Zone system understanderers: Curve/DR optimizing for shooting off screens

[Chris Millar]

Gidday,

 

Imagine I have filmed a scene with a digital video camera of a sunny scene, something that would have a high Dynamic Range (DR) - and or a CG or composite scene in which the thing in er, real life that it is emulating would have a large DR...

 

I am now filming this back onto film (like on ya know film) off an LCD or HD CRT and am getting a DR of say 6 stops - try these quick numbers on for size:

 

%0 'black' = 4EV

%18 'grey' = 7EV

%100 'white' = 10EV

 

(measured with a spot meter off an LCD screen)

 

For any given clip I can find the max and min luminance values of the clip (in Nuke) and optimized it so the clip achieves these values - this ensures that we are utilizing the screen fully.

 

I have control over the curve 'in post' and also of course have control over the push development of the film - the way I see it, the push will expand the limited DR of the screen output to cover the latitude of the film (right?) but, and here is the real question: what happens to the distribution of levels across the new/pushed curve ? Especially shadows and highlights - will I have to play with the curves in the file to counter any unwanted effects of the DR expansion in the push ?

 

The goal is to learn how to optimize filming off screens so that the footage will be inter cuttable with in camera footage - you could take it for telecine and one light it for instance...

 

I understand all the other issues re. filming off screens (color cast, shutter angles, flicker and the rest) - not to mention the futility many of you will be prepared to comment on but I'd really like to keep this on-topic and just deal with luminance/DR/gamma issues (please :rolleyes:)

 

Project will be shot on B&W neg if it's of interest - could just as well be reversal - and yes, I develop it myself ;)

[Bryce Lansing]

Your question is a bit confusing, are you asking if you can shoot a scene digitally with only 6 stops, then shoot a screen of that footage with film to get more DR?

[Chris Millar]

Here it is a bit more simplified:

 

What happens to shadow and highlights when you push a film to expand a scene DR of 6 stops to make full use of film negative ?

 

The issue is that I'm shooting something that already has a characteristic curve on it (as I'm shooting something already shot).

 

It's hard to write the question adequately - need diagrams or at least hand gestures - I feel that the question is valid however ...

 

:)

[Brian Rose]

It would seem to me that the grain would be bumped up a bit, and you might lose a bit of clarity, as well as have some color aberration.

 

Since you operating by the zone system, what about flashing/pre-exposing the print film by -4 or-5 stops? That could boost shadow detail, with less impact on the film's physical structure, as would be the case if you manipulated the chemical processing via push method.

 

BR

[Chris Millar]

It would seem to me that the grain would be bumped up a bit, and you might lose a bit of clarity, as well as have some color aberration.

 

Since you operating by the zone system, what about flashing/pre-exposing the print film by -4 or-5 stops? That could boost shadow detail, with less impact on the film's physical structure, as would be the case if you manipulated the chemical processing via push method.

 

BR

 

hmmm - yes grain ... The end goal is in black and white film, hand processed roughly so I can accept grain - but in saying that I have found some old CRT stock with a high gamma which will help reduce the required push and it's ASA is very low (single digits) so I'm hoping for small grain too (it's made for it basically).

 

Hmm - pre flash - yes, now we're talking ! Would this be the same as pumping the shadows up a little in the original file (curves in Nuke, FCP etc...) ? Which is not to say the black level, but the shadows as a whole...

 

Since I have control over the original footage (CGI/digicam/comp/whatever) and I can play with its extremes I like how the zone system can be used directly, no half baked cine version here ;)

[Charles MacDonald]

Since I have control over the original footage (CGI/digicam/comp/whatever) and I can play with its extremes I like how the zone system can be used directly, no half baked cine version here ;)

 

Sorry if my Bias is showing but I always figure that Film can capture more dynamic range than most digital systems

 

The Web page for the multiprocessing group does have an article on using multiple exposures to increase dynamic range when going between video/digital and film. I am not sure if this is what you are thinking or or indeed the exact opposite, it is in the lower part of this page http://web.ncf.ca/aa508/MovieProcessing/

[Dominic Case]

Hmm - pre flash - yes, now we're talking !

If I've understood the question, then preflashing is the last thing you'd want to do.

 

You've got an original scene with a Dynamic Range of - say - ten stops. You've shot this on a digital camera. Your display device is giving you a maximum Dynamic Range of six stops (EV 4 - 10), and you can manipilate the curves of your images so that you utilise this range fully.

 

You want to shoot the image off the display device onto black and white film and produce an image that looks as if it was the original scene shot directly onto film. (or is at least intercuttable with it).

 

Film negative has a much wider dynamic range than six stops, and it would be able to capture all of the original scene, albeit with some toe and shoulder shaping. So if you simply shot on camera negative, you would end up with a very low contrast image compared with the original.

 

Flashing the stock does indeed lift shadows, by moving them from the toe of the curve up onto a slightly steeper part. In fact it extends the dynamic range of the emulsion. But in doing so it reduces the overall density range of the negative, and therefore reduces contrast, particularly in the shadows. It's normally used to reduce contrast. You want to increase contrast.

 

Nothing you do to the display image will increase the range on your final negative, it is limited by the maximum brightness of the device. It's why CRT film recorders gave way to laser recorders some years ago. The high contrast stock you shoot on is the only viable method of getting the range you want in the negative. And the short answer to your question has to be: "do tests".

 

Create a grey scale on your display device, in half-top intervals through the 6 stop range it is capable of. Measure on your spotmeter, and correlate that against the values in the digital image.

 

Shoot that onto your hicontrast stock at a bracket of exposures. Normal processing should be about right - as you say the stock is designed for the purpose. Then you should be able to measure the density of each step on the negative and plot it against the original input values to see what happens at each end of the curve and the middle. You can measure the density by projecting the negative and using your spotmeter again, in the absence of a real densitometer.

 

Basically you are looking at the process that is used to calibrate any film/digital/film i/o system, such as the DI chain of scanning, grading, & recording back to film. The principle to follow is a closed loop, of unity characteristics.

[Chris Millar]

I've posted this question on a couple of other forums also - I said 'Hmm - pre flash - yes, now we're talking !' as Brian was the first person not to start talking about flicker and frame rate issues (the common malaise with this kind of carry on) - it was exciting to finally get on topic somewhere ;)

 

Right, with that being said yes, you've understood me correctly (did I explain myself clearly or was it a bit jumbled ?) anyway, I think you're only person so far who has understood the question from word go and been able to provide an answer... which is the usual answer, TEST...

 

heh heh

 

I was just trying to ball park it and also start off a discussion that might introduce me to some concepts I am yet to grasp but in the absence of further advice I'll just get down to it. Yes, I'll shoot the charts and I'll edit together a small section of the clip that is most representative of the full luminance ranges also (spatially and temporally).

 

I'll be projecting and spot metering yes - guess I'll do it outdoors, its summer after all (down under also)

[Chris Millar]

The stock, by the way is 7374 'television recording film' - I have poopy poop pooper loads of it - ex NZ National Film Unit if the stories are true, passed from person to person, Peter Jackson is a rumored ex owner...

 

ha ha :rolleyes:

Edited March 3, 2010 by Chris Millar

[Dominic Case]

ex NZ National Film Unit if the stories are true, passed from person to person, Peter Jackson is a rumored ex owner...

only indirectly . . .

 

NZ National Film Unit was privatised some years ago and became Avalon, and at some stage the lab became just "The Film Unit".

 

Then it went up for sale again, and Peter Jackson bought it rather than see it close down or go into foreign hands. He then built a top class sound mixing facility across town, moved the lab into the same place, added a digital facility and called it all Park Road Post. Moving a lab isn't easy, but in his spare time, he relaxed by making a film or two - or three. Did quite well. No bad for a lab owner.

 

That 7374 stock is probably quite old. I can't remember the last time I came across it.

[Chris Millar]

ah,

 

got ya - and yes it is old, rusty cans ... But its been used on a few music vids over the years and I've had success with it doing timelapse also - it's faster outdoors than in, spectral sensitivity all over the show ;)

 

Just to get things back on topic someone elsewhere suggested I use HD projectors - 3 chip DLP's he said I'll get more DR from them from black to white (assuming a good set up) - how much more though ?

 

Specifications found online for both screens and projectors are a load of BS basically.

 

Also an idea re. a kind of backwards HDR - I still need to think about it but the general idea was that I find the average luminance of the video footage then split the footage luminance wise (this is the bit I'm foggy on and need to think about, is it even possible?) - anyway, so imagine you have two streams, one shadows one highlights (relatively) - now expand each of these to cover the full range of the screen (so original 126 becomes 256 of the shadows and original 127 becomes 0 of the highlights) - underexpose the shadow stream then wind back and overexpose the highlight stream (goal that you 'connect' back 126 to 127 continuously)

 

How the flip do you split a clip luminance wise ?