Quantifying Sensitivity

[paulcurtis]

Im looking for information, i've searched but not found exactly what im looking for and im also aware that 'specifications change' so historical threads might no longer be the case.

Im hoping some current red owners can help me

Im trying to quantify the sensitivity of the Red, compared to say the XDCAM range. None of the manufacturers really standardise on specs so i can't go by paper. Im aware of the EX1/3 and the PDW700 to a much lesser degree. The PDW700 is an alternative to the red for this project, price is similar. (Please! - im not starting a 4k vs HD i'm well aware of raw and workflows).

I know that the red is 'officially' rated at around 200 iso, so can i get a good idea running iso200 on a dSLR, how similar would it be in real life?

Has anyone done a quantifiable test side by side? I've looked at adam wilts test but that's not really focused (no pun intended) on this aspect.

Reason is simple, im budgeting and planning and my concern with Red is that i will need considerably more light and force changes to my existing lighting package (and all the associated pain that brings). Also it looks like 5000k lighting can make a real difference or is this just hearsay? (Even the EX works better under that kind of temperature but it's still perfectly usable under tungsten - i would expect the same from the red)

any help, pointers or guidance very much appreciated!

cheers
paul

[Shane Betts]

Hey Paul

The camera's 'base' ISO is 320.

As for the colour, yes, consensus seems to be that it looks better under daylight than tungsten as the blue channel is a little less sensitive and therefore more inclined to noise under tungsten.

Can't tell you exactly how much noise - still waiting for our Red to arrive:-) Soon, soon, soon.

[Markus Rave]

I find this hard to compare. For the XDCAM you opt for working with disks. This means you change disks when they are full. You opt for 2/3" CCD giving you much more DOF. With Red you need to backup media on set. You will shoot with 35mm lenses, shooting in night situation you want to rent HS lenses. You will need a good focus puller. You will need someone backing up data. And this will be in the same price range as XDCAM? I have severe doubts.

I think you should compare again adding needed options. Don´t know about XDCAM but my RED I rate at ISO 320, use my meter and everything else falls in place. But it is definetely not like shooting 2/3" CCD. Takes more time and care on set.

Markus

[paulcurtis]

I find this hard to compare. For the XDCAM you opt for working with disks. This means you change disks when they are full. You opt for 2/3" CCD giving you much more DOF. With Red you need to backup media on set. You will shoot with 35mm lenses, shooting in night situation you want to rent HS lenses. You will need a good focus puller. You will need someone backing up data. And this will be in the same price range as XDCAM? I have severe doubts.
Markus

Markus, i'm aware of the workflow differences and what it takes to shoot with both sides, thank you. That's not at issue, i don't want this to be a Red vs Whatever thread! I like the Red, and if it works for my project i will buy or rent one (or two)

But i like to know what im getting into and it does make a difference if i *have* to switch to HMI packages, if i need a lot *more* light. Maybe i don't? I have location restrictions to consider as well as wanting to take advantage of existing gear.

Hearsay says that EX1 is ISO 640 and Red 320, but it's hearsay. Has anyone shot side by side? Are there empirical examples? If the EX is 640 that's a stop advantage but then EX at 3db of gain is perfectly acceptable and that would be two stops.

(Also i have to add - don't think for a second i am comparing an EX1 to a Red, Im using EX1 as an example because i have one sitting in front of me and i know it well and i know how it relates to some 2/3rd cameras)

These are all numbers though, i'd rather hear from people in the field.

thanks!
paul

[Charles Angus]

I'm sure this has been broached elsewhere, but we desperately need a new description of sensitivity for the digital age. I don't think I'm the only one who feels that ISO is fairly meaningless when describing digital.

[Daniel Browning]

Im trying to quantify the sensitivity of the Red, compared to say the XDCAM range.


The one true measurement of sensitivity is Quantum Efficiency (QE). That's the percentage of light falling on the sensor that gets recorded. Most cameras convert between 20 and 70% of the light that hits them.

Your post also makes me think that you also care about read noise performance, which is a separate (and very important) measurement, because it determines how noisy the shadows are, dynamic range, and other things. It is measured in electrons (per ADU or per square micron).

Measuring both of these aspects with a RAW camera is very easy. Few cameras have their QE and read noise in the produce specifications; more often it will be calculated by the first folks that buy the cameras.

Doing the same measurement with a video camera that lacks RAW is much more difficult, because the original linear light measurements have gone through so much in-camera processing.

One method is to shoot a wide dynamic range (e.g. Stouffer chart) with both cameras on the same f/stop and shutter, have them processed, and compare the output. The one with noisier highlights has lower quantum efficiency, the one with noisier shadows has more read noise. However, it's impossible to process them exactly the same because they're not RAW, so the test will not prove anything about the cameras themselves: only how the cameras respond to such-and-such processing (and in-camera settings).



None of the manufacturers really standardise on specs so i can't go by paper.


Of course, even if they did specify the real performance numbers of the sensor/electronics before in-camera processing, it wouldn't tell you what kind of image you get out of the camera because so much is done by the various curves, saturation, compression, etc.



I know that the red is 'officially' rated at around 200 iso, so can i get a good idea running iso200 on a dSLR, how similar would it be in real life?


The manufacturer suggested rating doesn't necessarily have anything to do with the sensitivity or read noise performance of the camera. It's just a suggestion about how much headroom to leave available for highlights. The convention for DSLR cameras has been to meter so that there is 3.5 stops of headroom from middle gray to clipping. But that's just a suggestion. You can shoot -1 EC and give yourself 4.5 stops.

Manufacturers take plenty of leeway how they meter and display ISO. The Canon 5D and 400D, for example, have a true sensitivity of 87 ISO, using the definition of 3.5 stops from clipping. But 87 is not a pretty number to the marketing department, so they call it 100 and meter as if it were 100, which gives a 1/3 stop difference in metering between Canon DSLR and other DSLR. They fixed this the more recent cameras (450D, 40D, etc.) so that it will meter for 3.5 stops.

Digicams, on the other hand, have a convention of metering for just 2.5 stops headroom. That means ISO 100 on a digicam meters the same exposure as ISO 200 on a DSLR. That also explains why digicams blow highlights sooner.



Also it looks like 5000k lighting can make a real difference or is this just hearsay? (Even the EX works better under that kind of temperature but it's still perfectly usable under tungsten - i would expect the same from the red)


Yes, it makes a real difference. Tungsten light has four times as much red as it does blue. The blue light is literally just not there. On top of that, silicon sensors are less sensitive to blue.

3CCD cameras use 33% of their filters for blue light, but Bayer uses 25%. Therefore the difference between tungsten and daylight is even greater with RED than with your EX.

[paulcurtis]

Hello Daniel,

I thought i recognised your name, i think we had a discussion from another board many moons ago about 3 CCD vs Bayer. I hope you're well? As you may remember i've experiented and messed with various sensors directly but at the moment i have my work-orientated-real-world hat on. Although i would say that whilst QE is an indication of base sensitivity there are so many other contributing factors that that value alone can't really be used.

But in view of the fact that specifications are just that, i was hoping to find someone that's shot some practical side by side tests with Reds and other cameras in the same low light so i can get a 'feel' for what to expect as i am not in a position to test myself.

cheers
paul

[Daniel Browning]

Hello Daniel,

I thought i recognised your name, i think we had a discussion from another board many moons ago about 3 CCD vs Bayer. I hope you're well?


Yes, I remember you now. We did some e-mail, too. I'm doing great, thanks.



I was hoping to find someone that's shot some practical side by side tests with Reds and other cameras in the same low light so I can get a 'feel' for what to expect as I am not in a position to test myself.

I don't recall seeing any public tests that compare sensitivity and read noise in a way that I would consider useful. Hopefully there are some out there that you will find.

[Júlio Taubkin]

Hey Paul, another user on the boards, mcgregor did a very simple and not scientific side by side test between the two cameras:

http://www.reduser.net/forum/showthread.php?t=15645

He also did some Dynamic Range tests that maybe you can recreate at the lab with the EX-1 to get some comparisson material:

http://www.reduser.net/forum/showthread.php?t=17756

Hope this helps.

Cheers!

[Deanan]

I don't recall seeing any public tests that compare sensitivity and read noise in a way that I would consider useful. Hopefully there are some out there that you will find.

There are tons and tons of variables in a camera system that ultimately affect how you choose to rate a camera and evaluation criteria for noise is far from quantifiable. The only worthwhile thing to do from a practical standpoint is to shoot some tests for the type of project you're shooting and put it through your full pipeline.

[paulcurtis]

There are tons and tons of variables in a camera system that ultimately affect how you choose to rate a camera and evaluation criteria for noise is far from quantifiable. The only worthwhile thing to do from a practical standpoint is to shoot some tests for the type of project you're shooting and put it through your full pipeline.

Deanan, nice to hear from you - i hope all is good, it's been ages!

I suspect i'll bite the bullet and rent one for a day or two and find out for myself when i have a chance, i was hoping that perhaps people on here had been through the process in similar circumstances. The macgregor stuff is not really low light.

cheers
paul