Welcome to our community

Be a part of something great, join today!

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 :)

Stouffer of Mysterium-X...

Status
Not open for further replies.
Jim. All those boxes arriving at your doorstep are Red One's awaiting sensor upgrades. [Think the owls in the first Harry Potter movie].
 
Graeme Nattress said:
The measurement methodology is similar. The actual chart used is different. How the graph is plotted is different, and I think more accurately now. QUOTE]

because when I have tested DR along with other people.. Ive seen more in the 6-7 stops range with red one.
Click to expand...
 
Zakaree Sandberg said:
Graeme Nattress said:
The measurement methodology is similar. The actual chart used is different. How the graph is plotted is different, and I think more accurately now. QUOTE]

because when I have tested DR along with other people.. Ive seen more in the 6-7 stops range with red one.
Click to expand...

Maybe you need some help with testing? DR is defined as signal/noise. How were you testing that?

Jim
Click to expand...
 
Zakaree - testing dynamic range is a non-trivial task. It's darn easy to screw them up - I know. I've screwed up enough in my time. However, now I have a test platform that is controlled and repeatable. Anyone could go out there and shoot a Stouffer (make sure the room lights are off as any stray light will reduce your measured DR in the shadows), and plot the results.

What I have personally found - to measure DR you need accurate charts, controlled environment, patience, and a double check. With these graphs I also double checked the results with some charts that Deanan had shot, to ensure what I show on the graph is not a fluke.

Then we get to the gnarly and somewhat political issue of "the definition of DR", and where the only reply that makes sense is "your milage may vary". Not saying you're wrong, but I am saying that most likely you're not measuring the same thing I am. (and probably using a completely different method of measuring it too)

Graeme
 
Gavin - there are two improvements - one in absolute measured DR, and the other in the noise level at that DR. There's more range, and there's less noise in that range.

Graeme
 
Ok, we're just waiting...
But is a sort of Karma for all of us here.

Anyway, this time round, we can actually go out and shoot!

With "just" a Mysterium...
 
It seems to me that Arri is in trouble here. 2k vs. 5k. They have approx. the same DR and three times the price. Their system is not modular and they don't have Redcode. So why are they building these cameras. Seems like a development mistake anyway I look at it.

http://www.arridigital.com/teaser/sensitivity
 
Gavin Greenwalt said:
So if M is 11.3 and M-X is 12.25 isn't that less than a stop of improvement?
Click to expand...

If you measure the RED ONE to the same quality noise floor... it is 1 1/2 stops difference. I guess Graeme just said that?

Jim
 
Arri A-EV is apparently a 800+ EI equivalent camera system which records 1-60 fps and records in a ArriRaw format - uncompressed and in real time. The 800 ISO needs to be proven by 3rd parties to be believed. The data records to a S.Two drive, or media recorder. I am not Arri, but I believe the A-EV was designed to compete with F35 and Genesis directly and the Epic indirectly. Pricing is aggressive against the F35. Essentially, Arri's is taking shots across Red's and Sony's "bow" - if you want to use a nautical term. Uncompressed 3.5K should be an interesting storage issue...
 
I asked this earlier.....but no response yet.

"If we expose so we have 6 stops over and 6 stops under, what ISO would we have to set the camera to?
Assuming, as usual, that with increased ISO you lose some dynamic range."

I thought I was well behaved with this question by not asking about "native" ISO. :smile5:
 
Stephen Williams said:
Hi Evangelos,

Is that 12.2stops measured the same way as the 11.3 stops from Red One, just trying to work out how much improvement there has been.

Stephen
Click to expand...

Zakaree Sandberg said:
because when I have tested DR along with other people.. Ive seen more in the 6-7 stops range with red one.
Click to expand...

Yes, Stephen this graph is similar to what was the old one... the calibration that Graeme is referring to I suppose is actually density measurement of each step in order to be totally accurate.

As Jim says the DR is a function of noise versus density steps (DR stops), so a more useful info would be a chart that shows the noise deviation on each stop of DR, expressed in exposure stops... i.e. in the Stouffer step 30 which is 10 stops down in the shadows, what is the noise deviation expressed in stops? in the case of Mysterium sensor my measurements along with DoP Alfonso Parra measurements showed identical numbers (0,02stops difference) the 30 step was having a noise deviation of at least 1 stop, so yes it was barely visible to be included in a graph like the one is posted here, but the noise of 1 stop was so high that was almost unusable...

It seems that the 0.36 or about 1/3 of a stop is a noise level threshold that the DoP's is getting when empirically testing with multiply exposures the DR (or latitude) of a camera.

Comparing the graphs alone, we can expect a clear difference of 1 1/2 stop better performance, which I want to stress that this is like you had a Porsche that was running with 160 mph and now its running with 480mph...!

Thats quite big...

Zakaree, if you had 7 stops of DR of your taste, now you will have 8 1/2 of your taste...

Graeme Imatest...
 
Is "squeee!" an appropriate response? For a full-grown adult-type man-person?

Great work as always. Really looking forward to the cameras.
 
Evangelos - Imatest is nice enough, but just by altering the development curve it gives wildly different DR results, and hence in unreliable for my testing. DR of a sensor should be measured on linear light data, and I could never get Imatest to undo a totally accurate REC709 gamma curve back to linear light... Perhaps a problem at my end, but at least when I do all the math I know exactly what is going on and can answer questions.

Graeme
 
because when I have tested DR along with other people.. Ive seen more in the 6-7 stops range with red one.
Click to expand...

Keep in mind that you are going to have a very very hard time finding any type of display that can show more than 7 stops of range at a given time. You are never going to plug a camera into a monitor and see 12 stops. Or go to a theater and see 12 stops. Thats just the range that is on the negative (digital, or film), the whole trick with lighting is how do you work with the latitude of the capture medium, and the display medium, and get the image you want out of it.

Just my opinion, I'm in a room full of guys smarter than me ;-)

Nick
 
Interesting. So I presume the noise floor is more sudden now. I thought signal just got eventually overwhelmed by noise. So the m-x blacks out before the last 2/3 stop becomes apparent?

In other words the last half stop is less noisy than the last half stop on the REDOne? Can we not get at that last half stop if we want REDOne levels of noise?
 
From a non-technical perspective, I do like Evangelos's method on measuring noise as an x-stop variance. Could that not provide us with a testable, comparative numerical measurement of what is, to most of us, such a subjective thing? If the chart results could be accompanied by those kinds of numbers on noise we maybe could make some kind of objective appraisal of noise vs exposure. What do you think Graeme?
 
Shane Betts said:
From a non-technical perspective, I do like Evangelos's method on measuring noise as an x-stop variance. Could that not provide us with a testable, comparative numerical measurement of what is, to most of us, such a subjective thing? If the chart results could be accompanied by those kinds of numbers on noise we maybe could make some kind of objective appraisal of noise vs exposure. What do you think Graeme?
Click to expand...

I wish we had time to spend on everyone's wish of how to measure DR. The net result, which will be verified in short order, is that the M-X sensor is "bring a smile to your face" better than M. You can dissect that any way you want (in your spare time)... but the image speaks for itself. The charts and graphs on M don't seem to matter much anymore. Everyone knows exactly what kind of image it makes. The M-X is 1 1/2 stops better. Trust me... when you see the image, you won't need to check out the charts to verify that fact.

We take quick breaks in the action to deliver these tidbits, but Graeme's value is not in the classroom. He is at the core of finishing the work on fine-tuning this program.

Jim
 
Gavin, it's hard to describe in words exactly what I'm seeing. And If I tried to explain it, I could very well be wrong... There's no sharp change from noise-free to noise-floor. It's a continuum, but it starts later than in Mysterium, and doesn't get as noisy.

Graeme
 
Status
Not open for further replies.
Back
Top