WARNING: RED has -3 stops less latitude than the D-21&F35

[Tim Sutherland]

I think an important consideration in ETTR is to make the decision by scene, which most likely means that you TEST the camera before shooting it, just like you would if you've never shot Kodak Vision 3 before.

The reason I think it's important per scene, or maybe per dynamic range envirnonment, is that you can sometimes be opening up the iris 1/3 of a stop to fill the histogram and then close it back down for the reverse, if you are matching shots. Now you have DOF changes and other considerations.

I think this is a decent hybrid between the 'old' ISO based system, and the histogram reading situation, because you can use the histogram in your tests, and say, for this 2pm daylight exterior, we are going to rate the camera at 500ISO and fill in the shadows a bit, in order to keep the skys blue. That way you have a base exposure setting to help with continuity of lensing, etc.

I think that's one of the big differences between using a histogram for stills and a histogram for motion. In motion, a lot of times you make aesthetic choices with the lens, shutter speed, etc, and then you work the lighting/DR (as much as you can control it) around that (although, you can't just pluck numbers out of the air, it requires testing). This allows you to have that continuity in a scene, especially matching elements, that are sometimes (often?) not necessary when exposing for digital stills.

Hopefully my point is clear here.

 

[Dylan Macleod CSC]

Given that each camera manufacturer has a different post and monitoring workflow, and acceptable noise is a matter of personal taste...

I'd suggest that "hands on" camera tests generating inconsistent results - and follow on debate and additional testing - are here to stay.

I am hoping that ARRI's new Dynamic Range Test Chart is going to help level the playing field when it comes to camera to camera comparison of dynamic range.

http://www.arricsc.com/pdf/2009_09_ARRINEWS_EVF_DRTC.pdf

I have been in contact with ARRI so that a side by side analysis can be performed with many of the camera systems mentioned here - and adding in Kodak and Fuji stocks.

I know that there are systems with more dynamic range than the RED. That didn't stop me from buying one.

 

[Hans von Sonntag]

Exposing Red with an ASA less than 320 can have very unfortunate consequences.

Frankly, I think that throwing around numbers like 160 could lead people to make terrible mistakes.

Opinions vary, but exposure never lies...

Good shooting and best regards,

Leo

Leo, developing at ISO 160 or changing metadata in camera to 160 and using this for exposure does indeed no good to the shots. But taking a lightmeter setting it to 160 ASA will bring you near to the acclaimed 320 Metadata ASA Graeme is referring to.

Basically there are two user groups with Red. The ex film people who like to rely on stock, ASA indices and lightmeter and the video folks who love to see what they are doing. Metadata ASA and "legacy" lightmeter ASA don't match, IMO. This confuses people. In doubt follow Red's advises. But then don't use a lightmeter.

Hans

 

[Axel Mertes]

I agree with Graeme.

Exposure of a scene can be controlled without even thinking about an ISO setting nor actually selecting one. In fact I can do the same perfect exposure with a RED ONE at ISO 2000 as I can do at ISO 160. Because its RAW it is always the very same data. The only thing that changes is that my preview will look either too bright, too dark, or somehow right when going to e.g. ISO 320.

The ISO setting is a preset for interpreting the data, either for live monitoring or later quick review. Remember, you can always override or alter the ISO settings on any R3D file.

Its nice if a camera can catch a really wide dynamic range. But a DOP / AC should be able to adjust the camera and lit the scene to a point where the dynamic range almost fits the scene. IMHO its their job. The less they care about this, the more latitude is required to compensate.

If one day the perfect sensor would be there, that would catch every single photon for sure, then we might never more discuss to light a scene. We would just point and shoot. DOPs jobs would be down to framing "somehow". Well, I clearly not expect such a day to come in my lifetime.

One more thing:

The cameras that have been "compared" here have drastically lower resolution than the RED ONE. If you downscale the RED ONE image to the same resolution as they have by nature, you immediately gain dynamic range. That is simple math.

With all respect to the poeple involved, as long as I have not seen proof that this has been taken into account in the comparison, the whole thing is just wrong IMHO.
I do not say that the RED is better or close to F35 etc. I did such tests myself and know there are differences. And the F35 low light performance is incredible - no question. But the values stated have to be carefully examined.

At which resolution has each camera been tested exactly?
IMHO only a 1920x1080y scaledown of all of them would be a valid comparison. Clearly not for sharpness, but for dynamic range.

I can use my good old low dynamic range Hi8, expose just to its limits, downscale that to 24*18 pixels and will for sure have the widest dynamic range of them all... pristine!

In stop motion - were exposure time isn't an issue - we use so called "frame averaging" to increase the quality of the exposure. Noise goes down drastically. You may shoot 100 frames with a Hi8 body and the resulting image with frame averaging looks far better then the most expensive camera would do in a "live shot".

As Greame pointed out, dynamic range is a function of signal to noise ratio, the higher the resolution is, the higher the comparable noise will be.

Apples to apples, pies to pies.

Axel

 

[BradWright]

Brad - I just did my experiment here under 2800k Tungsten, and didn't see unnacceptable noise going with the light meter's recommendations.

Graeme

I'm just wondering is there a standard test for measuring noise levels by comparing changes between frames on a test chart. I wrote an App that I never finished which compared stuff like this. This seems like the missing link in the scientific analysis of measuring noise floors on locked down shots. I really should finish that App.

 

[Stephen Williams]

Hi Sanjin,

So this is the second time you have been anti G.B.! IIRC the first time was in the build 8 days.

Geoff has shot an awful lot with RED one, because he understands the cameras limitations he is able to get the most out of it.

Stephen

Geoff Boyle FBKS, cinematographer and leader of

CML - Cinematography Information & Discussion

did extended digital camera testings at the

Hands on HD - the annual high definition-event in Hannover.

"The RED has just under 3 stops less latitude than the D-21 or the F35 and
about 1 stop less than the SI-2K. However, it can look gorgeous if you bear
this in mind, it definitely favours overexposure and protecting the
highlights rather than exposing for the low end."

LINK>>>



I write now and would consider this suggestion as a sort of Warning Message to RED Team:



"Please RED do not start ship EPIC, EPIC-X or RED ONE upgrade to Mysterium-X

before you don't FIX EXPOSURE LATITUDE of the new sensor (aka Mysteruim-X)."

This would help us not to go again throughout the same circle of discussions, confusion and testing

about Mysterium-X comparison with the other digital cameras like Arri D-21, Sony F35, SI-2k, Phantom HD Gold, etc,....

And the most important thing is to make RED1 M-X, EPIC-X, EPIC, etc,...

a real competitive camera system with the other digital acquisition systems.

Simply just make it better than any other sensor at the competition."

 

[Steve G]

I agree with Graeme.
Its nice if a camera can catch a really wide dynamic range. But a DOP / AC should be able to adjust the camera and lit the scene to a point where the dynamic range almost fits the scene. IMHO its their job. The less they care about this, the more latitude is required to compensate.

I think that all dop's would say that of course they would love to be able to light to perfection everything that they shoot. Sadly, budget and time pressures mean that we don't always have the exact lighting tools or resources to be able to do that. I shot a commercial recently which could have been done on the red, but we were doing a night shoot and we only had vehicle access and budget for one small cherry picker, which then meant that we had to limit the size of the main "moon light" to an 18k to cover a very large field. I knew that by the time I softened that down to get a nice spread, I would be dead if I used a red, so I went for the P2 Varicam 3700, which is a fantastic low light Hi Def camera.

The cameras that have been "compared" here have drastically lower resolution than the RED ONE. If you downscale the RED ONE image to the same resolution as they have by nature, you immediately gain dynamic range. That is simple math.

So, given that more and more stuff that we shoot is being watched on the internet or on tv screens, as opposed to cinema screens, do we really need 4 k resolution ? I don't know the answer to that one. (Certainly, the last thing I shot on Red was posted at 2k) Maybe arri, with their new 2 k camera, feel that the "good enough" scenario comes into play in terms of resolution.

 

[BradWright]

I agree with Graeme.


One more thing:

The cameras that have been "compared" here have drastically lower resolution than the RED ONE. If you downscale the RED ONE image to the same resolution as they have by nature, you immediately gain dynamic range. That is simple math.

With all respect to the poeple involved, as long as I have not seen proof that this has been taken into account in the comparison, the whole thing is just wrong IMHO.
I do not say that the RED is better or close to F35 etc. I did such tests myself and know there are differences. And the F35 low light performance is incredible - no question. But the values stated have to be carefully examined.

At which resolution has each camera been tested exactly?
IMHO only a 1920x1080y scaledown of all of them would be a valid comparison. Clearly not for sharpness, but for dynamic range.

Axel

Excellent point! 4 to 1 down sampling to 1080p is a significant way to reduce noise if done correctly. This would explain some of the huge gap in reported DR.

One of the problems I noticed when shooting the Red is that the 720p output for monitoring is a significant detractor when compared to monitoring at 1080p. There are digital artifacts that occur in the down sampled 720p output that aren't present in the final image. The impression you get from the 720p output is that everything is very soft and a bit nosier than the high res R3D.

If you are shooting side by side and the F35 is feeding a 1080p and the Red is producing a 720p image, you are going to be biased towards the F35.

 

[CsabaJuhasz]

If so, why such experienced cinematographer, Geoff Boyle doesn't take it into consideration and do this test such a way?

 

[Eki Halkka]

If so, why such experienced cinematographer, Geoff Boyle doesn't take it into consideration and do this test such a way?

That's a good question... but funnily enough, most camera shootout tests out there have done something wrong with the red footage, so i wouldn't be surprised if...

 

[Stuart English]

HD-SDI outputs

HD-SDI outputs

To be noted is the 1280 x 720p PVW output from RED is intended as a HD video assist, it's not Program video.

And as we do NOT add any sharpness (edge enhancement) in the camera, yes it will look a touch soft v's other cameras.

Also the noise floor and color purity of the HD-SDI outputs are significantly better since Build 20 firmware.

 

[Hans von Sonntag]

With all respect to the poeple involved, as long as I have not seen proof that this has been taken into account in the comparison, the whole thing is just wrong IMHO.
I do not say that the RED is better or close to F35 etc. I did such tests myself and know there are differences. And the F35 low light performance is incredible - no question. But the values stated have to be carefully examined.

At which resolution has each camera been tested exactly?
IMHO only a 1920x1080y scaledown of all of them would be a valid comparison. Clearly not for sharpness, but for dynamic range.

I can use my good old low dynamic range Hi8, expose just to its limits, downscale that to 24*18 pixels and will for sure have the widest dynamic range of them all... pristine!

In stop motion - were exposure time isn't an issue - we use so called "frame averaging" to increase the quality of the exposure. Noise goes down drastically. You may shoot 100 frames with a Hi8 body and the resulting image with frame averaging looks far better then the most expensive camera would do in a "live shot".

As Greame pointed out, dynamic range is a function of signal to noise ratio, the higher the resolution is, the higher the comparable noise will be.

Apples to apples, pies to pies.

Axel

Hi Axel,

isn't the RedOne effectively a 1080p/2K camera? How many 4K cameras are out there... One reason why the RedOne holds up so well is the power of oversampling, IMO. Here I heartly agree. But since the comparison standard is 1080p and oversampling is part of the RedOne's technology we compare oranges, to oranges, IMHO

Hans

 

[Sanjin Jukic]

Hi Sanjin,

So this is the second time you have been anti G.B.! IIRC the first time was in the build 8 days.

Geoff has shot an awful lot with RED one, because he understands the cameras limitations he is able to get the most out of it.

Stephen

Hi Stephen,

there is no need to "advocate" G.B. from you here because whatever he and his collaborators

did with those tests there are going to be published and available to the public soon at CML pages.


The main idea behind that sort of warning was a reason that RED should not RUSH with shipping Epic-X

before Mysterium-X is not totally ready for the (bright) future in a comparison to the competition.



Take these two possible "scenarios" to count:



Scenario-1:

Next year at October 2010 somebody doing a comparative digital camera test.

They found that new ARRI camera has 3 stops more latitude than RED EPIC-X.

some of the main Hollywood cinematographers like a R. Deakins, M. Ballhaus, R. Richardson, etc, ..(just to name a few top cinematographers)

that before were mostly shooting film now at 2010 suddenly switching to digital (ARRI new digital camera)

because they said that it has equal sensitivity of the main film stocks like a Kodak Vision 3 500T...



Scenario-2:

Next year at October 2010 somebody doing a comparative digital camera test.

They found that new RED Epic digital camera system has equal or even more exposure latitude than new ARRI digital camera system.

Also EPIC has 5K resolution and with new RED de-Bayer software you can get a pure 4K resolution that has equal as 4K film scans.

With almost the same exposure latitude and resolution as film scans at 4K RED EPIC digital camera system with its specs "attracted"

some of the main Hollywood cinematographers like a R. Deakins, M. Ballhaus, R. Richardson, etc, ..(just to name a few top cinematographers)

that before were mostly shooting film now at 2010 suddenly to digital (RED EPIC new digital camera system)

because they said that it has equal sensitivity of the main film stocks like a Kodak Vision 3 500T...



What I would like to "hear" is something like a "Scenario-2" with Mysterium-X upgraded new RED cameras for the next year.

Just in the case that "Scenario-1" could possibly happen then a lot of professionals will be going to switch on ARRI digital just because

it would have the same price range as a brand new full EPIC RED digital camera system between Euro 30.000 to 70.000 it depends on the configuration.



Also do not take all those names and comparisons so serious just because any future is unpredictable anyway.



That was about a little warning before the main game start next year.

 

[Shawn Nelson]

3 stops? Come'on, i've repeatedly heard 1 to 1.5 stops

 

[Felix K.]

Sanjin, the Mysterium-X sensor is done. Nothing can be changed there. We saw pictures of it already. Your clever "warning" (what would RED be without you) is a bit late. In 12 months, I'm pretty sure, we will be discussing what the Monstro-Sensor can do.

No good cinematographer will make up his mind about digital cameras and then stick forever to whatever was best at the time of his decision. They will always keep an open mind about technical advancements.

Why is it so important what other people think about RED or who works with it as long as you are happy with your cam (which I thought I could read hidden somewhere between the lines of all your posts)?

P.S.: Ballhaus more or less retired as D.P.

 

[dustin lindgren]

Let me get this straight, Sanjin... you're concerned that your $17,500 camera isn't considered, by Geoff Boyle, to be just as good as a $300,000 camera?

You really need to stop worrying that not everyone in the world thinks your RED ONE is the greatest camera ever made (the same applies to the lenses you use) and just go out and shoot with it!

Exactly! Stop comparing cameras to RED that cost 15X more. Its a tool for people who want to use it. Go buy a $300,000 camera and tell me how much better it really is at the end of the day.

 

[Leif Thomas]

I can verify the results of the shootout on the Hands On HD workshop. My Red One (and 3 others) were involved and we saw some strange behaviors on all 3 cameras.

A copy of the original RAW material (and DPX-Sequenzes / ARRI RAW) is on my personal HDD and I will try and get some more information out of the tests when I find my time.

Under the given conditions (judging Redlog VS C-Log VS S-Log) there was no doubt the Red was lacking at least 2.5 stops. I will try some different post-routines, but with our grading system on the workshop the D21 (rated 200 ASA) got more latitude in highlights AND shadows than the Red One (320 ASA, build 20 / 21 beta) both on wide open Master Primes, conformed on a low-light / high contrast nightshot with a fire-breather.

On (unfiltered) 3200K it went even worse, but I think Geoff will get there in detail soon...

Unfortunately I'll be out shooting stuff the next 3 days, but I'll watch this thread closely for possible questions.

 

[sergio arguello]

cmon all this is just no longer needed or useful; superfluous the camera is DOPE!!
if DR is your focus then shoot film. All this brings to mind how many times i had to fight
with producers for REVERSAL film...get over it and sharpen your skills within the parameters of your equipment...stop waiting for 22 stops of latitude cause when it comes youll be out of a job...cause i can hear the producer saying "ill just shoot it"

just sayin

 

[Dominic Cochran]

The main idea behind that sort of warning was a reason that RED should not RUSH with shipping Epic-X

before Mysterium-X is not totally ready for the (bright) future in a comparison to the competition.

Sorry Sanjin

http://reduser.net/forum/showpost.php?p=487262&postcount=188

 

[Stephen Williams]

Hi Stephen,

there is no need to "advocate" G.B. from you here because whatever he and his collaborators

did with those tests there are going to be published and available to the public soon at CML pages.

.

Hi Sanjin,

You may remember David Stump's test of Frankie. 5000 people downloaded the tests, the Champagne was flowing until GB plotted the results against other cameras.

For some reason the Red One seems to perform badly in shoot-outs, lets hope Epic does better.

IMHO competition is a good thing Horses for Courses as GB would say.

Stephen