Latitude Testing Since Build 16 Release?

[Meryem Ersoz]

Has anyone done any? This once-fiery discussion has evaporated into thin air since Build 16 was released. From looking at my footage, it looks like more latitude than previous builds, but has anyone actually done any independent testing on the issue?

I've just been reviewing a large batch of recently shot Build 16 footage, and it is simply stunning. Stunning. Prettiest. footage. ever.

I guess that is all that we really need to know. Tests, schmests.

I am curious, though...

 

[Jannard]

We have the same latitude test results via Stouffer. 11 stops. It is all a matter of how much noise you can stand. Build 16 noise is much lower so the 11 stops seems much more reasonable.

Jim

 

[Stefan Christou]

Will Scarlet's 12 or so stops be a bit noisier than One's 11?

 

[sbcooler]

Very good question...am curious about Scarlet...if it can handle the dynamic range as well as the RED. Anyone?

 

[Esteban Sosnitsky]

Jim, how do you objectively judge the amount of noise one can stand. What do you compare it with? Whats your standard? I have been doing all kinds of tests in order to figure out "my working latitude" but still no answer. Although I still dont have my own camera on hand... soon though... soon.. A couple of friends helped me with their cameras on this tests and the results are no where as close as 11 stops.. so.. I guess its a matter of noise acceptability?

 

[Joel Kaye]

Jim, how do you objectively judge the amount of noise one can stand.

He can't judge how much noise YOU can stand. It's totally subjective. The only way to compare is vs. another camera you might use instead.

If you know you're going to shoot RED then do your own exposure test for scenes you are going to shoot.

 

[Esteban Sosnitsky]

I am sorry. I am not asking him to judge my tolerance to noise. I probably misspelled or should rephrase. I wonder how much noise does he objectively can stand before saying: ok thats not usable or worth to rescue. I know I am shooting with red, and I have done 5 tests with no luck so far.

 

[Stefan Christou]

Will Scarlet's 12 or so stops be a bit noisier than One's 11?

I still want to know

 

[Eric Lange]

HEY ESTEBAN,

Nice to see you well and alive on RED! Somewhat belatedly thanks for dinner at the SPIE and good conversation, we must get back in touch!

I have been wanting to hassle you about how its going with your new RED ONE!

Maybe we could build you a kick-Ass stereo camera with 3 arc second precision, for medium format and RED ONE.

Check out some of my later posts and greedy wish lists for the proposed high res still and lower res dynamic camera. That would dramatically improve the workflow for our 3d stuff.

Hope you are well!

Will lurk around for a bit of moral support, but must get back to things,

Eric Lange

 

[Eric Lange]

Jim, how do you objectively judge the amount of noise one can stand. What do you compare it with? Whats your standard? I have been doing all kinds of tests in order to figure out "my working latitude" but still no answer. Although I still dont have my own camera on hand... soon though... soon.. A couple of friends helped me with their cameras on this tests and the results are no where as close as 11 stops.. so.. I guess its a matter of noise acceptability?

I am not a cinematographer, but here's a weird idea. The guys that did the Hubble Space Telescope imaging had this nice technique when working with hopelessly noisy imagery. For their deep space work they would image the same narrow piece of sky hundreds of time. They would in effect sum all of the imagery and then divide by the number of images. Astonishing imagery would come out of incredible noisy dim images. I.e. the larger fraction of the noise had been averaged/processed out.

Perhaps what you could do is lock the camera down very steadily and at your particular settings, control lighting etc. Then shoot a hundred frames or so (or less). If you are able to batch process, to sum and divide the imagery you could produce a comparison shot. This shot would be a synthetic theoretical maximum that your conventional shot could be compared against. That way it might be easy to see how much noise YOU WANT to live with for a particular scene or set up as a basis for comparison. In essence it would be a relative scale rather than an absolute scale, tailored more for the immediate practical requirements. The number of shots required to reach an asymptote to wash or average out the noise, e.g. five shots, 50, or hundred?

Just a crazy idea, maybe you guys already do stuff like that ?

 

[Eric Lange]

Oh yeah, If forgot to mention as from above if you then subtract frame 1 from the averaged 100 frame composite then you can SEE where your noise really is or where its coming from.

I'm sure you cinematographers already do something like this?

If not, maybe EstebanRed this might give you something to get your teeth stuck into!

Cheers

 

[Fredrik Callinggard]

I believe that the latitude is somewhere between 4 - 5 stops depending on circumstances and the dynamic range somewhere between 9.5 - 11, here also depending on circumstances.

 

[Illya Friedman]

The guys that did the Hubble Space Telescope imaging

Hubble imaging sensor = DALSA



I.

Illya Friedman
DALSA
Digital Cinema Division
www.dalsa.com/dc

 

[Esteban Sosnitsky]

I just bought a Phase One back p65+ it hasnt came out into the market yet, but their press release says its sensor is also made by dalsa. (actually co developed with phase one)
They claim 12.5 stops of dynamic range, now that is awesome too!!! I mean 12.5 stops!!! I already own a Phase One and I know how much noise is enough. With the RED I am having issues, basically because I don't know how to monitor it. Maybe I am just ignorant in the topic but I just don't have a 4K monitor to judge my footage. On the computer the screens lie, I would have to make a film out and see for myself on a theater screen. BUT If I take my proxies in the worst case scenario into a CRT SD monitor even in the worst conditions it looks great. So what should it be??? For me?? I still dont know.

 

[Eric Lange]

Hubble imaging sensor = DALSA



I.

Illya Friedman
DALSA
Digital Cinema Division
www.dalsa.com/dc

Hey Illya, very cool indeed.

Just two weeks ago I downloaded the specs on the Dalsa 28MP chip (FTF 6146C), and the 38 MP (FTF 5066C) as well as the application notes on binning.

We really wanted to see if it was technicaly possible to grab a higher frame rate from a smaller portion of these sensors. For our applications and software and hardware, which are best described as a stereo photogrammetric inverse animation systems we need high res capture to provide good 3d key frame modeling data and the low res dynamics to guide the animation of the high res captured data. We were thinking of approaching LEAF (as we are already designing high precision stereo 3d cameras around their backs) to see if we could squeeze out just 14 frames per second from a small area of the chip.

A week later we saw Jim’s announcement of the possibility of a double duty camera that could do exactly what we want (the power of vapour ware!). From a technical stand point we would rather grab an under sampled version of the data from the whole area of the CCD, rather than a binny sub sample. Here I think CMOS (and their transistor arrangement, may have the edge). The CCDs I believe would yield better tonal information, however I feel the CMOS would offer a greater range of functionality. Unfortunately, convenience usually gives way to quality, (just look at how film gave way and is giving way to digital). Soon there will be a generation that has no visual memory of what film used to look like, so that all worries about visual noise will literally die out.

Illya what do you think about the prospects for CCD technology being able to deliver high res static and sub sampled dynamic (at a good frame rate)?

[ESTEBANRED I sent a PM to clear things up J]

 

[NateWeaver]

Unfortunately, convenience usually gives way to quality, (just look at how film gave way and is giving way to digital).

Nobody at the upper end of the spectrum picks Red because it's more convenient. At that level, 35mm is the most convenient because the post is as old as dirt, and nobody has endless conversations trying to figure out how to do it.

Sometimes quality in a format has less to do with the actual image, and more to do with what it frees the director up to create.

 

[NateWeaver]

Just a crazy idea, maybe you guys already do stuff like that ?

Yes, it's been used before. It can introduce weird motion artifacts if used too heavily.

 

[Eric Lange]

Nobody at the upper end of the spectrum picks Red because it's more convenient. At that level, 35mm is the most convenient because the post is as old as dirt, and nobody has endless conversations trying to figure out how to do it.

Sometimes quality in a format has less to do with the actual image, and more to do with what it frees the director up to create.

Nate I like your post, especially "and nobody has endless conversations trying to figure out how to do it.".

Although I literally grew up in the film business, one of my main technical backgrounds is in the dark art of photogrammetry. We used to use huge Zeiss UMK 1318 format cameras on glass plates up until the mid 1990's to effect high res 3d capture etc. The huge plates were then scanned to 7.5 micron res. Huge files, and excellent lenses and imagery. Even though a medium format back has a lower total resolution than large plates, its much more effective at getting the job done. I remember 12 years ago when I was at the Getty (as part of a research fellowship) having to persuade everybody that digital photography was the way to go (for 2d and 3d documentation of complex archaeological and historic sites). Bitter arguments would ensue between various technical and executive staff members, and even when you showed (actually carried out) how a large site can be recorded to a global resolution of .5 mm across its entire surface, some of the die-hards were unconvinced as to the merits of digital. Back then people were trying figure out how to tame this difficult digital stuff too. In only six years (later), they were all using digital on their field exercises and had installed a large magneto-optical juke box for automatically managed (triple redundant parity checking) of all the data.

I still miss the quality of large plate transparencies, but not the difficult work flow. In thrirty/fourt years no one will have a living memory of what a silver halide medium can produce.

In the case of RED, people and developers (as you seem to hint) are still trying to figure things out, but within six years its a workfow that should be very well sussed out ( I hope). However, I do believe in terms of a long term archival medium some sort of film output would be prudent, if there comes to be a truly softcopy work flow, as it is far too easy to loose digital data. At least film may degrade and be digitally restored later on. I cannot even comprehend the amount of digital data that will have to managed for film archive purposes in the future if we really go to completely digital workflow and delivery system. Maybe by then RED et. al. will be the biggest data storage company in the world.

As far as the 100 frame averaging technique, its good to know people use that, as a test (I meant that the camera would be locked still on a very stable mount to achieve this, I.e. a static test). Thanks for the info.

 

[Eric Lange]

Nobody at the upper end of the spectrum picks Red because it's more convenient. At that level, 35mm is the most convenient because the post is as old as dirt, and nobody has endless conversations trying to figure out how to do it.

Sometimes quality in a format has less to do with the actual image, and more to do with what it frees the director up to create.

I suppose a better question would be what do you personally consider as the limitations of digital/RED versus film in terms of creative freedom.

I am not a cinematographer and would like to hear it straight from the horses mouth (so to speak).

 

[Blair S. Paulsen]

Exciting times

Exciting times

Now that this thread has been completely hi-jacked I can't resist jumping in. Once humans become adept at operating in a particular way they achieve a comfort level that can allow them to return their higher level processing functions to less mundane tasks. For a large majority of the top craftspeople and creatives in the motion media business that means 35mm film and all that goes with it. That said, there are many specifics of dealing with 35mm film that hardly "free up" the creative process.

Less weight, better ergonomics, longer run times between media replacement, accurate hi-res monitoring, innovative metering & focus tools, lower lighting requirements, better image quality - wide dynamic range - large color gamut - resolution, etc all matter and if RED, or anyone else, can deliver those attributes along with lower operating costs then they will have the most popular cameras right up to the highest levels of production.

Jim & the Red Team have changed the game and the other players will not sit idly by. I forecast a decade of advancement in camera design coming up that will significantly change the way we interact with our tools.