dan kessler

Just to clarify my previous statement a little more... there's a common misconception that CG software does all the work for you. All we need to do is push the right button. While the software is impressive, even high-end packages like Maya and Houdini just don't work that way. They're great big toolboxes that give you CG equivalents of hammers, screwdrivers, pliers, maybe some jazzy things like table saws and milling machines. You want a fancy weapon attached to the elbow of a live-action character? Sorry, there's no software on earth that can do it. You need an artist who knows how to use the tools.

Time-consuming frame-by-frame animation matching CG to plate action with pixel accuracy, then lighting and compositing CG, again, to match the plate. Done all the time by experienced CG artists.

Boy, we're covering a lot of ground here, but -- yes, it was Jerry Lewis the comedian, actor, director who first stuck a video camera onto a Mitchell. He was making his movies for Paramount and they bankrolled the development of the video playback system for him.

Not to confuse the issue, but hopefully make it very clear... if your end product is the HD transfer, then you have the options described above. If, on the other hand, you were going the photochemical route, as David Mullen asked earlier, the answer would be different. When color timing via the film printer (neg-pos contact, the "old fashioned" way) your options are somewhat limited to overall scene brightness, color balance, etc. If you want extensive digital control over every aspect of an image destined for film output, then think digital intermediate workflow, i.e., hi-res scan the original neg, manipulate it digitally, then record back to film.

I think this is a common way of converting spherical zooms to anamorphic. You end up with double the focal length and half the speed, if I'm not mistaken.

You say you're clueless, so it's possible that you might not understand the answer to your own question. Stated simply, this is the look you get with telephoto lenses. The longer the focal length, the more drastic this look will be. Before embarking on your project, you should acquaint yourself a little more with the basics of cinematography and with the camera you intend to use.

My cemented lenses DO NOT correct CA. Achromatic lenses consist of two or more positive and negative elements with different indices of refraction. They can be cemented or air-spaced. Together they add up to a positive or a negative lens. In my case, this is all for a home-built camera.

The amount of squeeze is determined by the ratio of the focal lengths of the lenses. For a 2x squeeze, the positive focal length is twice the negative focal length, with the negative lens positioned at half the positive distance. The system is afocal, meaning that parallel rays enter and leave. It's not an image forming system, but is designed to work with an image forming system. Rays enter the negative element and exit the positive element. If you are familiar with graphical ray tracing, you can draw this easily and understand how the system works. I've been calling my set-up a de-squeezer, but it's really a squeezer. The image from the ground glass has the 2x horizontal squeeze from the camera lens, and the viewfinder then applies a 2x vertical squeeze to restore the image proportions. No matter, it's the same result. I chose specific focal lengths to fit my design, but the actual values can vary. Of course, if you buy lenses from a place like Surplus Shed, you are limited to what they have in stock. If they don't have what you need, you're forced to go to a retail supplier, and you will naturally pay a lot more. All of these cylindrical lenses are simple plano-concave or plano-convex, so aberrations are definitely a concern. I made my negative element into a double concave by cementing two plano-concaves flat-to-flat. For this you start with two negative lenses that are twice the focal length you want, because the combined focal length will be half. This improves the shape factor, helping to reduce the spherical aberration of the overall system.

Robert - Your description of the hypergonar is exactly right. This was the original Cinemascope lens, by the way. I think the limiting factor at wide angles is the barrel distortion. As far as I know, there's no way around it unless you use a different lens design. As for my viewfinder, I used the same principles, although my anamorphic lenses are not "in front," but "in the middle." The objective lens of the viewfinder is a symmetrical design. I separated the two elements so that the light rays between them would be parallel and put the anamorphic lenses in there. I reduced the spherical aberration by making the negative cylindrical lens a double concave shape. Spherical and chromatic aberration are further reduced by using a stop in the viewfinder system. It works quite well. I bought all the lenses from a place called Surplus Shed for less than $50. The main reason for building stuff is because we can't or won't spend thousands of dollars to buy the best equipment. With off-the-shelf lenses, there are, of course, limitations. They work very well with relatively narrow viewing angles at moderate to slow speeds. For a photographic lens, that means using design principles that have been around for 150 years. Needless to say, though, wide, fast, well-corrected photo lenses are far more demanding and anamorphics are more specialized. Can't build those quite so easily with parts from Surplus Shed.

I, too, have been down this road, having built a de-squeezer for a camera viewfinder. Building a photographic-quality anamorphic adapter is much harder. The first problem is that almost all off-the-shelf cylindrical lenses are simple plano-concave or plano-convex shapes. While you can indeed put together an appropriate combination of focal lengths to obtain the desired amount of squeeze, the system will suffer greatly from spherical, chromatic and other aberrations. You need achromatic lenses to overcome these aberrations. I know of no company anywhere that manufactures achromatic cylindrical lenses in a wide range of sizes and stocks them as standard items. Getting custom optics would be very expensive and defeat the whole purpose of building it yourself. The only other simple way to reduce aberrations is to stop the system way down, meaning that you could only use very small apertures. Cannibalizing existing adapters for their parts doesn't seem like a good solution, either.

Simon, just received and read that article you cited. Very, very informative! Thanks!

Well, that's more trouble than I would've asked for, but if you do get the chance, that would be awesome. Thanks, Stephen.

I understand, Simon, but a snug fit to what gauge? Here's what I'm driving at. I once made some registration pins using #35 drill blanks. The blanks have a nominal diameter of .1100, with a tolerance of +0000 -.0002. I ground the flats to match the nominal BH perf height of .0730, using the same tolerances. So, these were definitely within print specs for BH perfs, but how do they compare to a Mitchell or Panavision? Do they gauge to the upside, say +.0002 -0000, or something else? Do they start with a larger nominal size, smaller, what? They have to have a gauge, something they consider to be their standard. Surely they don't just grab any old piece of film and stick it on.

Appreciate all the feedback so far. I knew about the Mitchell legacy in Panavision cameras, but does anybody know the actual pin dimensions and tolerances? I suppose I could rent a camera, take it apart and mic the pins, but I thought I'd ask around first. :)

Andy Warhol

I'm hoping for some authoritative feedback regarding the registration pins in both of these systems. I'm familiar with the specs and tolerances for BH perfs, but I want to know if camera makers use these exact same specs for the pins, or do they use something else, e.g., oversize? Also, which pin is the full-fitting pin in Mitchells? Do Panavision's dual full-fitting pins really give superior registration? The Mitchell scheme adheres to time-tested tool and die making principles, so I'm curious if Panavision does have a better mousetrap. Thanks.

Here's a brief overview of the typical fx pipeline: The camera footage, or background plates, are scanned in. The 3d animators and effects artists will model, animate and track their elements to match the plates. The cg lighters will light the elements to match the plates. The cg elements are then rendered, usually as multiple layers, i.e., shadows, diffuse lit, spec lit, etc. The rendered elements then go to the compositor, who performs the final task of blending them seamlessly into the background plates. The final composite is then rendered out, and that is the image that goes finally to the film recorder.