EX1r_9db.jpg
CX760 – f2.0, 29.97p@1/30shutter +18db gain
http://dl.dropbox.com/u/4564537/ex1%20vs%20cx760/CX760_18db.jpg
My questions is this:
What in the hell is Sony putting on this tiny, little, HEAVILY disadvantaged, 6 megapixel, bayer pattern sensor that allows it to have a respectable showing against the EX1r??? The EX1r is the winner,…but given the CX760’s heavily outnumbered specs,..the comparison should not be this close. At 18db, the CX760 should look like a joke…but somehow it doesn’t!
Another question:
If they can squeeze so much output out of a inferior (on paper) little sensor,…why can’t that same technology be applied to 3 bigger and higher output sensors and produce proportional performance increases?
I understand QE (somewhat) but something is just not adding up the way it should.
Last question:
You have seen a PMW200. Is Sony using it’s “best” sensors and “best” processing on that camera? What are the “Handycam” guys doing that is so right?
Cliff
I tend to agree with the comments from Cliff. As an EX1 owner, I’ve always felt that its noise performance leaves something to be desired. I realise, of course, that there are myriad factors involved here (dynamic range, colorimetry, resolution, coding artefacts etc.) and, ultimately, the operational advantages of ‘professional’ gear like the EX1 (e.g. the fantastic manual lens controls) make the investment worthwhile. However, the surprisingly good CX760 noise performance cannot be ignored.
By the way, I’m happy to admit that when I first became interested in acquiring professional video gear around four years ago, I was totally ignorant about the true meaning of the half inch and two thirds inch sensor sizings – I actually thought that the EX1 would contain sensors having a diagonal measurement of around 13mm! I now, of course, know better (i.e. the EX1 sensors are approx 6mm in size). But is it possible that many current newcomers are unaware of this truth, and therefore imagine that cameras such as the EX1 contain sensors comparable in size to Super 35mm?
Yes, video sensor size equivalents stems from 4:3 sensor sizes, the EX diagonal is just over 8mm and a 1/3″ sensor is 6mm and 1/4″ is 4mm, which as you point out a striking difference between the 27-28mm of Super 35mm and why those big sensor cameras produce such stunning pictures.
The CX760 and NX30 and all that family do produce good pictures, when the light is good. As soon as light levels drop away though the overall image quality deteriorates as resolution falls away and smear increases as the noise reduction ramps up. There is a night and day difference between these cameras and cameras like the EX1 PMW-200, especially on a big screen where you can really see the resolution differences. There is also the way the footage handles in post to consider. These small cameras have certainly come a very long way in recent years, but you cannot escape the laws of physics.
Don’t get me wrong. I love my EX1r. I want it placed in my coffin with me when I die and it is easily my weapon of choice in a run and gun shoot. (My girlfriend hates it, she thinks I love it more that I love her!)
I didn’t push the cameras to their maximum gain last night. I think I will set that up again and see those results. (I’m almost afraid to do it now)
Here is the baseline, 0db for each camera.
CX760
http://dl.dropbox.com/u/4564537/ex1%20vs%20cx760/CX760_0db.jpg
EX1r
http://dl.dropbox.com/u/4564537/ex1%20vs%20cx760/EX1r_9db.jpg
As anybody can see, at the starting point of the race, the EX1r has a HUGE head start and that is exactly as it should be. The interesting thing to me though is how much ground those tiny little CX760 photosites covered 18db later.
EX1r – F1.9, 29,97p@1/30 shutter +9db gain
http://dl.dropbox.com/u/4564537/ex1%20vs%20cx760/EX1r_9db.jpg
CX760 – f2.0, 29.97p@1/30shutter +18db gain
http://dl.dropbox.com/u/4564537/ex1%20vs%20cx760/CX760_18db.jpg
Keep in mind,….this is a race between a Ferrari F1 and a 3 cylinder Yugo. The Ferrari wins but the Yugo is not the “distant second” that we all expect and should be laughing at. The CX760 makes a GIGANTIC Olympic leap from it’s 0db starting point.
In my humble opinion, the CX760 is trying much harder the the EX1r. This is not sitting well with me.
It it wrong for me to expect a PMW200 to exhibit that same kind of distance covered in it’s 18db jump?
In my mind, I’m expecting to see (key word) “proportional” performance increases in image quality when amplified at high levels.
Yes,….I’m only talking mostly about noise. Dynamic range and resolution is another subject, agreed. (although I would not call the resolution difference between the two “drastic”)
I find these results fascinating.
How odd,…..
Your scenes are static scenes. You can see the loss of resolution in the CX760 footage and it lacks the overall quality of the EX1 image.
I’m not questioning that the CX760 isn’t a great performer, but it just doesn’t produce as nice an image as the EX1 and the PMW-200 will be that bit better again. The real test come in a dark scene and how the cameras handle dark noise and motion. A still image is just what noise reduction circuits like.
In my opinion, the EX1r is a fantastic all around performer. It’s optics, sensor specifications and configuration allow it to easily produce the superb image that it does. This, I wouldn’t call “remarkable” because it all is, as it should be.
This little CX760 is HEAVILY disadvantaged in every category and specification on the list. It’s photosites are extremely tiny. The fact that Sony can bring this terrible sensor configuration up to such a “good” image quality,…is….well?….truly “remarkable”.
Did Sony create a “new” sensor set for the PMW 200? Or, did Sony take the exact same 2007 EX1 silicon and optics block and simply “add” a simple noise reduction algorithm in post processing. I want to believe that the PMW 2000 contains Sony’s absolute latest sensor and processing improvements.
Can I get the same PMW200 image with my EX1r using “Neat Video”?
Alister, if you have more EX1 vs. PMW200 comparison materials, would you be able to share it with us? I’m very interested in the noise performance between the two. I’m debating the idea of buying a new PMW200 or keeping my EX1r and going down the FS100 road. (I shoot allot of stage performances, sometimes in very low light)
Cliff