June 18, 2011

Tessive's Time Filter Brings Film's Dreamy Motion Rendering to Digital Cameras (and Eliminates Rolling Shutter Too)

It's been one of my pet peeves ever since the advent of digital cinema cameras: at the same frame rate and shutter speed as a film camera, digital sensors looks more stroboscopic. There's more jutter and I find it distracting. For a long time I couldn't explain what I was seeing until Adam Wilt explained why in a piece about RED. Now Adam's the first to bring my attention to a new product called a Time Filter from Tessive. If you want a smoother shutter rendering on your digital footage, this is a serious innovation -- brilliant, in my opinion.

At the standard shutter speed of 1/48th of a second on a 24p shoot, digital cameras have motion rendering that bothers me as too stroboscopic. See if you can spot the difference in these two shots:

So what's the difference between digital and film cameras in this regard? Once again I'll turn it over to Adam:

Film cameras usually use a mechanical shutter at some distance from the film plane; their shutters do not sharply and suddenly expose and then cover the film, but perform a soft-edged, “fade-in, fade-out” exposure due to their defocused, penumbral sweep across any given point on the film plane. 24p digital cameras with 180-degree shutters (or 1/48 sec shutters, if you prefer) tend to show more visible “stuttering” or “judder” on motion than film cameras do with the same shutter angle or time; the sharp-edged motion blur is the culprit. Some shooters try to compensate by opening the shutter longer, say, to 1/32 sec or 270 degrees. This increases the blur duration, partially (but only partially) compensating for the sharp edges, at the expense of increased blur overall.

I've experimented with slower shutter speeds in the past, and it works pretty well on many shots. But I get the feeling the difference is more pronounced for me than it is for most. Maybe I just have sensitive eyes. But I also think part of the magical nature of 24 FPS film is the smooth, penumbral motion rendering, and that has been lost with the switch to digital cameras. From their tests, it seems that the Tessive Time Filter brings it back. Here's how it works:

Yes, it's a drop-in matte box filter. Brilliant. Okay, so if this thing is so great, why aren't we all using it right now? Drawbacks include a loss of two stops of exposure (though with today's super-sensitive 800 ISO cameras, some might embrace this exposure loss), and you need a genlockable camera -- a DSLR won't work. Oh, and this first version costs $14k. Check out Adam's full article below (it's tucked into a post about Cine Gear, about which I'll have more soon), which is a must-read for camera nerds.

Tessive was demonstrating the Time Filter at this year's Cine Gear, and while I'm far less concerned with wagon wheels turning backwards -- I'm chiefly interested in the motion rendering on everyday shots -- FreshDV got a nice demo from them, which shows that not only does the Time Filter demonstrate better motion rendering but in fact eliminates rolling shutter problems, because it's effectively a mechanical shutter:

Tessive

At $14k you're not going to be buying one of these unless you're a rental house, but that's exactly what it's designed for: renting. My own forthcoming feature is going to have a very active camera (at times), and I'd love to get a more filmic motion rendering without the drastically increased (and unrealistic) budget that would come with shooting on film. Great work, Tessive.

Links:

Your Comment

18 Comments

I am not sure you quite understand what Tessive's time film is doing. You speak a lot about how tessive removes the digital strobing effect on digital motion picture cameras. The reality is, Tessive's filter performs the same on both and digital and film camera. Digital and film cameras perform identically in terms of strobing...unless you are using the digital camera incorrectly. They both expose the frame the same amount and are either 100% exposed to light or 0% exposed to light. There is no gradation in either format. A film frame's exposure does not "fade down." In fact, old ASC manuals have a chart with proper panning speeds to eliminate strobing on FILM cameras. Even before digital cinema really existed.

See this page for more specific information http://www.tessive.com/home/time-filter-faq-1/time-filter-technical-expl...

The article you linked to from Adam is also incorrect.

What tessive's filter does do, is help eliminate rolling shutter. Please get your facts straight before spreading around incorrect info across the web!

June 19, 2011

0
Reply
Manny P.

I feel even though Ryan is misstating a few things in this article that there is no reason to be so abrasive about correcting him. Yes, film cameras do also strobe and this will do nothing for how sensors record information but as things have been brought to light on this blog in the past, the erronious statements were corrected with out ego, since this is one of the few blogs that realize that people can be wrong.
I have no problem with you correcting these facts, as I was going to do the same, just leave all the over active angry bullshit out. It's really not needed.
I am also excited about how this will "change the game" for all of cinema and I hope it leads to cameras with this as the standard for the shutter operation in the future.

June 19, 2011

-2
Reply
Darrell

Thanks for the comment, Manny. If Adam Wilt is wrong, then I feel no shame for having followed his lead. However, we're not the only ones who feel that digital sensors yield more stroboscopic footage than does film:

http://reduser.net/forum/archive/index.php/t-58298.html?s=cd43973b989b1e...

If the "penumbral sweep" of a mechanical shutter is not a valid explanation, can you think of anything that might explain this perception (that others share as well)? I'll gladly correct the post but simply saying "that's incorrect" doesn't give me much to go on with offering a valid explanation as to why people (other than myself) also feel this way.

June 19, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

Adam Wilt is correct about penumbral blur "softening" the shutter in film cameras. Since the shutter has some distance from the film, the edge is necessarily slightly out of focus with respect to the film plane. Different camera configurations will have slightly different blur. The Tessive filter takes this to a much more engineered, mathematically defined level. While film cameras do have some blur at the edge of the sweep (so the transition from 0% to 100% transmission has a slight ramp to it), the Time Filter provides a designed curve throughout the entire frame time.

When we look at the frequency response of a camera system, we do it analytically (the transform from shutter response to frequency response is pretty easy to calculate), and we also do it empirically with a measurement system we've built for this purpose. Mostly it makes pretty graphs that I enjoy looking at, but it really tells a much more thorough tale of the way motion is represented. The difference between shutters with a slight blur and sharp digital shutters is difficult to see, since in the main the frequency responses are the same. The falloff happens in the extreme high aliasing frequencies. But when there's blur, there's a frequency difference.

I hope that helps. I'm very thankful for everyone who has commented here (and other places.) It's very exciting to see people with such heartfelt interest in the science and art of cinematography.

Tony (the guy in the white lab coat)

June 19, 2011

2
Reply

Thanks for stopping by, Tony. As you can tell from the post I'm excited about your Time Filter and look forward to the opportunity to try it on a shoot.

June 20, 2011

-1
Reply
avatar
Ryan Koo
Founder
Writer/Director

Sorry everyone...didn't mean to come off as rude! Merely passionate.

June 20, 2011

1
Reply
Manny P.

Camera manufactures should just build these right into the camera as a shutter: Eliminating rolling shutter problems, creating a more organic "film look," keeping the awesome low light ability, and containing it in a small package.
Yet another dream for the perfect camera I guess.

June 19, 2011

3
Reply
Shii

Lemme know when we can get this effect with a plugin, so i can stop dreaming! lol

June 20, 2011

-1
Reply
alexander miller

June 21, 2011

-1
Reply

Wow, thanks for that -- 2 years ago!

June 21, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

I wonder how red's new "magic motion" will compare to this.

June 23, 2011

2
Reply
Matt

This is MASSIVE for film makers (who can afford such a unit). The artifact it is eliminating is very annoying even in big hollywood productions. Any film maker whos film has a budget for this kind of unit would be crazy to ignore it.

June 23, 2011

1
Reply

$14k??? Am I missing something here or is this just an LCD hooked up to a controller?

This could be made for under $50 with an Arduino and an LCD screen? (unless I am totally missing something here? Which I very well may be.)

I will say though, this does give me plenty of DIY ideas for a Digital Vari ND Filter...

I think the interesting thing will be if camera makers adopt this as a built in method of exposure over the digital "on/off" method currently used. It would certainly be cheaper for them to implement in camera, with all the syncing and what not.

June 23, 2011

1
Reply
O'Ryan

You're missing something. It's a translucent LCD. Like something you might see in a digital watch, perhaps. And hardly common, especially coupled with the the need expertly sycn'd to the millisecond with the camera's shutter.

June 24, 2011

0
Reply

even in this case that cant be that expensive to make using arduino and i think an implementation is worth the effort, even if this results in a vari ND as O'Ryan says
O'Ryan, do you have any experience with arduino?

June 25, 2011

-2
Reply

O'Ryan isn't missing anything. This was a cheap experiment and shows tha conpcet work quite well, synching is just a manner of fine tunning the electronic shutter, no biggie and certainly doesn't merit the 14k price tag. http://www.vimeo.com/5976527

June 25, 2011

0
Reply
Alfonso

14k sounds waaay too expensive for such a solution. But since it doesn't work on DSLRs I'm sure it's not meant to target indie filmmakers with a very low budget.

Here's an experiment that's 2 years old and basically shows that Oakhurst is talking about. A very cheap and effective solution. I would never pay 14k for something like this, especially after seeng that experiment.

Rolling shutter experiment: http://www.vimeo.com/5976527

June 25, 2011

0
Reply
Alfonso