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 :)

13+ stops....

david farland said:
Maybe as you find it so boring Charles, you can enlighten us lesser mortals to the reasons why a fair latitude test between other cameras & Red is impossible. Just a line or 2 of meaningful facts will be fine.....
D
Click to expand...

David... we would love to have someone do that. Who, out there, is knowledgeable enough to do it correctly, getting the absolute most out of each camera, is totally impartial, and has access to every camera? While they are at it, shouldn't they do a resolution test at the same time?

We aren't the ones to do it... we don't know how to get the most out of other systems. And they probably don't want to get the most out of ours.

If you could give a two line solution to all of the above, we would be getting somewhere... but then again, the test would be obsolete as soon as new sensors hit the market. :-)

Jim
 
Hi,

Paging Mr David Stump ASC.

Stephen

Jannard said:
David... we would love to have someone do that. Who, out there, is knowledgeable enough to do it correctly, getting the absolute most out of each camera, is totally impartial, and has access to every camera? While they are at it, shouldn't they do a resolution test at the same time?

We aren't the ones to do it... we don't know how to get the most out of other systems. And they probably don't want to get the most out of ours.

If you could give a two line solution to all of the above, we would be getting somewhere... but then again, the test would be obsolete as soon as new sensors hit the market. :-)

Jim
Click to expand...
 
Jannard said:
We aren't the ones to do it... we don't know how to get the most out of other systems. And they probably don't want to get the most out of ours.
Click to expand...


I like the way this was worded...

They should be afraid to get the most out of RED's!

:biggrin:
 
I was at an interesting seminar in Toronto last night put on by ARRI/KODAK and E-FILM-DELUXE.

The focus was on Digital Intermediate, but there was definitely a soft sell going on with regards to the superior dynamic range of film and how that dynamic range can be maintained throughout the DI process.

What I found really interesting (and relevant to this thread) is that ARRI talked a little bit about a system that they are trying to implement with SMPTE for levelling the playing field when it comes to measuring "dynamic range".

It is called the "Opto-Electronic Conversion Function" - very German! :-)

Basically it is a system of back lit circles that are capable of displaying a full 16 stops of light. With a system like this, as long as it has been calibrated properly, all you need to do is point the camera at it and read the measurements. No light meter error, imprecise stop setting or light contamination.

Seems like a very simple/fair and hard to F*&! with system.

Anyone else know more about this system?
 
Dylan Macleod said:
Anyone else know more about this system?
Click to expand...

Doesn't sound much different than a calibrated Stouffer step wedge, if it is lit perfectly even.

There are small factors that can throw off the results slightly, too, and should be considered in the testing. There is optical falloff from the lens itself, falloff due to cosine^4, PRNU, dust on the sensor, Callier effect, any other source of flare, uncalibrated shutter speeds, aperture activation variance, and then a host of issues in the software side. Many of the testing hardware issues can be resolved by subtracting only the differences between two frames for measurement.
 
DxO Labs noise test

DxO Labs noise test

Dylan Macleod said:
I
Basically it is a system of back lit circles that are capable of displaying a full 16 stops of light. With a system like this, as long as it has been calibrated properly, all you need to do is point the camera at it and read the measurements. No light meter error, imprecise stop setting or light contamination.
Click to expand...

DxO Labs have just come out with a system for comparing still cameras on a qualitative basis -- they use a similar system for their noise measurement:

http://www.dxomark.com/index.php/eng/Technologies/Measurement-definitions/Test-Protocol

> The DxO Noise chart is a transmission chart consisting of a set of circular holes equipped with neutral density filters ranging from D=0 to D=4. The density filters are positioned on a circle and the chart itself is placed against a light table delivering a uniform lighting across the whole chart. When shooting such a chart with the camera being tested, the sensor sees a wide range of light levels with a 1/10,000 ratio from minimum to maximum.
 
Alan Fletcher said:
DxO Labs have just come out with a system for comparing still cameras on a qualitative basis -- they use a similar system for their noise measurement:

http://www.dxomark.com/index.php/eng/Technologies/Measurement-definitions/Test-Protocol

> The DxO Noise chart is a transmission chart consisting of a set of circular holes equipped with neutral density filters ranging from D=0 to D=4. The density filters are positioned on a circle and the chart itself is placed against a light table delivering a uniform lighting across the whole chart. When shooting such a chart with the camera being tested, the sensor sees a wide range of light levels with a 1/10,000 ratio from minimum to maximum.
Click to expand...

This is exactly what the Opto-Electronic Conversion Function looked like.

Unlike the Stouffer test wedge, the points of light are isolated so they do not contaminate their neighbour.
 
Daniel Browning said:
Doesn't sound much different than a calibrated Stouffer step wedge, if it is lit perfectly even.

There are small factors that can throw off the results slightly, too, and should be considered in the testing. There is optical falloff from the lens itself, falloff due to cosine^4, PRNU, dust on the sensor, Callier effect, any other source of flare, uncalibrated shutter speeds, aperture activation variance, and then a host of issues in the software side. Many of the testing hardware issues can be resolved by subtracting only the differences between two frames for measurement.
Click to expand...

Is the Stouffer capable of displaying 16 stops?
 
LilRed said:
We just worked with David Stump a couple of months ago.. fine person. (Had him sign my hat). Didn't he test the prototype Red One one-upon-a-time? :)

~Sarah
Click to expand...

Yes it was David,
 
Okay…..my idiot take on it…(and I mean that!)
Why can’t we do a fair latitude test with Red & others camera?
I thought the Stouffer 41 step wedge test could be used to comparatively measure latitude within the ‘luminance’ domain. If the noise gets unbearable in the low range you stop. If you stop measuring any grey in the high end you stop. However I understand this to be a linear line and it’s possible to apply a slight S curve to this. I guess adding a curve will result in increased stops in the high & low ends at the expense of the mid tones where the s curve (ev. vs stop) changes more rapidly. You get the min/max +/- Ev range but actually count the number of Stouffer wedges and I’m guessing the count will be down.
Now in regards to the individual chrominance channels, the results will be similar. A noisy blue channel or clipping red channel and the latitude of the channel will be reduced. If you apply a curve to one of the colour channels to ‘rectify’ this or increase saturation, the image will suffer reduced overall latitude or banding (I guess).
For lots of 3rd party cameras this fine tuning is done in the camera, (presets designed by experts) but Red allows ham-fisted fools like me to have a go on it and I guess if comparative testing and post is done by people who have a poor understanding of the Red in this regards, the results can be unflattering for Red.
The point of all this is to try and identify the common mistakes made by Red users and raises the question of what test elements should be studied to provide indicators of the strengths or weaknesses in a camera performing in real life.

Dave,
 
Jannard said:
Who, out there, is knowledgeable enough to do it correctly, getting the absolute most out of each camera, is totally impartial, and has access to every camera? While they are at it, shouldn't they do a resolution test at the same time?

Jim
Click to expand...
There was a very thorough and reasonably expensive series of tests done by a large Post house in Sydney this year, comparing 35mm film, Genesis, Red, F23, and others.

The tested were screened over several weeks and viewed by many in the industry.
All the Red footage looked bluish in comparison to all other formats. Sunrise exteriors, etc...
It looked really unattractive compared to other cameras.

"We found that the Red Camera footage comes out cold"
I thought "WTF?"

It left me completely unsure about their ability to pull the best out of the footage in many ways, including latitude, which at that stage could be seriously degraded in transcoding by not following certain steps.

I left quietly P$#%ed, and wondering if they had an agenda to push the telecine suite that they'd invested so much money in. Either that, or they were clearly not up to speed with processing Red footage.
They kept stressing that they had no agenda except science...

It's very hard to find someone who doesn't have an agenda.
 
If you are going to look at a RED, process the file out to a 16bit Tiff in RedAlert. Set ISO to 320, turn off matrix. That will show you what was recorded without any image processing clipping at either end. You can set the gamma to either REC709, REDLOG or Linear Light, as they're all full range and don't induce clipping. Exposure compensation should be set to 0 so as to not induce any clipping.

Graeme
 
Graeme Nattress said:
If you are going to look at a RED, process the file out to a 16bit Tiff in RedAlert. Set ISO to 320, turn off matrix. That will show you what was recorded without any image processing clipping at either end. You can set the gamma to either REC709, REDLOG or Linear Light, as they're all full range and don't induce clipping. Exposure compensation should be set to 0 so as to not induce any clipping.

Graeme
Click to expand...

There needs to be a huge button in the RedAlert interface that makes all of that happen, and that should be the default setup, even before metadata, to ensure that footage is transcoded at it's full potential.
 
Make sure you use Build 17 or greater...

Jim
 
Mark Pugh said:
There needs to be a huge button in the RedAlert interface that makes all of that happen, and that should be the default setup, even before metadata, to ensure that footage is transcoded at it's full potential.
Click to expand...

Mark! Where's your movie gone bro? I'm still waiting for the release.
 
Ancient history...

Ancient history...

In this thread http://reduser.net/forum/showthread.php?t=10362 there was a very good example from Jbeale on how someone can analyze a step chart...

The original file that I posted is not on my FTP any more, I can put it back if someone asked me... but it was shot with build 14 I think (or 13?)... so an up dated test with someone that will use a Build 16 or 17, with the RED alert settings that Greame is stating above will show easily everything...

It just needs a good work on how you will evenly light the step chart also it needs a good work in the stray light and its a perfect solution...

If someone follow the original Cinematography.com thread it will see the setup I used...

A Danes Picta TS28 was used in this test with 14 stops, Stuffer has 13 2/3 range...
 
Rubber Monkey - love them.

Rubber Monkey - love them.

Nigel Stanford said:
Mark! Where's your movie gone bro? I'm still waiting for the release.
Click to expand...

It's done a bunch of Film Festivals - Toronto Film Festival, Melbourne, here and there.
US Distributor (Darklight) seems to be sitting on it, but I'm not sure ...

Folks this is something I shot last year... "Acolytes" - in Australia... Rubber Monkey supplied the camera - a Viper that's probably off in a museum somewhere.

Nigel and his boys did an incredible job, flying in from New Zealand at ridiculously short notice, DITing and helping the editors beyond the call. Day and Night, writing code to keep things working better... a really memorable effort, full of genius and passion and fun.

Pictures looked good, though, and audiences seem to like it.
It won a cinematography award last month - all thanks to you...
 
You must log in or register to reply here.
Back
Top