Would like to get suggestions on how we can measure the quality of a video and still photography. Metrics will be important to improve two of the banes of multi-copter photography :
- vibration (x/y/z vibration that affects vid/photo quality/clarity)
- stability (roll/pitch/yaw instability that affects vid/photo quality/smoothness)
Ideally we want a sky platform that comes as close a possible to a 4-axis tripod or 7-axis camera slider.
So maybe a standard printed page used in photography? Attached is a video from a try-storm session toying with the idea (didn't have the standard printed page). The platform is a quad with a hard mounted GoPro 3+ black set on 1080s-30fps for video and 7MP-M/FOV for photo. The quality isn't important as I'm just trying to figure out how this should be measured.
P.S. my lack of pilot skill was hindering tackling this area of interest, so my brother built me a little practice quad. he makes me practice 15 minutes every morning in the man-cave. so thanks, Jim, because now i can find a 4 second clip where the copter is relatively in front of a target. i know ... i still suck but am getting better :-)
Replies
I like Jello but only in the cherry and strawberry flavor on my plate.
to eliminate this nasty video Jello, try using a CCD camera instead of a CMOS (CCDs do not use a shutter principle to read the pixels and are thus immune to Jello); or as our friend Robert says, use very soft dampeners. I say very soft because that is my personal experience : even kyosho zeal pads in my training quad was too hard to eliminate jello on a Mobius camera. Moongel and soft foam, combined have shown good results at reducing jello.
Hi Robert, Hugues,
Certainly true that CCD eliminates most kinds of video jello because it does not use a rolling shutter, but rather a "global" shutter.
The problem with CCD cameras, is it is often hard to find good and affordable ones that will let us get the quality of picture or video we are looking for.
And although I have had a lot to say about Kyosho Zeal for vibration isolation for our flight controllers accels, cameras are a whole other ball game.
For the kind of short moment high frequency vibrations that screw up the accels, Zeal usually works fine, but for cameras, all frequencies and moments of vibration are problematic.
For a camera you truly need a tuned system designed to isolate it from every form of vibration.
Unfortunately that can mean suspending the camera so loosely that it's own mass starts to conflict with the motion of the airframe inducing secondary and out of phase motions.
A camera setup that is tuned to work with one airframe may work very badly with another.
It isn't insoluble, just tedious and needs tuning for each setup.
Gary, I agree with your assessment of CCD camera options. They are great in theory with the global shutter, but then you actually go shopping for decent cameras with a CCD and find there are none. Most of the options are poor quality, with cheap lenses, low resolution, etc. So maybe there's no jello, but you have a lot of other problems giving you poor quality. The vast majority of people doing this use a quality CMOS camera, and just figure out how to eliminate vibration problems.
From what I have seen, the number one problem for cameras is angular vibration, not linear. So it's better to design the mounting system to be sure you're not getting angular vibration, even if some linear gets through. For sure, you actually don't want the mounting so soft in the quest to eliminate linear vibration, but then end up with angular vibration being created due to airframe movements. Brushless gimbals play a big role in eliminating angular vibration, because when the camera is balanced, they actually don't pass on much angular vibration by their very nature.
Hi Robert,
There are some quality CCD video cameras like Point Grey, but unfortunately they are limited in the formats they support and are not really oriented towards normal photographic uses.
And, you are certainly right, a properly working brushless gimbal can at least compensate for any sort of rotational vibration (and they appear to do so quite effectively).
But the linear vibration at least on a CMOS camera still tends to induce jello and a variety of less than desirable visual artifacts.
In a video camera, it seems like the Sony balanced steady shot system with a real time gyro stabilized optical mechanism actually does work very well, even at full zoom (saw a video of it on a multicopter zooming in on a distant buildings window with it rock steady at full 30x).
Unfortunately there isn't anything similar for still cameras.
I dunno Gary, in my mind, straight linear vibrations acting on the actual CMOS chip really should have no effect at all. Think about it. If you're capturing light rays coming from an object 100+m away, what possible difference does it make if the sensor is moving up and down by 0.001m?
The linear vibrations only cause problems through a few different mechanisms:
1) Linear vibrations could cause the lens system to move around if it's not built ruggedly enough. This will cause movement in the light rays hitting the sensor. Similarly, if the chip itself is not firmly fixed, then it could move inside the camera relative to the lens.
2) Linear vibrations can create angular vibrations if the gimbal mounting system is not built properly. If the CG of the camera is not coincident with the centroid of the damper system, then linear motion will become angular vibration.
I say this based on my experience with the Protos 500 Heli. It actually vibrates, but since the mounting system is laid out well, it still takes good photos.
I agree, the BOSS system is spectacular when used with a gimbal. I'd really like to get one. Still cameras don't need a similar system unless your system is shaking quite a lot.
the largest artifacts that i've seen from high quality cmos are from the tilting of objects when:
- moving past tall vertical objects at the edge of the screen
- panning across the front of vertical objects
Yeah, that's a fairly well known problem. I think some cameras can somewhat fix it themselves, or maybe in post-production. But generally the solution is just "don't do that".
But there is such a thing as too soft. Though that actually has a big dependence on the damper geometry.
CCD does help I guess, I've never used one. I've actually not a lot of problems getting good video from a multirotor. It's pretty easy. Helicopters are another story.
Hi Rob, I saw in another post you said you used the LotusRC NEX5 gimbal. What controller do you use with it ? Alexmos ? And if yes, what are your Power, P, I, D settings (roll & picth) and which motors are used (how many turns ?). Then what dampers other than the ones delivered with the gimbal do you use ?
Have you a video posted somewhere with your NEX5 and this gimbal ?
thx
Hi Hugues,
I'm actually using the VirtualRobotix gimbal controller with it.
I'm not using the dampers it came with. They might work on a multirotor, but I don't think they were going to be good enough to work on the helicopter. So I'm using a damper system of my own design.
Here's a quick sample of the video from the system. It's not great, but I think that's mostly due to external factors. There was actually a bit of wind at the time, the heli is not as stable as I'd like, and also it was in Loiter mode which can be quite "busy" as you may know. You can see the landing gear in the frame towards the second half of the video, so you can see what the gimbal had to deal with. I think it did pretty well. Most importantly, no vibrations from the heli.
-
1
-
2
of 2 Next