Recent Info

Jannard said:
We also suggest including resolution and noise testing at the same time.
Jim
Click to expand...

Exactly. The balancing of needs - resolution vs. dynamic range vs. noise. As others have mentioned, these results are a "before" picture in the development cycle. As long as everyone understands that ; it can be a good basis for discussion. Certainly, there is a mountain of comparison to be done when the camera is nearing shipment.

I, for one, would rather see the balance done right to match the needs of a 4K digital cinema camera and not characterize the sensor for more dynamic range at the expense of more noise (for example). Although, I'm still trying to translate my film knowledge with these digital discussions (so forgive me if I'm getting some stuff wrong). Is there a whitepaper on hd camera testing somewhere ? Show me emulsion results and it's easy, but this stuff...

And Stephen, obviously, with a camera in the development stages, they're going to be highly choosey as to who, how and when they test and release data/footage. I would be amazed if they acted any other way. The fact we've seen "test" footage along the way is evidence of an open development process, but not a crazy-open one. ;)

That chart, for me, is just a pencil sketch. One among many sketches to come. Including dynamic range results, form factor, exterior shots, noise tests, low light response and most importantly - will it serve my aesthetic needs as an artist ?

:beer:

RED#'s cautiously optimisitic
 
evinsky said:
I don't know how tha 8 1/2 stop conclusion was arrived at but if youv'e seen any of the Red footage projected, it seems obvious to me there is more lattitude there than 8+ stops.
Click to expand...

That's probably my fault for choosing a bad title for this thread back in November: http://www.dvxuser.com/V6/showthread.php?t=78368&page=3

"8 1/2 stops" related to the range of the test, not some declaration about the range of the sensor. Here's me feeling foolish back then:

boothba said:
Just for clarity I'm not saying that Red has 8 1/2 stops of Exposure latitude. 8 1/2 stops simply represented the range of images I downloaded from CML. If they had posted the F22 series I could have gone further. I should have been more clear on that.
Click to expand...
 
Dynamic Range of Opinion

Dynamic Range of Opinion

If those of us who have been quaffing the Kool Aid are the highlights and the folks who simply will not believe anything Red claims are the deepest shadows I sense a latitiude of about 7 stops on this thread. It has always puzzled me that there is not a way to make a 15 stop digital camera by simply interpolating the sensor data over a wider range but I guess that just shows my naiveté as an engineer. In any case, I find Jim's assertion that they will be able to get more DR after they have more time to work the sensor out easy to believe. Over the years Sony has refined their DSP profiling to create significant improvements so if Graeme and the rest of the Red Team are as good as I think they are, then we just need to give them a little time.

Blair S. Paulsen
RedOne #19
 
You might notice the roll-off near the highlights of the others on the chart. That is not a natural event. It smacks of electronic knee control, which will apparently extend dynamic range. It makes sense because we know that the others have image processing prior to output. Our shot has no correction (knee control) or any other processing. Additionally, with some great work on our sensor settings by the team, we have been able to get a full 11 stops. If you really want to compare these cameras, the playing field needs to be level. We are all in on a head to head comparison... all things equal. And once everyone gets comfortable that the RED camera excels in this area, we can then concentrate on resolution, color and "feel" of the footage.

Jim
 
There is real, scientifically measurable dynamic range, and then there's subjective dynamic range.

We measure dynamic range as the difference (in stops) between white clip and when you can no longer determine any detail in the shadows.

Subjective dynamic range is often greater than measured dynamic range due to various image processing techniques, and your perceptual tolerance to noise.
 
I agree Jim.. it would of been great to see the Red footage with a DSP curve since its being compared to the other cameras with curves, instead of being the only Raw sequence on the chart.(you can tell by the linear line of the data plot. For some reason Dalsa isn't on there, which would of been a much more important comparison, raw to raw.
 
I wonder what the RGB footage would look like, curve wise ?

And do they test separate colors with these tests ? - three curves - R,G,B ?
 
I cannot be a naysayer but I'm not a yes_man either, however I fully agree with Jarred and, decidedly, with Jim too. Of course, 'cause his words of wisdom.
 
I don't get it. :huh:

It seems like a useless test unless the cameras & test procedures are matched, and they're all production versions. If that makes me a "yes" man than so be it.

I need a vacation.

RED#'s cautiously optimisitic
 
Exposure Chart Analysis

Exposure Chart Analysis

There are several questions regarding testing methodology raised by "the chart" .....but in any case if you want to record a wide latitude signal, the slope of the linear portion of the response curve - which is the area between the toe and shoulder in film or between the toe and knee in gamma corrected RGB video - should be relatively shallow.

So its interesting to measure the slopes of these curves between two points that should not be affected by any RGB domain knee or toe processing - such as 50 and 200 - and note their exposure values. (being RAW, the RED footage had none applied, but it is an option within REDCINE)

. . . RED D-20 F-950
50 -3.00 -0.50 -1.50
200 1.75 4.25 1.50

Diff 4.75 4.75 3.00 stops between upper and lower values.

The very similar slopes suggest that the latitude capabilities of this early prototype RED and the D-20 are similar, and both are better than the F-950 series camera as set up for it's test.

It also suggests that the equivalent ISO rating of the RED camera is incorrect (correcting it would move the RED response curve to the right) and its sensor gain is too high (correcting it would move the RED response curve down and reduce the gradient of the slope, which in turn would increase the stop difference between the 50 and 200 values above the estimated 4.75)

Since the exposure test we have reduced the analog to digital converter gain by approximately a factor of 2, which should increase the measured latitude by about a full stop and further reduce the sensor's already low noise floor.
 
So would that give the sensor an ISO of around 100? I know it's just speculation, but 200 ISO is what has been mentioned in the past.
 
evinsky said:
So would that give the sensor an ISO of around 100? I know it's just speculation, but 200 ISO is what has been mentioned in the past.
Click to expand...

Hi,

I would be quite happy to have a lower camera ASA rating if that meant more Dynamic Range. I find lower speed films tend to have a greater DR too.

Stephen
 
Stephen Williams said:
Hi,

I would be quite happy to have a lower camera ASA rating if that meant more Dynamic Range.
Stephen
Click to expand...

And more Lights.

Depending on how RED performs, 'pushing' the RAW footage in REDcine may be an option (don't know). But, for low key lighting, I don't want to rent three vans full of lights to achieve it.

IMHO, the best way to approach it (which I'm sure RED already knows) is not to try and hit some artificial benchmarks (by sacrificing more noise, for example). The technology should be optimized and characterized to achieve its 'native' ability. Let the chips fall where they may. Taking a gander at Graemes post; it seems that's exactly what they're doing...

Can't wait to see the design changes you guys have also done. :beer:


RED#'s cautiously optimistic
 
ISO is nonesense. Digital cameras don't have one. They just don't work like that. It's a fiction so film camera users can guess an exposure for a digital camera using a light meter. You can effectively think of the iso as whatever you want - it slides the under / over exposure lattitude around until you get it in a region you're happy with.

Graeme
 
I always thought that what digital cameras had a was a gain control for the A/Ds which, on DSLRs at least, is callibrated in ASA equivalents. Since the inherrent signal to noise ratio of the sensor does not change, I guess that what setting the gain does is affect the exposure at which the A/D clips to white. So this is similar to, but not quite the same as a film ASA.

Nick
 
Sure, it's calibrated in ISO, but as you have the ability to decide how noisy an image you can accept, you can also choose what ISO to rate the camera at. The camera doesn't have an ISO, but you can choose it to have whatever ISO you want, within the bounds of visual acceptance. That's what I was getting at.

Either way, by us working to lower noise levels, that means you can gain it up further in post without it looking as noisy, and that can be thought of as equating to a higher ISO, even though the sensor gain has gone down.
 
You must log in or register to reply here.
Back
Top