BB_DF

WARNING: Geek speak ahead...

The standard SAVS camera mount is what engineers would call an underdamped system. I have experienced the bobbing behavior myself and seen its effects in nearly all unprocessed SAVS videos. The airframe is stable, but the camera just keeps on a-bouncing. The worst case scenario occurs when the camera bounce is great enough to destabilize the airframe, causing an oscillation between the airframe and the camera arm - certainly a reverse of the intended function.

This type of oscillation is completely unnecessary and easily remedied. There needs to be SOME isolation between the airframe and the camera, but this would be only to remove higher vibrational influences caused by the drive systems. Without active (gyro) stabilization, the intended effect of making up for piloting and wind variables is pretty much a lost cause, and in my opinion is made worse by the present design for the reasons mentioned above. Simply put, the camera and video image is often less stable than the airframe itself.

The best solution, I think, would be a 1" wide strip of very thin, soft foam wrapped around the camera support rod enough times to slide into the rubber ring retainer for a snug fit. This would still allow isolation of higher frequency vibrations, but would eliminate the bobbing action. I couldn't find any of this type of foam at the moment (very common as packing material), so I tried another approach. I reversed the direction of the rubber ring and slid a plastic ball over the rod. I'm still looking for a thin foam sheet, but this is already a big improvement in the video quality.

Cheers,
Bruce

Sky High

I totally agree and have thought of using a piece of thin rubber sheeting inside the round isolator with the CF rod passing through. Like you, I have also thought of wrapping some thin foam around the rod and stuffing it in the retainer still leaving a little bit of space for movement. That would definitely suppress those bobbing tendencies. However, the bobbing does slightly counteract stick inputs on all axes to further stabilize the image. So, if you take out the free motion as it is now, any sudden movement(s) in any axis(es) will be translated to the image. Some earlier proto-types of DFI's current design used a series of rubber bands forming a matrix circle for the rod to bounce from if it traveled too much but there was still at least 1/4" of play in that circle for travel. Please post some POV video of the results of your idea.

BB_DF

Here is some raw footage. Still pretty green at flying, but there's no bouncy-bouncy in the camera now, just pilot & wind funny business. At about 30 seconds, there's a controller hiccup onboard and the DF dropped about a foot or two. I've got a theory on what's causing this (TI was enabled). Regardless of the camera setup processing will be needed for smoother video, but right now I'm mostly interested in looking for snipers on the roof . Tomorrow at 8:30 I've got a demo flight for a local P.I. Hope I don't make a fool out of myself - wish me luck!

Ciao,
Bruce

http://www.youtube.com/watch?v=6_DytqT3dJk

Sky High

How long have you had your DF? I know it hasn't been too long but that flight was very good! Ti makes all the difference doesn't it? The shadow at the end of the video is really cool too because of how the rotors seem to be static while the DF is still flying. Good job!

BB_DF

Well thanks for the kind words! Got the SAVS on Monday, actually. Still sweating bullets half the time when I'm flying it, but yes, the Ti is amazing. It would be impossible for me to stay oriented at that height without it. I think the DF is a really fine machine, and I have also found the folks in Saskatoon quite helpful.

Sky High

I couldn't tell by the shadow but was that flight with that additional dampener shown in your first post? Just put some nav lights on it and believe it or not it will be even easier to fly near dark and after because orientation will be easier to maintain.

BB_DF

Yes, same rig as shown in the first picture.

BB_DF

Another test today with the new camera mount setup:

http://www.youtube.com/watch?v=X0biMhGpLXA

BB_DF

This dampener was made from a foam mini-roller.

Sky High

Is that a hard plastic center core? If so, that would most likely transfer alot of high frequency vibrations to the camera.

BB_DF

It's an improvement over Mod #1, but you're right - I need to dump the core. When settling with power (descent), it picks up some rotor vibration. On ascent, however, it's smooth as silk. The best idea is still using thin foam wrapped around the shaft, but I haven't found any yet. I've thrown away huges sheets of it in the past. Doh!

Later:
I took another 1/2 roller and made one cut lengthwise, opened it up and peeled out the plastic center. Then I wrapped it tight around the rod, slid it into the rubber retainer and let it expand. Will test tomorrow. Here are some raw video clips from today. I need to slow down my pans - hope you took your Dramamine!

http://www.youtube.com/watch?v=EYzQZHxU6PY

Sky High

Great flying, again! You're going really high now. I have noticed with mine, yours and others that forward flight always has more vibration than when hovering. I think this is due to the wind resistance pushing the ball back thereby allowing the camera boom to touch the retainer. I have never gotten good results with forward flight without using post processing. Maybe adding the foam dampener in the retainer will take care of that issue. I have some 1/8" foam that I will try in the retainer. What you have done with the thicker foam, with the core removed, may work even better because it is one piece and won't try to unroll or slide. I can't wait to see your video with it's use.

Sky High

Here is the dampening solution with wrapped foam and a fairly secure fit. It probably wouldn't hurt to tape the ends to prevent it from unrolling over time. I think the solid piece like yours is the best solution. The bright yellow should help with orientation too. I can't fly and get video right now because of flight board/video board compatibility issues. [:@]

BB_DF

Yes, very much so. It's surprising how visible it is even at fairly high altitudes.

Two videos from today, the first is with the foam wrap shown above:

http://www.youtube.com/watch?v=t1q0FmfugC8

The second is the standard SAVS setup:

http://www.youtube.com/watch?v=4-ttMhHOx5k

So basically, it's a choice between a high-freq vibration, or a low-freq oscillation. The mod vibrates while descending, and the standard setup oscillates while climbing. The wind was calm in both cases, so it's a pretty good A-B comparison. I cut a lot of shaky stuff out of both videos. I think if it's windy, though, the foam dampener will be the way to go.

And how about the de-shake software - which case do you think would be easier for it to correct? With the smaller vibrations, the outside frame won't have to jump around as much, but it might be too fast for the algorithm to handle smoothly(?).

dart340

Hi

I was just looking at your two videos, I would say that the standard SAVS looks sharper than your modified foam wrap video.
Its hard to tell with the compressed video but if you play your videos at the full resolution and you pause the video, are the still images sharp? If both videos are sharp I would consider the foam wrap but from my experience it is easier to stabilize a bobbing video with sharp images on each frame than blurry images on a non bobbing video.

Why don't you post a full resolution still frame of each and then we can compare the two.

Sky High

I'm thinking all of that vibration is being transferred though the rubber attachment under the flight board. Have you ever held your DF while the rotors spin at flight speed? Spin it up to 3/4 throttle and hold it in your hand. Then slowly oscillate your hand like a gimbal and feel all of that vibration that the DF creates when the rotors pulse to correct for flight. Put your finger on that rubber boot and/or the camera boom nearest the flight board. You'll feel less vibration if you keep the DF level but even more when you oscillate it. The software processes high and low frequency fairly well. You can see several demos in [link=http://www.rcuniverse.com/forum/m_4094795/mpage_1/key_/tm.htm]this[/link] thread including a split screen comparison that I did in [link=http://www.rcuniverse.com/forum/fb.asp?m=4140876]this[/link] post and then how the software works in [link=http://www.rcuniverse.com/forum/fb.asp?m=4136887]this[/link] post.

BB_DF

That's a great observation, and might be a crucial part of the solution. It could also explain why the jitter seems to appear at odd times, which might have something to do with panning yaw and Ti corrections. So the jitter might be occurring along the z axis, rather than just x and y. That could make it a bit more difficult to design a dampener. Up until now I was thinking that what is needed is a mechanical version of a mid-pass filter to dampen the low-frequency bobble and high-frequency jitter.

My next project is to get up to speed on the de-shake software. I cut the foam down to 1" in length, but if your theory is right, there's nothing there to stop the longitudinal component. Also the foam should really be VERY soft, so that it just acts to center the rod in the ring.

Sky High

The earlier picture of the foam showed it to be much tighter on the boom. Now it looks like there's alot of play for the boom to move around. Can you get some aerogel?

Way before the SAVS mount was released there were a couple of innovative people that made some 3 stage isolaters below. The first 4 are from davidtkal and the last two are from 1videoman1. So with the SAVS mount, the rubber connector is the first stage of dampening from airframe vibration, the boom itself is the second to counteract airframe movement during flight and the foam insert is the third to further remove high frequency vibrations. To me, the first stage is the most important because that's the closest to the airframe. That hard rubber isn't the best shock absorber but it's better than a plastic ball joint hard mount like with the older Eyecam system. It's a perfect example of a chain being as strong as it's weakest link. Great stabilization isn't going to prevent RF hits in the video though and WIFI along with other 2.4 devices are the new threat to this system.

BB_DF

That's quite the rig there! I notice the battery is not located to offset the weight - that must have been a bear to trim.

The photo makes it look like there's a hole through the middle of the foam roll, but it's wound tight on the support shaft like the previous one.

Old Man Mike

Maybe this is a little off subject but have you noticed the small consumer video cameras now available with Opitical Image Stabilization? The gyros in these cameras may be able to eliminate the shake much more effective than the mechanical approach. Two cameras in the 8oz catagory with this feature are the Panasonic SDR-S150 and the Sayno CG65. If the LCD screens are removed and power obtained from the main helicopter battery, the weight could approach a managable payload level of around 5 oz. You may also be able to feed the video driving the LCD screen into the SAVS transmitter for real time transmission but I think I would be OK with having everything recorded to the SD memory chip.

Mike

BB_DF

Hi dart340,
Sorry I missed you post previously. My codec is giving me an error when I try to grab a frame, but these two pictures were exported from Movie Maker. They both look pretty bad to me. There's no doubt that when it stops bobbing, the standard setup will be better isolated. What is needed is something between the open movement limiter and the foam support. Something along the lines of the two rubber band inner dampeners, but much weaker, positioned where the rubber ring is. It should only act to gently center the arm to reduce bobbing. That's just too long a moment arm to be supported from one end only.

Mike - active stabilization is really the only way to completely solve this problem at the camera. However, post-processing software might make the added complexity unnecessary.

- Bruce

Sky High

Old Man Mike, your idea is a good one but as with digital still cameras, you have to figure out how and where to mount these cameras for steady flight.

If you are recording to a DVR instead of an uncompressed format like DV or to a PC or portable HD, then your content will be softened by the compression of the DVR and you will never get clear images.

BB_DF, there are at least four inherent issues that plague the current video system as far as quality. The first is that the camera has only one CCD, but at least it's a CCD instead of the older CMOS. However, it should be noted that higher end CMOS cameras are now rivaling CCD cameras. The second is that with video, there are scan lines, 480 in this camera, that make up the image and tend to smear when moved quickly in various directions and that aren't present in megapixel cameras. The third is that the even though the image is good considering what it is, the recorded image is from a wireless signal and will not be as sharp as a native recording because of stray RF or from the helo, however slight. The fourth is the well controlled but not totally eliminated high and low frequency vibrations from the airframe and wind.

I have managed to get some clear images pulled from video by the simple method of checking each frame in the area I want to pull from and choosing one at the highest peak or lowest fall of the ball's bounce whether it's up or down, left or right. This will usually yield a clear image because for that instant in time the ball is practically perfectly still because it has met it's limit of travel and is preparing to move in it's reverse motion. It's like a bouncing ball has a pinnacle of it's bounce and for an instant it is suspended without motion until it begins it's descent.

I never pull stills from stabilized processed video because that process performs a push or zoom on the image to do what it does and that just blows up the scan lines and softens the image. You can read about and actually see how that works in [link=http://www.rcuniverse.com/forum/fb.asp?m=4136887]this[/link] thread. [link=http://www.rcuniverse.com/forum/m_3845986/tm.htm]Here[/link] is a thread with clear video stills that I have pulled from native unprocessed video from previous flights.

BB_DF

But other than that, Mrs. Lincoln, how did you enjoy the play?

Those stills from the other thread are pretty darn good, and the dancing frame thingy is very interesting. I wonder if DFI does any tricks to get such good promotional videos?

[link=http://www.uavp.de/index.php?option=com_content&task=view&id=37&Itemid=55]LOOK HERE[/link] for some stunning hi-res photos taken from the UAVP.

Sky High

Yes, those are impressive. [link=http://www.rcuniverse.com/forum/m_5151083/tm.htm]Here[/link] and [link=http://www.rcuniverse.com/forum/m_4400659/tm.htm]here[/link] are some high resolution stills taken from the DF.

BB_DF

What does the de-interleave step do? Also if anybody can email me a copy of "draganflyshake.vcf" for Vdub, mine seems to be corrupted, and there's nobody at DFI today. I can open the other two .vcf files with Outlook Express, but this one appears to be fubar. PM me and I'll give you my email address.

Thanks!
Bruce