I'm not sure I understand how you can get those numbers using a camera that only shoots 1000 fps. Each frame would be 0.001 seconds long, so how can you determine that ball-ball collision (for example) is only half a frame long?
For example, using a 1000 fps camera, if a certain tip-ball collision had a duration of 0.0011 seconds, and the impact just so happened to occur right before the end of a particular frame, it could actually appear to be in contact for three frames. Repeated video tests would help, but there's no way to measure the duration with a precision of less than 0.001 second (1 frame). We'd simply need a higher frame rate to be more precise.
I'm not saying you're wrong, though. Maybe you used a different method to measure the durations?
As far as the resolution, yeah, I hear ya. At 1000 fps, my current cameras' resolutions are crap indeed.
There are a couple of lengthy posts earlier in this thread that discuss this camera, as well as a brief discussion about alternative high speed cameras. You can also check it out at www.edgertronic.com . Specifically, here are some of the numbers at fairly high resolutions:
1280x1024 @ 494 FPS
1280x720 @ 701 FPS
866x672 @ 1000 FPS
640x480 @ 1849 FPS
1024x128 @ 4658 FPS
The very wide aspect ratios (like 1024x128 @ 4568 fps) would be great for filming cue/tip action at very high speed, since the motion is (basically) linear. We really don't need a 16:9 aspect ratio when a "strip" of video would suffice. I'm sure you know what I mean from your experience with your 1000 fps camera. This camera just does the same thing at a higher resolution and framerate.
One of the things that has been frustrating for me is not being able to get helpful slow motion video of the transmission of forces through the balls as they sit on various template racks (Accu-rack included). At 1000 fps, in one frame, the cue ball hasn't hit the rack yet, and in the next frame every single ball is moving. 10,000 fps or so would likely give us a lot of insight into what's happening and why (as well as how you might be able to improve the Accu-rack even further if possible). Just for fun, I'd also like to teach folks how to make the Magic Ball Rack do a little wave during the break. It's something that I figured out while filming hundreds of MBR break shots. I can't think of any use for it, but it probably affects the break in some as of yet unknown way...
Anyway, to get 1000 fps at 1080p, you'd need a camera from Vision Research's Phantom line, which starts in the tens of thousands of dollars (or rental for $5000+ per day). The Edgertronic camera that we're considering is capable of 700 fps @ 720p for just under $6000.
Twixtor and similar interpolation programs do a decent job of emulating slow motion, but only with relatively slow movements (like someone jumping up and down). Think of a firecracker - filmed at traditional 30fps, it's there in one frame, and completely destroyed in the next. How would Twixtor "know" what happened in between? Filling 999 frames with a gradual "morph" from unexploded to exploded simply doesn't work. Anything it could possibly generate is simply make-believe - it will never be the same as a true high speed camera. That being said, as long as you don't push it too far, Twixtor and others do a pretty decent job. As long as you don't go too far over about triple (or maybe quadruple) the original framerate, it looks pretty good.
I'm not "against" Twixtor or anything - I've even used it (and others, like VirtualDub/AVISynth) to "turn" a 240 fps video into an emulation of nearly 1000 fps, and it looked okay. But that was just a video of a cricket jumping away from a tarantula - again, a relatively slow movement when compared to tip and ball interactions.
Thanks,
-Blake
Last edited:
