Reservations to close...

One of the things I'm interested in is the effectiveness of speed ramping software in post, partially for some of the reasons mentioned above.
If you have a very limited selection of lenses, and you're working with a small light package, and space constraints, want to keep a certain aesthetic and resolution etc... How good is the speed ramping software compared to in camera change in fps ? I know it won't be as good, but is it comparable ? And wouldn't 4K REDCODE RAW give you more information for those extrapolations ?

Out of curiousity, what is the best speed ramping software on the market today ?

I've done in camera S16 before, but I had a wider selection of lenses and had a DP do the calculations.
But, since Malick operates his own camera and knows the DP side in and out, then can I do any less ? :beer: :laugh::beer:


RED# wish I was higher
 
D_and_G said:
How good is the speed ramping software compared to in camera change in fps ? I know it won't be as good, but is it comparable ? And wouldn't 4K REDCODE RAW give you more information for those extrapolations ?

RED# wish I was higher
Click to expand...

Standard speed ramping in post is inferior to an in-camera intervelometer for two main reasons:

The first reason is steppiness. Because an intervelometer will mechanically ramp the camera motor (along with the aperture), the curve from slow motion to fast motion will be smooth and the viewer will not notice any steppiness. A standard post solution would be to shoot slow motion in camera, and then apply a speed ramp curve (or vari-speed) which basically removes or drops frames from the native footage to arrive at the desired speed. This can result in distinct pops or stutters or steps which are very noticeable in things like camera pans and walk cycles, That said, certain speeds produce better results - 200% works great, 150% is not bad and 125% is perfect for television because you just remove the 3/2 pulldown and your left with progressive frames again (if shot at 24fps)

The second problem with post ramping is shutter speed - or in post we'd call it motion blur. A post vari-speed of 200% will result in half the motion blur of the native footage - or rather the same amount of mo-blur per frame, but not enough for its new speed. This is similar to the look of a narrow shutter angle on a 535. If you shot film of a city night scene at 6-fps you would get some pretty cool in-camera streaks / long exposure traffic. Compare this with shooting @ 24fps and vari-speeding in post by 400% - the traffic would be too sharp. Post vari-speeds are often very easy to spot for this reason. Watch a movie like "Lock Stock and Two Smoking Barrels" and you will find shots filmed at 300 fps with Photo-sonics. They speed-ramp the hell out of it in post and the resulting images are unnaturally sharp with almost no motion blur. We did this on the movie "The Cell" which shot stuff at 1200 fps - crazy. Photo-sonics makes a hell of a noise!!!

Now then, on to the good (or slightly better) news. There is a sexier post solution which was first used in the famous 'bullet time' sequence in the Matrix. It turned out that the shot of Neo dodging bullets was not long enough (not enough Nikons!) so they had to slow it down in post. Previously a step frame (repeat frame) or dissolve frame approach was used but in this case they created new digital frames (inter-frames) by morphing from one frame to the next. A morph from frame 1 to frame 2 would create a brand new frame in between (frame 1 1/2) - new pixels, new composition, new body position. This is now pretty common software. In discreet Inferno it is called "motion estimation", and in AVID it is called "fluid motion" and several other platforms and several other plug-in companies have their own versions.

While all of this sounds great it is not a perfect solution by any means. Often "motion estimation" has a very hard time rebuilding the pixels for the new inter-frames, and the result is unwanted image 'tearing', bending and dissolving. It is especially problematic when the scene changes rapidly from one frame to the next. It will work great on something organic and soft like clouds but go ahead and try it on a medium shot of a guy walking in Times Square - all of the background buildings and signs will have a weird morphing effect (around the guy's body as he walks) because the software can't interpolate between an obscured sign and a unobscured sign. Plus it's expensive (i.e.. very slow renders). At the end of the day you run it through the software (or plug in) and then spend hours per shot fixing the problems.

I just finished the vxf on an Acura spot which required comping a 360 degree rotating shot of a car (shot green screen) into 4 separate environments. The production lacked the budget to shoot each environmet on film with a proper circular dolly (night shoots, big sets, couldn't get the stop) so they instead decided to shoot digi-stills - approx. 24 (pseudo-dolly) stills per environment. The plan was to motion-estimate these stills in post and slow them down to 10 second shots (300 frames @ 30 fps). Long story short, it didn't work. The software couldn't handle the architectural detail from the 4K stills, as well as the fact that the panning / dollying revealed lots of new info in each new frame. We ended up using photogrammetry (kinda) to completely rebuild each environment in 3D and did it all CGI.

Also I don't think Red's sharpness and resolution will help with motion-estimation software - I think it will hurt. Just more pixels to tear and rip and ultimately fix.

So..... in summery I think the quality of post-ramping still does not match that of in-camera ramping. The post-route just offers you more options as you are not "locking in" on set. If I was really nervous about a particular shot (explosions, big stunts, one-takers) I would shoot in 2k windowed @ 120 fps - that's a no brainer. Still it would be great for the Red to incorporate some form of intervelometer, one which works mechanically (ramps shutter speed, capture rate & aperture compensation) rather than digitally (dropping frames). Dare to dream, eh!
 
:)
Thanks for the post Boothba ! Very informative.

I remember we did a shot using a 180 degree shutter, 8 fps, then reprinting 3 times to get the 24 in the optical printer, for that motion blur you were talking about . And the first time you speed ramp your film camera it can be a little scary (especially when your watching your counter). Multiple takes can get expensive. But, this digital acquisition is a whole different story.

Thanks for laying it out. I guess I'll do it in camera on the RED, or at least get as much of the way there as possible. I guess it will also depend on how well 2K scaled or 4K REDCODE cut with 2K windowed @120fps cut together...

Dare to dream. That's a good RED slogan. :beer:


RED# wish I was higher
 
D_and_G said:
:)

And the first time you speed ramp your film camera it can be a little scary (especially when your watching your counter).

RED# wish I was higher
Click to expand...


Yah, if - big if - the Red One offers us proper speed-ramping it should at least be much safer than film to ramp in-camera because you will able to scrutinize high-res playback before moving onto the next setup. Post ramping software will always get better and post solutions are certainly an option - provided you have the some time (and $$$) in post. You might try to do a few takes of ramping in-camera and then shoot a 'safety' at 120 fps 2K windowed as a backup.

With regards to matching 2K windowed to 2K scaled, keep in mind that there is a pretty good post solution too. If you roto the subject (actor) and throw a defocus filter on the background, you can sorta 'fake' the shallow DOF of S35mm 2K scaled. It's not perfect, but with some effort you can get it looking pretty good.

You also might want to send the editor a bottle of scotch along with the rushes with a big note: "please don't butcher my footage". I find commercial editors in particular are always speed-ramping because they have to fit the content into exactly 30 or 60 seconds. I work with an offline editor who regularly cuts 31 or 32 second spots, and asks us to speed-ramp the finished piece back to 30 seconds in the online. He even speed-ramps the audio in dialogue sequences. And it works.

Hey D_and_G, I just noticed you're a Toronto guy. I work at Technicolor on Ontario street. I just got Red #478. Perhaps we could swap notes when our cameras arrive!
 
Jannard said:
We have all the info we need to make good production decisions. It is likely the program may change a bit and we want to lock down the reservation holder's program. We think that a bit of notice is only fair.

Jim
Click to expand...

Is program a euphemism for price point?

I'll re-read the thread and see if I missed something.
 
boothba said:
Hey D_and_G, I just noticed you're a Toronto guy. I work at Technicolor on Ontario street. I just got Red #478. Perhaps we could swap notes when our cameras arrive!
Click to expand...
Sure :) Sounds good. And if mine arrives before yours (since I'm in the 200's) then you'd be welcome to check it out...Your notes would be more valuable than mine.:beer:


RED# wish I was higher
 
Yes, Toronto as well. Somewhere in the middle of both you guys (in the 300's). Would be good to source each other out should we need second cameras on shoots etc.
 
Jim,

Is it possible to guarantee the 300mm lens reservation now and to schedule its delivery to late 2007 when the 18-85mm lens will also deliver ?

That is, can we have the delivery of both lenses at same time? When your two RED lenses will be available, without compromise both uses.

Grateful for your help.

Best,
Emanuel
 
You must log in or register to reply here.
Back
Top