ASA Rating?

[Patrick]

Considering the use of Cine/SLR lenses is there an ASA/ISO rating for the camera? Someday I may invest in a hand held waveform monitor, but nothing beats the good old lightmeter.

[David Mullen ASC]

The DP told me that he was advised to consider the camera as being in the 250 to 320 ASA range, so he chose 250 ASA.

Now remember that this camera puts out a RAW signal right off the Bayer-filtered sensor. Giving it different ASA ratings really means processing the signal in order to boost the gain, like a video camera does.

With cameras being recorded in RAW mode, you don't boost the signal being recorded, you can only boost it later in processing or color-correction. So the 250-to-320 ASA rating is basically to give it a low noise floor, but you could rate the camera at 500 or 1000 ASA if you chose to, record the signal, and then in post, brighten it to look normal -- but at the cost of increased noise. However, if the camera is fairly low in noise to begin with, often a boost like that is not objectionable.

I saw a demo of some 4K Dalsa Origin material shot at night (and this camera is about the same sensitivity as the RED camera it seems) in a mix of real streetlamps and a few tiny movie lamps (like a Tweenie). So the effective ASA was probably in the 800 to 1000 ASA range once the picture was brought up to look normal in brightness. And the noise was hardly objectionable, though there were no extreme corrections to the color balance that could have made the noise pop in certain colors, which is the danger of working with a low signal. Sometimes the noise looks fine when you boost the image from 250 ASA to 1000 ASA, let's say -- but as soon as you try to make it more blue or more red, the noise jumps out at you. So you always have more flexibility in color-correction with a properly exposed image.

ASA ratings for digital cameras are somewhat flexible -- it all depends on the noise level you can live with or find acceptable. It is not unusual to use such a camera at 250 ASA outdoors but at 500 ASA indoors and not really see much of a difference in noise level.

Once you test the camera to learn its effective exposure range, you could certainly just use a light meter like with film and just keep an eye on the extreme ends of the exposure range. A waveform is more useful in telling you if large amounts of information are close to being crushed or clipped so you can decide whether to expose a little more up or down to hold more information. But for general photography, you tend to base your exposure on the subject itself, often a human face, and how you want it to look in terms of luminence.

[Graeme Nattress]

Great answer David - you're really grokking this RAW stuff.

Graeme

[T. Singh]

With cameras being recorded in RAW mode, you don't boost the signal being recorded, you can only boost it later in processing or color-correction.



Don't know how RED does it, but most DSLRs do boost their signals depending on the ISO setting even in the RAW mode. They use an analog amplifier before digitizing the signal. This keeps the quantization noise down.

Regards,
Tripurari

[Graeme Nattress]

Yup, you can do both digital and analogue gain. And then you can also do funky stuff in code with the RAW data. There's lots of possibilities.

Graeme

[Brook Willard]

I was able to configure the EI of the prototypes to 1,000+, though I never really checked to see how it looked. :laugh:

[Alexander Nikishin]

I'm hoping for 800 at the minimum.

[Stephen Williams]

I'm hoping for 800 at the minimum.

Hi,

You must like ND filters!

Stephen

[Alexander Nikishin]

For low light situations ofcourse. :clown2:

I'm sure that with the Mysterium sensor being as clean as it is, it can hold up cleanly at 800.

[casey warren]

How will the REDONE's ASA compare to a Canon 5D or Mark II?

100 ASA on the 5D is different from the 100 on the Mark II, rather odd. I found that the 5D noise level is much more muted.

[Yuval Shrem]

At the show the approximate ASA ratings I heard were 200, 250, and 320 ASA. Seems like 200 ASA is its true native measurement of the sensitivity of the sensor, while you can push it to 320 without creating too much sensor noise. Please do correct me if I'm wrong.

However, to my questions about using RED-ONE in low-light conditions, the answer Ive received was that using gain (which was not yet enabled in the cameras used for the demo-shoot) will make it possible to use it even for very low light situations. They also mentioned that they will try to optimize the compression codec to avoid compressing the added noise created by using gain. Graeme said he hadn't yet had the opportunity to fully test that, but judging by his supremely unbelievably amazing work on RED-CODE-RAW so far, I have full confidence in him to deliver.

Bare in mind that considering the astonishingly clean and noiseless images RED-ONE is currently generating, the analog noise of 6 or 12 db of gain will still not bring it to an unusable state, but will probably seem like the difference in grain levels between 200 ASA and 500 ASA film stocks. However, if you're really hoping for "at least" 800 ASA of native sensitivity you are simply being unrealistic.

[Alexander Nikishin]

Not native, just a minimum gain level to obtain 800 asa.

[Matthew Verkler]

Bare in mind that considering the astonishingly clean and noiseless images RED-ONE is currently generating, the analog noise of 6 or 12 db of gain will still not bring it to an unusable state, but will probably seem like the difference in grain levels between 200 ASA and 500 ASA film stocks. However, if you're really hoping for "at least" 800 ASA of native sensitivity you are simply being unrealistic.

It's all about signal to noise ratio, and what is appropriate for a given project. If you want that gritty look anyway, gaining up (whether in camera or in post with the raw files or both) won't be a problem for you. Look at Minority Report or We Were Soldiers. As I understand, they were both shot on 800 ASA film for all scenes, DAY AND NIGHT, precisely to get more grit. And Minority Report had all that effects work, and they still shot 800 for the look.

Jim may not like the sound of this as he is protective, hehe, but I personally want the ability to gain up in camera WAY beyond what might be considered "acceptable" by some. Then I can make the choice for a given project BASED ON TESTING. For some projects, it might make sense to shoot 4K at a very high gain, and then bring the final down to 2K to reduce the grain to the level desired. Downsampling smooths grain right out, especially if you add a slight blur to the 4K file before downsampling, which acts as a low pass filter and decreases aliasing. We watch plenty of films where the DI was at 2K, so for some projects, the increased sensitivity for shooting is worth less res in final display.

RED gives us a very clean 4K image as a basepoint. That means that we can push the image much further in any direction we choose versus other digital offerings and film itself. I tend to do most of the pushing in post, where I can recover from my "experiments", but if I know where I'm going and if it will get me even further, I'd love more adjustment range in camera as well.

[Kevin Halverson]

The concept of using an ASA rating for a digital based system is fairly subjective. Since the gain before the quantizers, the quantizers themselves and the post quantizer processing will all have an effect on the resultant value assigned to any given quantity of light reaching the sensor array. While I certainly understand the desire to have a number to work with that allows for the initial exposure setting, I would think that a better approach would be to define the performance in terms of S/N ratio for a given gain (both analog and digital) with a known illumination level. This would allow for a more translatable "rating" to other types of systems.

[Graeme Nattress]

We have not yet defined any "true" ASA of the sensor - all this is relative and in flux.

Graeme