Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Mysterium's dynamic range

D

David J. Cain

Member
Joined
Jul 26, 2009
Messages
5
Reaction score
0
Points
0
Location
Toronto
Website
www.davidcain.ca
Hi all,

Newbie to Reduser and having a little trouble finding any earlier posts related to the dynamic range of the Red One's sensor. Could someone help out?

Much obliged,

David
 
most people agree between 8-10 stops depending on conditions.
 
Thanks Brandon. I've read as much. I'm curious though to see if there is/was a full-blown discussion about it - including what some have done in the way of tests, gray scales, waveforms, etc.
 
It's complicated because there's a difference between total measurable dynamic range of the sensor and working or usable dynamic range -- a few stops of detail are buried at the bottom in the noise floor so generally fall into black once you time the image. But then, the same is true for color negative, the theoretical range is somewhat wider than the practical one.
 
Thanks David. I'm looking to determine the practical range. If I may...

I've got a series of spec spots shooting next weekend. We're using the Red and I was going to meter the shoot as opposed to having a waveform on set. I need to determine the camera's dynamic range, so I thought I might set up a gray scale in one of the camera bays, shoot it and compare spot meter results with a waveform. Is that a reasonable approach?

Sorry for the DR 101 question, but I'm appreciative of any knowledge you can pass on.
 
Also since Mysterium is a daylight sensitive sensor you'll lose some effective DR due to white balancing depending on the desired white point in your scene. Although Build 20 is supposed to reduce this loss through some clever software.
 
We're shooting mostly exteriors; one interior that will be keyed using a daylight fixture through a picture window. So it's my intention to stay with the camera's native colour temp (Build 17) as well as rating it 320.

I'm running the test tomorrow using a small HMI to the light chart. I guess I'll get a pretty good sense of things then. Thanks for the feedback all.

David
 
David, my recommendation is to use your meter for ratio and the camera's histogram for stops. A waveform won't give you the whole picture because it's not a video camera - it records RAW which is way more than a video signal can actually contain. The histogram is your friend. Basic theory is to expose to the right. If you search the forums for "ETTR" you should find some great information (so long as the search function is working today:laugh:) You should be looking for a histogram with whites over near the right, but with a little gap for safety and rolling off in post, and try not to paint yourself into a corner by letting stuff fall into shadows if you might want to bring them back up again. That's a recipe for noise. Just remember when you're setting your ratio that you need slightly tighter ratios than neg due to the tighter DR. So where you'd normally (neg) have a stop, you'd maybe set it to 2/3, etc. Aim to come back with a "thick" neg and grade it in post. Don't treat it like a video camera and try to lock it off in the camera. If nothing else, you'll just waste the cameras potential.
 
How come it is so small? I suspected they would be moving towards 16 stops with new techniques and new sensors.
 
Wayne Morellini said:
How come it is so small? I suspected they would be moving towards 16 stops with new techniques and new sensors.
Click to expand...

Well we aren't sure about the full capabilities of the new Sensors for Scarlet and Epic, BUT with the increase in bit depth and resolution I am sure that we'll see a boost in dynamic range, how much I don't know, but I imagine it'll be significant. (At least on the higher end).

As far as 8-11 being small... Film on its best day only has about 13-15 max right? (I think new Vision3 tops out at around 14 or so I think.)

More realistically you see far less dynamic range than that in most cases, especially with the degradation of the prints during each step in the process.
 
Gavin Greenwalt said:
Woooeee! 16? Now THAT would be quite the sensor. I don't think anyone has hit 16 without using ND sensor sites yet have they?
Click to expand...

There has been sensor technology for years that does 27 stops while keeping color constancy, and ultra-low light. There is another sensor company with cheap technology that does 120db of s/n (which I imagine you could squeeze 20 stops of latitude into). But they wont supply me to develop my own sub $1000 3D pro video camera. Samsung has bought into it, and maybe have bought the company by now.

For years there has been multi-slope and variable gain high latitude techniques, usually around 16 stops, on very cheap sensors. These older techniques however, do not produce the best results, and color constancy suffers.

Film latitude is not necessarily ideal either, but the 8 stops is disappointing. I have been working towards an high latitude post exposure editing process for many years. The sort of systems I have looked at for a long time, was to enable much smaller crews, moving much to post, this would allow smaller budget products to increase quality. The studio systems I have been contemplating are also aimed to greatly reduce the man power needed, and patentable.

Not all the technology in the Red is the latest or most advanced. I would be curious about its fill factor too, now closer to 100% can be had by back illuminated techniques, which would make it easier to include higher latitude and global shutter. However I am happy with what Red has coming, but latitude is one thing I would like more of (I have techniques in mind for that 16 stops anyway).
 
Last edited:
Don't forget that currently raw files are encoded as 12-bit linear, so impose an absolute 12-stop dynamic range anyway, regardless of what the sensor can capture (which is significantly less than 12 stops, as previously pointed out).
 
Gavin Greenwalt said:
Also since Mysterium is a daylight sensitive sensor you'll lose some effective DR due to white balancing depending on the desired white point in your scene. Although Build 20 is supposed to reduce this loss through some clever software.
Click to expand...

That should be "Mysterium, like ALL silicon based sensors, are naturally more daylight balanced than tungsten". That's because silicon doesn't respond too well to blue light and tungsten is blue deficient light.

You don't loose any DR due to white balancing. If your white balance causes the data to clip, you can bring the exposure back until it does not. No clipping, no loss. If the highlight color is funky, you can correct that quickly and easily with the DRX tool. B20 does indeed make for better looking, less noisy images though. That indeed is true.

General comment on DR:

Discussions over DR are always interesting, seldom boring, but... everyone has their own definition of DR, what it is, how to measure it, etc. It's not the hardest thing to measure, not the easiest either. Also, what it comes down to is a personal noise tolerance, because that's exactly what DR is - the difference between the brightest recordable light, and the darkest detail before it gets too noisy for you. To suggest that their is any consensus on a practical DR is, I think, stretching things too far.

The Mysterium sensor uses 12bit analogue to digital converters, so the practical and theoretical limit is somewhere between 11 and less than 12 stops.

Graeme
 
You must log in or register to reply here.
Back
Top