If you’re using film to gain insight on your systems,
are you filming in slow motion?
And if so, what camera are you using?
Level 1 is something like an iPhone camera,
level 2 is a handheld camera, that is capable of a little more than the iPhone,
level 3 is a high frame-rate slow mo oriented camera.
I wonder if to gain more insight, it is worth it to spend money on an expensive camera to shoot high speed operations in high resolution slow motion?
I’d suggest a newer iphone, perhaps a pro max variant, as their video qualities are high and the slow-motion video will be sufficient for all but the fastest movements on previous years’ robots. If you still struggle after that, maybe consider how much a camera with the capabilities to shoot at the framerate you require would cost, and if the need for slow motion capture is that high in comparison.
Some android phones are slo-mo capable as well. Also some action cameras have a super slomo option. Definitely look around and have people check their smartphones first.
Stuff like a phantom cam is way too expensive to justify what it will be used for. In many cases of you also put a good backdrop or indicator system you can still get some of this info even with full speed and then frame by frame analysis.
The above picture is a large scale test by the mythbusters to capture a sneeze. That’s a much faster and smaller object than a note. What’s useful is they taped out every X feet to be able to measure the feet per second of velocity and the travel distance. You can change the backdrop to measure any unit size and match the scale of the item (biological emissions) you are trying to film
I looked into this recently. If you want faster than 240fps, choices get expensive and limited. There’s a couple affordable -ish Sony cameras that will do 960fps. There are some demos on YouTube. It you want better than that, the Chronos camera is probably your best bet. https://www.krontech.ca/product/chronos-1-4-camera-bundle/
In any case, most of these cameras will get you only a few seconds of recording at best. Lighting is also a consideration when running high frame rates. Sometimes you need to flood the subject with an extraordinary amount of light so your video is not too dark.
All of that said, high speed video is cool and allows you to see things you never knew were happening.
We borrowed a friend’s recent-ish Samsung phone that has the capability to shoot 960fps at a reasonable resolution. Markedly more useful for watching the gamepiece-shooter interactions than 240fps for sure.
As far as I could tell, to do meaningfully better than that, we would really need to step up to at least a Chronos v1, which is several thousand dollars.
I own a Cronos 1.4 and it is awesome. I would highly recommend one even though they are expensive the diagnostic data you can get from using one is pretty crazy if you want to take the time to count frames. I would highly suggest not getting the lowest amount of RAM, it was an extra $400 for the middle tier and it doubles the record time.
We take HSV requests too! I can’t promise we’ll get to every one of them, but if anyone is looking for something specific we are happy to do our best to accommodate.
To answer OPs question: we have gained significant insight using both cellphone and real HSV cameras. @Nate_Laverdure linked some of the better studies we’ve done with our Krontech camera. Here is a link to an analysis we did using just a cellphone camera: FRC 95 The Grasshoppers 2020 Build Thread - #78 by JamesCH95
That looks like it is processing the frames with an algorithm to stretch out the time so it looks like 960. Helpful, but not as useful for frame by frame analysis.
I like my Pixel 7a, but another option is to reach out to a local university or company and see if they have one of those mocap systems that can track stuff and see if they’ll let you use it.
Each camera on those costs more than a car (I think each one in one of the labs I work in is $20k+ USD, and they have roughly 20) so they’re probably unobtainable for an FRC team.
Top notch tracking though.
A lot of this can be done in whatever film editing software you are using. Just make sure that the camera is up to framerate that it can be done without streaking.
Most important aspect here is understanding what frame rate will be enough for your needs. I work in custom automation, and rarely need more than 240fps to get the info on if a mechanism is or isn’t meeting design intent and why. You can get that with an iPhone or a GoPro pretty easily.
May not let you watch your flywheel velocity in 5 degree increments this year, but given how pricey more expensive equipment can get I would definitely recommend staying on the lower-cost end of the spectrum. Would bet it’s good enough for what you need.
Yeah, I suspected that’s how it works and I just confirmed. It records at 240 and then interpolates 4x to slow it down to 960. I thought the quality looked a little too good and the motion a little too blurry.
I’ve used a Nikon point and shoot that does 1200FPS for 3 seconds at very low resolution. That was good enough to see a very slow moving bullet.
I’ve also run a Phantom camera that goes up to 1M FPS (at some tiny resolution like 4x100 pixels). But there is not much in FRC that would need anywhere near that capability.
Slomo is generally filmed at 120 fps (or higher), which is achievable by most flagship phones, so you could get solid slo-mo even from editing a regular video to half speed, I say just stick with the phone unless you literally need to spend money.
I would like to add that most films are done in 24 fps; high fps is considered 60 for recording, so this lets you comfortably go to half or even 1/4 speed on a 120 fps camera.
I will note that I have seen a prototype that looks decent at 240fps but awful at 960fps. The extra frames helped a ton.
While most new high-end phones can shoot at 240fps, it’s not a major investment to try and look for an old used android phone (cracked screen, broken wifi may make it even cheaper) that can do 960.