Ben Syverson

Post an image... 16mm at 1080p will definitely have some visible grain, but it's impossible to know how much is too much without seeing it.

Well, setting aside the fact that nothing in the natural world emits a pure spike at 400nm, it will result in an a dark, dominant blue record, with differing amounts of red and green depending on the recording medium. Viewed on a display with decent gamut, your eye will readily accept it as a deep violet. Film or digital. I'm having a hard time understanding how spiky lighting spectra relates to this discussion. It's true that some saturated colors may appear dramatically darker when illuminated with a spiky source such as LED or Fluorescent versus broad spectrum daylight. However, neither film nor digital have very spiky spectral sensitivities, so they aren't exacerbating the problem. The problem in that case is with the light source.

What Karl just posted is really the key. Is it possible to 100% reproduce the spectral energy of a color, using only RGB/CMY? No. Take the example of a cup full of ultramarine pigment, illuminated by sunlight. Even with such a simple and pure setup, the pigment will reflect distinct spikes in different areas of the spectrum. Any three component imaging system decimates the spectrum to just three overlapping zones. So whether it's film or digital, the reproduced color will be some mix of R,G,B instead of that original highly complex spectral "signature." In other words, color reproduction is a hack. It's a trick. It's lossy compression. Whatever you want to call it, it works because our eyes operate in a similar way, with three broad zones of spectral sensitivity. Our eyes can't tell the difference between the rich spectrum reflected by a red shirt in daylight and the same color simulated with different percentages of Red, Green and Blue lasers. Our eyes aren't spectrometers. So that's why this "you can't reproduce violet" nonsense is such nonsensical nonsense. First of all, there is no color standards board to specify what "violet" is. What are we even talking about? A pure spike in the spectrum at 400nm? Are we arguing that the spectral response of a CCD/CMOS is closer to the eye's spectral response than what can be achieved with color negative? They're both imperfect. The question itself is ill-defined, which is why this debate is so inane. The bottom line is: film and digital can both capture "violet," however we define it. The exact mix of RGB/CMY produced will be different for every emulsion or sensor. We can argue till the cows come home about which is more "accurate," but if we're really getting this pedantic, they're both inaccurate. Yet they both can produce a color which is perfectly acceptable to our eyes as "violet".

You were right in the last thread, and you're right in this thread: violet CAN of course be reproduced via film.

Here is that generic dye gamut overlaid on top of the Adobe RGB gamut. It speaks for itself.

And as you can see, violet is covered well—look down at the 400nm corner. So... I don't understand what this debate is about. Very pure violets can be captured and reproduced on film.

I have to call bullsh*t on this. Here's your eye: Here's Portra 400 (similar to Vision stocks): According to Kodak, color negative film is far, far more sensitive to "violet" than your eye. Please do not spread misinformation about film.

Look at the spec sheet for any color negative film. There's plenty of sensitivity in the violet range of the spectrum. Saying "nobody has ever seen violet on the cinema screen" is totally absurd. What about The Joker in the Batman movies? What about that scene in The Game that's shot entirely in black light?

It's called a "sunstar," and it's caused by the shape and number of the iris blades when stopped down. Typically you'd see sunstars starting at about ƒ/8, becoming progressively more prominent as you stop down. But it varies by lens. If you stop down a lens and look at the aperture, the more "spiky" or angular it looks, the more apparent the sunstar will be.

Karl is right... saturation IS color contrast. If you take a low-saturation image and increase the contrast, the saturation will shoot up. Bleach bypass has high overall contrast yet low color contrast, which is why it looks so distinctive. But no color negative film is anywhere near that divorced from reality. In fact, aside from slight differences in characteristic curves and even more slight differences in spectral sensitivity, all modern color negatives behave pretty similarly. At the very least, none of them are so extreme that they can't be acceptably matched to each other by a skilled colorist. That includes ECN2 to C41.

I realize the choice of Super 8 is mostly an aesthetic rather than budgetary one these days, but I'm wondering how people keep costs down when using this format. Using Pro8mm as a reference, 10 rolls from stock to scan is $672, or $67 per roll. By comparison, 35mm shorts can be had for 15¢/foot, processed for about the same, and scanned for 32¢/foot, for a total of $62 per 100' roll. I realize that a 50' load of Super 8 runs longer than a 100' load of 35, but have we really reached a point where it's literally cheaper to shoot a daylight load of 35 than Super 8?

Hi everyone, I'm looking to pick up a few more 35mm 100' daylight spools with cans. If you have some, shoot me a PM or email (firstname @ firstnamelastname.com). Thanks! Ben Syverson

Needs a better bitrate, or else it will be useful to exactly no one.

I'm 100% with Brian Rose on this... I imagine that same conversation in my head when I watch one of these scenes. It's like being bludgeoned over the head by the director and DP, both screaming "THIS PART IS INTENSE! IT IS VERY EXCITING!!" More often than not, it's just a complete turnoff. Motion blur is not exciting to me. Nor is using a 45° shutter for no reason. Saving Private Ryan was an interesting experiment, but it's been done to death overy the past 12 years. Now it's just a boring shortcut. If your camerawork and editing are more dynamic than your action, you have a problem. This is not restricted to shaky cam; I see it all the time, especially with young directors. It'll be a wide crane shot of a city that transitions into a dolly shot through a doorway and then it goes into a close-up of a flower. Stuff like that makes me roll my eyes and think "oh boy, the director was real excited about this shot." The subject is boring, so the shot is boring. Overly elaborate camerawork is a band-aid that only excites other filmmakers, not the audience.

The 35mm/2 is an old lens, but it's still good -- especially for video, where lens sharpness is not as critical... I'm a big fan of those less popular lenses, like the 28/28... They tend to be small, light and incredibly cheap. The only downsides are less "pro" build quality and slightly slower stops.

The stills and video are essentially identical between the 7D and 550D. The 7D gets you better frame rates and a more professional body. Personally I would go with the 550D and put the monetary difference into lenses... The 5D is a great camera, but less flexible for video. The DOF is extremely shallow, and you don't get the 720 60p mode.

These cameras are terrible for charts... You really need to shoot with them in real-world situations to see if they would fit with your style. Personally, the aliasing really really bugs me, but it wouldn't bug me enough to go with HDV over DSLR. But that is just one random guy's opinion...

Brian has a good point. After years with a terrible tripod, I invested $500 in a carbon fiber tripod that I can fit in any bag (the Gitzo 1550T) and a real fluid head (under $300). And both of those purchases were on the extreme LOW end of what I should have spent. Honestly, with this hobby (if it's a profession, you're renting), there is no limit to how much you can spend, but $1000 on great camera support will benefit you more than $5000 on a camera.

My guess is that DSLRs will spell the end for DOF adapters like the Letus... It's hard to imagine people continuing to pay $1000 and up for adapters when a $800 Canon will get the same "look" without the weight, bulk and hassle. The DSLRs are not as sharp as dedicated video cameras, but the success of adapters have already shown that DOF is more important than sharpness to many people. While it has its good points, HDV is a dying format... HDV tape drives are expensive and finicky mechanisms, so manufactures are really pushing for flash memory formats. If you're really hell-bent on investing in HDV, I would spend as little as possible, as that gear is basically already obsolete and will only lose value. I definitely would not break the bank for the XLH1 or its expensive proprietary glass. If you put a little of that money into a DSLR and the rest into some good glass, you'll be investing in something you can use for years. Just swap out the DSLR every year or two with the latest and greatest, and you can take your lenses with you. Lenses retain their value much better than cameras... Anyway, that's just my 2¢ -- maybe an HDV defender will pipe up now :)

My advice is to get a Canon 550D and use the rest of the money to actually make a short. People get so caught up in gear expenses. The world is full of Starbucks employees with £3000 cameras but no time or money to shoot anything. Once you start getting paying work, or investors for a feature, you'll be renting cameras, not using a consumer HDV camcorder. So get a camera that's cheap, will let you shoot some demos and shorts, and has multiple uses (it takes pretty good stills!). And hey, however ill-advised, people are shooting features on these things. Lena Durham's "Tiny Furniture" recently made a splash at SXSW, and it was shot on a 7D -- which is essentially identical to the 550D when it comes to video.

Paul, I'm have no real EE experience, and I haven't read all that much about USB 3, but my initial reaction would be "hey, 8 wires is 8 wires!" There's nothing mystical about wire, despite what audiophiles say. That said, I would keep the cable runs short -- the longer the cable, the more resistance, the more you're taxing the signal. Of course, for a render farm, you may only need to go about 12-18 inches at a time! Can you catch me up on why you're running a render farm? I'm guessing it has something to do with processing RAW frames from your scanning rig? I'd be curious what kind of processing you're doing... I've done a little work on GPU-accelerated film tools. Things like dust/scratch removal and color correction. On my laptop, they're realtime at 2K and near-realtime at 4K...

You almost certainly don't want to do that... From what I remember of my old K3, the spinning mirror would keep you from getting that C-mount lens close enough to the film plane.

If you're talking about compressed bitrate vs color sampling "rate," then it's truly apples and oranges. You could have a terrible-looking 4:4:4 image if compressed down to 1 Mbps, and you can have an amazing 4:2:0 image at high bitrates, especially since we don't know what the frame size is. We could be talking about 4K or HD; who knows. When it comes to bitrates, higher is generally better, but there is a point of diminishing returns. You need to test, test, test to figure out what that point is. If you're doing talking heads on a static background, you might find that 17-25 Mbps is fine. If you're shooting a surf movie, you might realize you need 100 Mbps or more. When quality is more important than price or time, you can always shoot a raw uncompressed format, such as 35mm film. However, it's not clear from your question whether you're talking about acquisition or the bitrate / sampling scheme of your DI or transfer... Perhaps you can elaborate?

Unfortunately, there are none. There's the 14mm/2.8, but it's a premium lens and not cheap at all. There's the 20/2.8, which you can find for $300-$350, but at 32mm equivalent, it's not that wide. Honestly, if you have the kit lens, I would use that at 18mm for the wide end. If you absolutely NEED a true wide, there's the Canon 10-22, Sigma 10-20, etc. Unfortunately, they aren't all that cheap, and the stops aren't great (usually around f/4). Wide shots won't look all that great anyway... That deep DOF and wide FOV grabs a lot detail, and tends to expose the aliasing artifacts that dog these cameras.

Samples do not equal lines. Pixels do not equal detail... Once you're able to fully separate the sampling rate from the signal, you can start to figure out what an imaging system can do. My advice is always to take the lowest lp/mm of the system (lens, scanner, sensor, etc), multiply it by two (Nyquist), then multiply it by the sensor dimensions in mm. You'll get the "real megapixels" of the sensor. For example, color negative tends to resolve around 60-70 lp/mm with decent contrast, your lenses can do better, and a good 4K scanner can most likely do even better. Assuming 65 lp/mm, a 2.35:1 extraction of Super 35 would give you 24.89 x 10.6mm, for a total of 1617.85 x 689 line pairs. To resolve such a frequency, you would need twice as many samples, which gives you 3235.7 x 1378 pixels, or 4.4 "real megapixels" on an ideal sensor. That number tells you the actual amount of detail the system can record. Because of Bayer sensors and AA/Optical Low Pass filters, you usually need 1.25 - 1.5X as many pixels as a theoretical "ideal" sensor to reproduce a frequency. So, taking the optimistic 1.25X multiplier, we know that we would need at least a 4044 x 1722 pixel sensor to reproduce the amount of detail on that S35 section. But of course it is not so simple. Grain moves around, which is good (error diffusion in spatial sampling). Pixels are fixed, which leads to all kinds of problems (aliasing, interpolation errors, various artifacts). Resolving the signal and having the signal look "good" or "natural" are very different things.