Craziest Idea to Date

[Paul Bruening]

I think I've come up with a way to cheaply (relatively) jam two Mitchells together to get cine HDRI. What I'm curious about is the intersecting values of the 2 paths of film. I've included two graphs. One is Kodak's exposure range graph for 5217. The other is two versions of that graph overlapped, representing the two paths. The format I have in mind is 2-perf converted by 4K scans and combined in computer.

 

1. Do you think the graphs need to be further compressed together, expanded apart (by how much) or are they about right as they are?

 

2. How many useful stops total do you think can be pulled from this approach?

 

3. How many stops difference would you ND the two units by?

 

4. Where on the overlapped graph would you place a single median exposure point for a.) sunny day (top of the lens) and b.) well lit by street lamps night exteriors (bottom of the lens (such as in David M's recent Chicago shots))?

[Chris Millar]

1: No idea

2: 22

3: 9

 

(I counted them on your graph)

 

4: I know its sounds funny to say considering what you're up to but:

 

'bracket'

 

(to find your average exposure ... or bias for shadows etc...)

 

show off! :D

[Michael Collier]

A wise man once said "Before a mad scientist truly goes mad, there is a time when he is only partly mad. This is when he will throw his best parties."

 

So when is this party?

 

 

Looks like your getting into a wold of complexity (bob and weave making lining each shot up tricky) double the cost of film, more than double the weight, etc.

 

But if you must then go for it. I would compress them just a bit. You want there to be a bit of overlap in the straight line sections, so you can dissolve smoothly between them. You will want to ax the non-linear part in both videos before combining them (probably by doing a luminance matte)

[David Bowsky]

Just started playing with HDRI in stills and wondered about how to get there with cine. Fascinated to see where this discussion goes.

[Satsuki Murashige]

Paul, just throwing this out there but I would love to see one of your photoshopped pics of your proposed 'two (!) Mitchell camera rig' (handheld config. would be especially awesome!).

[Chris Millar]

Paul, just throwing this out there but I would love to see one of your photoshopped pics of your proposed 'two (!) Mitchell camera rig' (handheld config. would be especially awesome!).

 

The blimped 3D version with home built video taps and steadicam :lol:

[Paul Bruening]

This is my first guess on question #4. Please, help me with what needs to be moved where or changed to make the graphs more accurate. There's no way to improve shadows or stops on the thick end of the scale on night exteriors and only little help with the thin end. Sunny days look much more useful on both ends of the neg (assuming I haven't royally messed something up).

[Paul Bruening]

It just dawned on me. I could be thinking about this all wrong. If I'm using the same, single lens and same ASA stock for both paths then any stops of ND won't do anything but make the filtered path uselessly thin. Is that right? This may actually be the dumbest idea I've ever come up with. Wouldn't I have to make the HDR path difference through separate lens apertures? Two lenses are no good for this.

[Phil Rhodes]

Isn't this quite a lot of work to go through just to avoid having to set a stop?

 

P

[Adrian Sierkowski]

Hell if you wanted to keep this simpler why not just build a cine camera with a prism in it and work it like a 3 chip CCD system, and bigger film. What I mean by this is using the prism to break the light up into 3 separate paths, one is set for exposure one is NDs a lot for highlight information and 1 gets fogged a bit for shadow detail. Then using mirrors etc you put all the images back onto 1 big piece of film, say 65mm or something like that, with ND on top, then normal, then dark, all within 1 frame, which then get overlaid by some software later on.

 

Does that make any sense? You'd still be using a big ass camera, and expensive film, but you'd have just 1 lens on the system so you can get all those racks etc and focus should be easier to keep. And also since you're just using 1 piece of film, you don't have to worry about syncing 2 movements etc.

[Paul Bruening]

Isn't this quite a lot of work to go through just to avoid having to set a stop?

 

P

 

The idea is to get stunning images. Two 2-perf Mitchells running short ends is still way cheap. Two 400' loaded Mitchell mags and cameras in a well made, fiberglass blimp is still lighter and smaller than one loaded Mitchell in my metal blimp. If it had worked, it would have been a massive leap in image quality for a short leap in equipment hassle and cost... if it worked.

[Paul Bruening]

Hell if you wanted to keep this simpler why not just build a cine camera with a prism in it and work it like a 3 chip CCD system, and bigger film. What I mean by this is using the prism to break the light up into 3 separate paths, one is set for exposure one is NDs a lot for highlight information and 1 gets fogged a bit for shadow detail. Then using mirrors etc you put all the images back onto 1 big piece of film, say 65mm or something like that, with ND on top, then normal, then dark, all within 1 frame, which then get overlaid by some software later on.

 

Does that make any sense? You'd still be using a big ass camera, and expensive film, but you'd have just 1 lens on the system so you can get all those racks etc and focus should be easier to keep. And also since you're just using 1 piece of film, you don't have to worry about syncing 2 movements etc.

 

2-perf is really, really cheap. Mithcell's are really, really, cheap (I already own them and their registration is flawless). I already have one beam splitter and veiwfinder optics. I already have the fiberglass and resin. All of that is my principle defense for this design. It was cheap.

[David Bowsky]

The idea is to get stunning images. Two 2-perf Mitchells running short ends is still way cheap. Two 400' loaded Mitchell mags and cameras in a well made, fiberglass blimp is still lighter and smaller than one loaded Mitchell in my metal blimp. If it had worked, it would have been a massive leap in image quality for a short leap in equipment hassle and cost... if it worked.

 

So, you are going for an expanded dynamic range effect, but still with the goal of a relatively realistic looking image rather than the more surreal HDRI stills where the range of the final composite is greatly expanded?

 

Re: Adrian. Beam splitting after the lens onto a single stock seems like a good idea. I'm trying to imagine the modifications needed to pull that off, but a little baffled. At the risk of stating the obvious, a system like you describe would also have an effect similar to putting a tube between the rear of the lens and the film plane, yes? Or does prismatic splitting change the rules there?

[Adrian Sierkowski]

@ Paul, fair enough. was just trying to think of a way to keep it simpler in terms of moving parts etc.

 

@ David, honestly I'm not that up on optics/theory etc, just a way that makes sense in my head. The prism gives you 3 images from 1, and then you'd use mirrors/lenses to move that image back to a piece of film. I know MTF would be affected and there would probably be some issues keeping it all sharp, and you'd loose some light in the transmission, of course.

Another issue would probably deal with the FFD of lenses you're using; though again I'm not mathmatically inclined (or experienced) enough to really follow through those calculations etc. Just a "what if" idea in my head.

[Paul Bruening]

So, you are going for an expanded dynamic range effect, but still with the goal of a relatively realistic looking image rather than the more surreal HDRI stills where the range of the final composite is greatly expanded?

 

Re: Adrian. Beam splitting after the lens onto a single stock seems like a good idea. I'm trying to imagine the modifications needed to pull that off, but a little baffled. At the risk of stating the obvious, a system like you describe would also have an effect similar to putting a tube between the rear of the lens and the film plane, yes? Or does prismatic splitting change the rules there?

 

Yup. Expanded detail to help with the normal, log, roll-off of film, taking the extra stops that came along for the ride (which might have been 8 to 10 extra stops if it had worked). If beautiful but not too funky looking images are called for then that's the goal. If surreal is called for then that's the goal. My current project could use plenty of both given it's low overall production value.

 

1/2 silvered mirrors are the more common approach these days. I'm not savvy enough to know the deep, engineering and physic behind beam splitting. I know that splitting mirrors have less glass, flawless reflective coatings if I don't buy the student grade off Edmund's, are more plentiful and are much easier to mount than prisms (did I mention, cheaper?). You lose a full stop in both paths on each mirror. I can take two hits on my ND'd path and keep a 2 stops darker viewfinder (compared to spinning mirror reflex) since that path needs darkening anyway. I can't get around loss in image quality. I can reduce the loss by parsing the quality of the mirrors available.

[Paul Bruening]

Another issue would probably deal with the FFD of lenses you're using; though again I'm not mathmatically inclined (or experienced) enough to really follow through those calculations etc. Just a "what if" idea in my head.

 

I had thought of a two path, Tscope framed, VistaVision pull-down where the two images were separated by two perfs and 1/2 leap frogged over each other. The optics were within reach. But, I can't put my hands on that dang VV camera body and medium format lenses.

 

If I turn Mitchell body number three upside down and go with medium format lenses I could get results that I would never find a use for (assuming the graphs are correct). I suppose I could get a shot of a person walking up to a 20K and still get full detail as they climbed into the housing. I'm not trying to be an smarta** to you, Adrian. Just providing the kind of example of where 3 paths would be useful.

[Adrian Sierkowski]

No, I know you're not. My biggest worry when dealing with multiple bodies etc would be getting them all to "work together," ya know what I mean? I mean even though there's no dispute about the reliability and stability of a Mitchell, but rather as you increase the moving parts, I at least would worry about it not working right, ya know?

As for the 20K shot; hell could be nice, almost frame and forget because you know you have the range to get it just right in post.

[Paul Bruening]

No, I know you're not. My biggest worry when dealing with multiple bodies etc would be getting them all to "work together," ya know what I mean? I mean even though there's no dispute about the reliability and stability of a Mitchell, but rather as you increase the moving parts, I at least would worry about it not working right, ya know?

As for the 20K shot; hell could be nice, almost frame and forget because you know you have the range to get it just right in post.

 

My two solutions for that was using my existing Aranda controller and motor and add another motor through a break-out. The head/tail bloop light would bleed into both paths for path matching and the controller would maintain frame sync. The idea for an under-carriage, toothed belt, gearbox and shafts design is doable at my level of engineering but a little more expensive than simply adding another motor.

[John Sprung]

Um, how about a much simpler approach? In total darkness, buff the anti-halation backing off of a fast stock. Then just bi-pack it emulsion to emulsion with a slow stock, shooting thru the fast stock's base? One camera, just a little flange focal distance adjustment, the registration problem gets a bunch easier....

 

Oh, yeah -- the emulsion and filter layers would be in the wrong order on the fast stock. You could test it emulsion to base with a long lens and deep stop, just to see what the exposure issues are. And if it works, special order the fast stock coated the other way around from Kodak.

 

 

 

 

 

-- J.S.

[Paul Bruening]

Um, how about a much simpler approach? In total darkness, buff the anti-halation backing off of a fast stock. Then just bi-pack it emulsion to emulsion with a slow stock, shooting thru the fast stock's base? One camera, just a little flange focal distance adjustment, the registration problem gets a bunch easier....

 

Oh, yeah -- the emulsion and filter layers would be in the wrong order on the fast stock. You could test it emulsion to base with a long lens and deep stop, just to see what the exposure issues are. And if it works, special order the fast stock coated the other way around from Kodak.

 

 

 

 

 

-- J.S.

 

I thought about that one too. Keeping the two stocks pressed together to avoid flutter required stamping mechanisms or air pressure or vacuum. That's beyond my abilities. The air idea meant more gear and/or more noise tethered to the camera. The other problem was the costs at Kodak to remake the front strip with switched order of emulsion and leave off the coating and/or a lab that could wash off the coating without altering the emulsion.

[John Sprung]

Technicolor used bi-packs without a whole lot of extra stuff -- no pneumatics. That might not be a real issue. A still camera test might be fairly easy to do. The big one is getting the special backwards emulsion made.

 

 

 

 

 

-- J.S.

[Paul Bruening]

Technicolor used bi-packs without a whole lot of extra stuff -- no pneumatics. That might not be a real issue. A still camera test might be fairly easy to do. The big one is getting the special backwards emulsion made.

 

 

 

 

 

-- J.S.

 

Interesting, John. As I read your post, it occurs to me that with 2-perf's inherent vertical pressure combined with the normal perf line pressures, flutter may occur insignificantly or not at all.

 

Now, how to get short ends for reversed layered, no back-coated 5219? 5212 is already a gettable stock.

[Paul Bruening]

Interesting, John. As I read your post, it occurs to me that with 2-perf's inherent vertical pressure combined with the normal perf line pressures, flutter may occur insignificantly or not at all.

 

Now, how to get short ends for reversed layered, no back-coated 5219? 5212 is already a gettable stock.

 

I could probably make a passable coating washer. Could I just fudge the normal layers of the top strip in post to dodge the special stock costs? A simple Adobe invert color command might do the trick. I only have to switch the blue and red channels.

 

The BiPack magazines already exist, as well.

[John Sprung]

First, shoot a deep focus back to front still camera test -- with a long lens stopped down. That should be quite doable, all you have to do is wash off the backing in a darkroom first. Such a test will tell you what kind of exposure you'd get on the second film. If that works OK, then talk to Kodak about the special coating issue.

 

 

 

 

-- J.S.

[Paul Bruening]

First, shoot a deep focus back to front still camera test -- with a long lens stopped down. That should be quite doable, all you have to do is wash off the backing in a darkroom first. Such a test will tell you what kind of exposure you'd get on the second film. If that works OK, then talk to Kodak about the special coating issue.

 

 

 

 

-- J.S.

 

Sorry, John. I was editing my previous post while you were posting yours. What do you think of that post trick to get around special film costs?