Ben Syverson

Negative latitude is in the other direction—highlights. Pushing doesn't give you more shadow detail; just brighter midtones and highlights. So the further you push, the higher the contrast. If you're going for a high-contrast look, it might be just what you're looking for. If you're just looking for more speed, it's probably simpler to go with 250D.

Well, I got the part today, and it looks like everything fits! The FDM version couldn't be manufactured because I made the wall thickness too small, but the MJM looks fantastic. I'm blown away by the detail they were able to achieve—the M42 × 1mm thread was smooth and worked the first time. Anyone who has had a 3D print made knows how crazy that is. Now I just need to shoot a test roll to confirm infinity focus.

Phil, Were you able to establish a good workflow for dealing with the scans? Let us know how the project is going!

Nice work! You may think about investing in a tripod—the shakiness of most of those shots was distracting to me. And the reveal shot with Batman could have used a little backlight to create some separation. As it is, you have a black costume against a black background. Keep it up!

Guy, this is another phenomenal post. I've been saving these to my hard drive for future reference. Fantastic information in there.

I think the reason most people get confused is that there's no consensus on what an "equivalent" lens would be. So first, you have to come up with a personal criterion. For me, the most important thing is horizontal FOV. Some people prefer diagonal FOV, because different formats have different aspect ratios. That has never made sense to me. Well, different strokes. Once you know your criterion, the calculations are exceedingly simple. Just take your focal length and divide it by the relevant dimension of Sensor 1, to get a format-agnostic focal length multiplier. To find the equivalent on any format, multiply the constant by the relevant dimension of Sensor 2. You have to decide if you want to compare horizontal, vertical, diagonal, etc. For example, if I like 28mm lenses on my full-frame Canon (36mm wide sensor), I would do: 28.0 / 36.0 = 0.7777. If I want the same exact HFOV on S35 (24mm wide sensor), I just do: 0.7777 * 24.0 = 18.6666. I could also find the equivalent for my 8x10 view camera (254mm wide sensor): 197.555, or about 200mm. If you always convert the same two formats, you can consolidate this process into a simpler "crop factor." For an HFOV conversion from S35 to 135, it's 1.5. Just multiply or divide by 1.5 to go back and forth. People get confused and think "a 50mm is always a 50mm on every format." Well, that's true, but meaningless. FOV is what you're interested in comparing, and FOV for the same lens changes with the format. Eyeballing your stills, I would roughly guess at this progression (all focal lengths assuming S35 format): 1. 75-90mm 2. 50mm 3. 30-35mm 4. 24mm It would be handy if you could pick up a Canon DSLR with a cropped sensor (Rebel, 7D, etc). The horizontal dimension of its sensor roughly corresponds with the horizontal dimension of S35, so it could be useful for previewing focal lengths.

I heard that Yogi-isms were mostly written by the sports writers of the day... Hence the classic Yogi-ism "I really didn't say everything I said."

That's... bizarre. I can't imagine what the benefit would be. It would reflect less light than a white card, and could lead to strange colored highlights on anything partially reflective.

Digital will always look great... until you have highlights. Negative film can easily roll off 10 stops of highlights, but digital must clip to ugly pure white eventually, because the sensor is linear rather than log, like film.

Heh heh...You know, I can't get too worked up by anyone who quotes the one of the greatest movie lines ever in their sig. Somehow, an exploding movie fits right into the Donnie Darko world, too. :)

"Naughty" according to whom? As a software engineer, I know how impossible it is to completely remove all bugs for every edge case. That means this software would wipe out a certain percentage of innocent hard drives. Why not just ship the movie with some plastic explosive that's triggered when the user opens Bittorrent? Would that increase sales?

My humor detector must be off today. Are you being serious? Because I'm sure it's obvious to everyone that this would lead to fewer disk sales, not more. I would switch to downloading illegally, because it would be safer than risking my production system. If my computer was wiped out because of the twisted logic in some movie studio spyware, I can't even tell you how many lawsuits I would file.

I haven't bought a movie in years, because there's no acceptable way for me to do it. I still go to see movies in the theater, and stream via Netflix, but I'm quickly losing the desire to "own" a movie. At this point, I'm not going to buy a DVD, because it's not HD. I'm not going to buy a BluRay, because my 5 year old projector is "too old" for the DRM, and my MacBook doesn't read the disks. I'm not going to buy the movie from iTunes, because 1080 is cut down to 720, and there are no bonus features. I would be happy to pay $15-25 for a download if it was on par with the BluRay. It would cost the studios less than pressing a disk.

In the optical days, it was called "step printing." In fact, the famous "slow-mo" scene in Reservoir Dogs was not actually shot at high speed, but rather step printed. Personally, I think step printing is the same category as freeze frames for pure cheesiness. But sometimes cheesy can work to your advantage.

Man, that's outrageous. Putting linear values into a DPX is like charging an electric car with a gas generator... Missing the point.

These days, 16mm high speed cameras cost less than 1000' of 16mm stock, and can go well beyond 200fps. I would just buy a camera and sell it when you're done with it.

I doubt very much he's been given linear DPX files, which would be close to useless. What is much more likely is that the DPX viewer he's using is linearizing the DPX (log->lin), but not gamma correcting it for display.

Keep in mind that the films are heavily augmented with CG and aggressive DI treatments. What you're seeing is about as far from a "straight" crop from S35 as you can get.

I was just getting those images in the ballpark... Literally about 60 seconds on each using Curves. Here are the Curves for the last image I posted: The only way I've messed around with DPX files is from within compositing apps. Personally, I would probably batch convert the DPX to low-contrast 16 bit TIFFs, then create QuickTime movies from the sequences. I'd cut in Final Cut Pro 7, applying color from the timeline. Knowing me, I'd probably write the DPX converter and the color tools too. I always make things more complicated.

$10.50? You're lucky! It's going to cost me $17 at the IMAX or $15.50 at the normal theater.

I'm doing two—one is FDM (ABS, coarse resolution but very strong) and the other is MJM (translucent acrylic, extremely fine resolution). I'll use the MJM to verify the design, and hopefully the FDM will work for actual use. FDM is extruded in 0.2mm layers, so I'm not sure how well it will render an M42×1 thread mount, but hopefully it's close enough that screwing in a lens will force the threads open. Either way, the front may need to be sanded slightly to get the flange distance just right...

I'm slowly teaching myself CAD, so I thought I would tackle an adapter to convert the Eyemo's Eymax mount to M42. Aluminum versions do exist, but they're exceptionally rare. So, I just sent this off to get 3D printed, and if it works, I'll make sure it's available to anyone who wants one!

Another...

Yeah, the stock is fine. The problem is with your DPX workflow. The screenshots from the lab confirm it... I spent about 1 minute on a couple of those screenshots, and came up with this:

Looks slightly out of focus to me, but grain-wise it looks about right for 200 speed film.