Mysterium-X...

[Daniel Browning]

Of course I do understand that the practical benefit of 'even useful gain' for imagers is a lot less than e.g. in audio,

I'm sure you know this, but I'd like to add a reminder that the only time analog gain is useful is when the camera designers cannot figure out how to get the late stage read noise (e.g. ADC) lower than the early stage (e.g. photosite) in the first place. If they could achieve that ideal (and many have), then analog gain has no benefit (only downsides).

 

[Lauri Kettunen]

I'm sure you know this, but I'd like to add a reminder that the only time analog gain is useful is when the camera designers cannot figure out how to get the late stage read noise (e.g. ADC) lower than the early stage (e.g. photosite) in the first place. If they could achieve that ideal (and many have), then analog gain has no benefit (only downsides).

Daniel, well put!

 

[Karl H]

Sorry if this has been answered but I cant find it anywhere. Jim's new specs for mysterium-x quote frame sizes as 2:1 at 4K and 5K.

What is the maximum resolution you can shoot 16:9?

 

[David Mullen ASC]

Sorry if this has been answered but I cant find it anywhere. Jim's new specs for mysterium-x quote frame sizes as 2:1 at 4K and 5K.

What is the maximum resolution you can shoot 16:9?

If the 2:1 Mysterium-X sensor is 30x15mm and is 5K, then assuming that 16:9 uses the maximum possible sensor height, then it would be 15mm tall and therefore a 26.67mm wide area out of 30mm. That works out to be 4.4K if 30mm wide equals 5K. Unless my math is off...

 

[Jeff Kilgroe]

I'm coming up with 4550x2560 resolution for 16:9. However, they have not given us actual specs yet. Assuming full 5K is 5120 pixels wide, we get 5120x2560 for 5K 2:1, which is 27.648mm wide assuming the mentioned 5.4um pixel size. Additionally there should be 2.352mm split evenly on the horizontal of overhead or look-around room or 218 pixels on each side for this look around.

Or at least that is how the math works out. What we actually get could be something different.

 

[Karl H]

Thanks very much Jeff and David, interesting thoughts.

 

[joshua csehak]

You are correct. The key issue then is WHY you would you choose to re-rate to ISO 1600?

And the most likely answer to that, in the scenario you describe, would be to provide lots of highlight protection.

Now I'm even more confused. Daniel Browning said a couple pages ago that you get more headroom for your highlights at low ISOs. But he showed it with a regular digital camera. Does the RED sensor work differently? (Out of curiosity, does film work like that? (A bit of googling says it might, but it's hard to tell.))

ps- sorry if this is off-topic!

 

[Daniel Browning]

Now I'm even more confused. Daniel Browning said a couple pages ago that you get more headroom for your highlights at low ISOs. But he showed it with a regular digital camera. Does the RED sensor work differently?

Yes. Most digital cameras use four types of ISO settings (or "gain"): metadata gain, useful analog gain, worthless analog gain, and idiot gain.

RED ONE uses only 1 type of ISO setting: metadata gain. So as Eric Haase said, the ISO setting on the RED ONE doesn't change anything about what's recorded to the raw file, but changing the exposure does.

 

[Eric Haase]

Can someone from RED comment on how much more sensitive the MX sensor is over the M? Specifically referring to basic sensitivity, is it 1 stop faster? 2 stops faster? Surely this has been quantified and compared in the development of the sensor. Now that we have the Epic announcement with prices, etc., I am hoping we can get specific details about the most important part of the camera (IMO).

 

[Chuck Z]

Eric,

Check this out:

http://reduser.net/forum/showthread.php?t=36368

 

[Eric Haase]

Eric,

Check this out:

http://reduser.net/forum/showthread.php?t=36368

I've read this and see that MX has more dynamic range. Jim says great images at 800-2000 ISO.

But as far as basic sensitivity what is the measurable difference? That is, for example, if I need 400 footcandles to achieve 43 IRE (green false color) in RAW viewing mode shooting an 18% grey card with the lens at a T5.6 on the Mysterium, how many footcandles do I need to achieve 43 IRE (green false color) in RAW viewing mode shooting an 18% grey card with the lens at a T5.6 on the Mysterium X? An answer to this easily performed test would give us all an idea of the difference in sensitivity in the sensor and not how much you can pull out of underexposed image. It would also avoid subjectivity in statements about low noise images at high ISOs to prove sensor speed. I think it's important to understand the DR of the sensor and what the basic sensitivity is in order to produce the cleanest possible images and make decisions about how to light, expose, and post process a scene.

 

[Ryan E. Walters]

Are you talking about at ISO 320?

That would be interesting to know.

 

[Ryan E. Walters]

Never mind- I just read that a more closely- you said RAW mode. I'm still interested to know the answer to the test...

 

[Eric Haase]

Never mind- I just read that a more closely- you said RAW mode. I'm still interested to know the answer to the test...

The test would be interesting. It is important to note that part of RED's approach is developing color science to get great looking images out of somewhat underexposed images. It sounds like they are making a lot of progress in this area with the new "FLUT" color science, etc. I think it's good to know both basic sensor sensitivity and also what you can get away with as far as underexposing and digging out a clean image.

 

[joshua csehak]

Yes. Most digital cameras use four types of ISO settings (or "gain"): metadata gain, useful analog gain, worthless analog gain, and idiot gain.

RED ONE uses only 1 type of ISO setting: metadata gain. So as Eric Haase said, the ISO setting on the RED ONE doesn't change anything about what's recorded to the raw file, but changing the exposure does.

I get it! So on the Red, if you set the ISO to 500 (assuming 320 is optimal), you're basically just tricking yourself into underexposing by a stop then, right? I'm glad I leave it at 320. I was afraid I was missing out on something.

 

[Daniel Browning]

I get it! So on the Red, if you set the ISO to 500 (assuming 320 is optimal), you're basically just tricking yourself into underexposing by a stop then, right?

Yes.

A lot of DSLR shooters have the idea that high ISO settings cause noise. In fact it's the reduced exposure, not the ISO setting, that causes noise.

The reason that ISO can be difficult to understand with digital is that the linear response allows us to choose whatever tone curve we want in post, so any given exposure may or may not be ideal depending on how the raw conversion and post processing. One man's "underexposed and noisy" is another man's "perfect exposure with smooth highlight rolloff". Similarly, one man's "overexposed with highlights that clip too soon" is another man's "perfect exposure with no noise". It's always a balance between highlights and shadows.

 

[Deanan]

Yes.

A lot of DSLR shooters have the idea that high ISO settings cause noise. In fact it's the reduced exposure, not the ISO setting, that causes noise.

The reason that ISO can be difficult to understand with digital is that the linear response allows us to choose whatever tone curve we want in post, so any given exposure may or may not be ideal depending on how the raw conversion and post processing. One man's "underexposed and noisy" is another man's "perfect exposure with smooth highlight rolloff". Similarly, one man's "overexposed with highlights that clip too soon" is another man's "perfect exposure with no noise". It's always a balance between highlights and shadows.

Agreed. ISO/sensitivity does not equal analog gain although thats a common perception. ISO/ASA ratings are awkward for digital cameras and ideally would exposure compensation instead.

 

[Mark Pugh]

Agreed. ISO/sensitivity does not equal analog gain although thats a common perception. ISO/ASA ratings are awkward for digital cameras and ideally would exposure compensation instead.

This is why ISO should probably be re-named and taken out of the Sensor menu. It would help clear up misconceptions that are constantly recurring in this forum.

 

[Deanan]

This is why ISO should probably be re-named and taken out of the Sensor menu. It would help clear up misconceptions that are constantly recurring in this forum.

Taking it out would lead to more confusion.
ISO is still valid for how it's used in this case which is both for a visual representation of how you're exposing the sensor and a tie in to the metering.

 

[Roberto Lequeux]

How can it be differentiated then? Gain doesn't affect dynamic range aside from a probable higher noise floor at higher gain settings right? Gain does belong in a sensor menu. I wouldn't be confused if ISO were elsewhere, knowing that you can develop at a higher ISO is related to how you shoot on set, but it still not a "sensor setting" but more of a "negative property" or something.