What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Rolling shutter Jello

Don

Don

Well Known Member
Am I correct that increasing the FPS will reduce the Jello effect that rolling shutters produce when there's motion in camera's FOV?

Are there any other ways to combat the Jello effect other than dealing with it in post-processing?

Thanks!
 
Increasing the FPS will not significantly aid with the "rolling shutter" problem. The issue is a byproduct of how the vast majority of low-cost POV cameras work (CMOS technology). Rather than capture a frame at a time, they capture a scan line at a time. The result is the image composition can change between scan lines at the start of the frame relative to the bottom.

"jelly" is caused by vibration while "warping" is caused by objects moving quickly.

You can reduce the jelly effect by reducing vibration - specially at key frequencies.
 
humptybump, great explanation of the phenomenon, makes sense. I recieved a Gopro as a gift for Christmas and have seen this already, now I have some understanding of it so perhaps I can combat the issue.

Thanks!

Doug
 
What Jello effect are you talking about? The one I'm familiar with happens to the whole image when the camera moves or vibrates. In that case, you need a more rigid mounting system.
 
Bryan - most occurrences of the "jelly" effect are related to vibration and most of the vibration is from the engine.

One thing I still have not done if go out and fly at a series of engine RPM while recording so I can see which settings cause the most/least effects.

Since every aircraft is different, with different resonance, and different camera mounting positions, I don't think any one aircraft's data will help any other aircraft.
 
humptybump said:
Bryan - most occurrences of the "jelly" effect are related to vibration and most of the vibration is from the engine.

One thing I still have not done if go out and fly at a series of engine RPM while recording so I can see which settings cause the most/least effects.

Since every aircraft is different, with different resonance, and different camera mounting positions, I don't think any one aircraft's data will help any other aircraft.
Click to expand...
I've had great luck using a three-legged RAM mount system. Suction cups (RAM items) attached to a camera work well in many locations on the plane. Zero jello. And very secure installation (if a glossy/smooth surface for the suction cups). Pic shown has a screwed ball at the aft location (for use in the tie down).

2012-11-03_10-29-43_841.jpg
 
Last edited:
Low Pass said:
What Jello effect are you talking about? The one I'm familiar with happens to the whole image when the camera moves or vibrates. In that case, you need a more rigid mounting system.
Click to expand...

Lowpass, It is definitely not the whole image. It's mostly in the bottom left side of the image and its a waviness like you'd get looking through warped glass. Which means it may be warping, per humptybump's explanation.

The mount I'm using is exactly like the one PerfTech shows here: http://www.vansairforce.com/community/showpost.php?p=534393&postcount=29 (except mine is on the left wing). It seems very solid but I'm open to ways to improve it.

In any event, my interest is in reducing or eliminating the flaw with the least amount trial and error. If the mount is the problem there's no sense in trying different camera settings and engine RPMs, etc. So I guess the first thing is to figure out if what I have is a 'jello effect' or 'warping'.
 
try the GoPro forums.....?

I'm certainly no authority, but there's lots of possible improvements...here's one of the better explanations I've found.

http://flitetest.com/articles/vibrations-and-jello-effect-causes-and-cures

Increasing the frame rate, or using filters seems to be the basic solution path, depending on your equipment. Try duct taping an old sunglass lense over the camera lense, and see if it improves it; let us know how things work out!
 
The use of a neutral density filter changes the shutter speed and hence it changes at which vibration frequency the "jelly roll" effect occurs.

What is interesting is that increasing the frame rate and adding an ND filter would be counter producing results - one is speeding the shutter rate and one is slowing it.
 
Good point Glen!

humptybump said:
What is interesting is that increasing the frame rate and adding an ND filter would be counter producing results - one is speeding the shutter rate and one is slowing it.
Click to expand...

...by sheer fluke I said 'frame rate 'OR' filter.....' but that's easy to miss, thanks for that clarification Glen! ( 'speshully for those of us who don't really know WHY things work, but use the trail & error method!)
 
The complicating factors here are multiplying quickly.

1. ND filters. These appear to be a lot more practical with the GoPro than with the Drift because of the physical geometry of the cameras. I'll play with shutter speeds and 1080p vs 720p first.

2. It seemed to me that using a ND filter primarily got rid of whacky prop images by slowing down the shutter speed. I think Glen essentially said as much - ND filters and increasing shutter speed are at cross purposes.

3. Vibration wasn't something I'd given thought to. I may buy some longer screws from ACS and try some different foams to reduce vibration.

4. The notion that vibrations are mostly cause by the engine seems a bit simple - I'd think the prop would contribute (a) vibration(s) and hence we balance them. I would also think the airframe may have some intrinsic vibration that isn't normally an issue until its exacerbated (flutter) or maybe interferes with low end videography. FWIW, flying commercially I have frequently watched wings oscillate. I'd guess not every vibration that might affect video results is visible.

In the end it looks like I can experiment with:
-engine speed (and when 702DA is finished, engine and prop speeds)
-vibration isolation
-shutter speeds (fps)
-scan rates (720p vs 1080p)

Thanks for the thought gentlemen and if I get any results worth reporting, I'll let you know.
 
Last edited:
Don - While there are lots of things vibrating on an airplane - engine vs prop vs other things - those can all be normalized to RPM.

In case someone wants to see a VIVID example of the phenomenon, here is a short video I shot a while back ...

http://vimeo.com/31002944

Jump to 2:10 and watch what happens when I change RPMs at 2:14 and then again at 2:22
 
Last edited:
Interesting Glen - that's worse than what I was dealing with but that is what I'm talking about.

I'll start out testing engine RPMs and see what my results are.

Do you have any opinion if you could effectively isolate the camera from vibrations without a lot of expense?
 
Don - I shared that clip because it is an ideal (and extreme case). I have a small bit of padding and it has no effect. Given I can map it to RPM, I just plan to map the effect to a range of my RPM settings. That way, I can know when I'll have unusable video.
 
more confusion

Based on my experience with a Drift camera on my Legal Eagle XL (LOTS of airframe vibration!), I think we are getting "shutter speed" and "frame rate" confused. The two terms are not referring to the same camera mechanics.

I have my Drift camera set to 60 fps because it reduces shutter roll (jello) effects. But the camera shutter speed can vary while still operating at 60 fps. Here are real-world examples:

http://www.youtube.com/watch?v=f9iforGIIQ0

This vid was shot in bright daylight at 60 fps. Notice the prop artifacts caused by the scanning of the shutter.

Here is a video shot in low evening light at 60 fps, notice how the prop is now blurring instead of "chopping":

http://www.youtube.com/watch?v=ndSqKW94WCQ

The reason the prop blur is enhanced in low light is because the camera decreased shutter speed in order to achieve proper exposure. The frame rate was 60 fps in both cases, it didn't change, but shutter speed did change.

Density filters have been recommended in order to increase prop blur by slowing the shutter speed. But high frame rate is the best method for reducing rolling due to airframe vibration. Using a neutral density filter will not impact frame rate, that is a setting in the camera menu.

Hope this helps,
 
Last edited:
Sam Buchanan said:
Based on my experience with a Drift camera on my Legal Eagle XL (LOTS of airframe vibration!), I think we are getting "shutter speed" and "frame rate" confused. The two terms are not referring to the same camera mechanics.

Hope this helps,
Click to expand...

Sam,

You're right I was using the terms interchangeably...I think. Thanks for picking up on this and trying to straighten me out. Something was bugging me from my SLR and film days but I had not put my finger on it. Let see if I understand the difference.

When I set the frame rate to 60 FPS, the I will always get 60 FPS regardless of the lighting. What the camera will change (in order to get the right exposure) is the "shutter speed." I'm further guessing that the shutter speed is really the speed that the progressive scan that makes up one frame is completed. I'm going to stick my neck out and say it sounds to me like, if you're shooting at 60 frames per second, the scan rate (or shutter speed) had to be 1/60th of a second (and maybe a bit faster to make up for lag) or faster. If the light was bright the scan rate (i.e., shutter speed) could go up several or many times the previous 1/60th of a second rate.

In bright light, I can imagine the scan rate may be 1/1000 sec or faster to get the right exposure but, the frame rate will still be 60 FPS. In dim light the scan rate (or shutter speed) would drop to 1/60 sec (or a bit faster) and fast moving stuff would be blurred.

Do I have the difference between frame rate and shutter speed correct?
 
Sam, yes. I was not clarifying very well.

Don, yes. In theory the shutter rate could only go as low as the frame rate since it would need to start a subsequent frame as soon as it finished the current one. At any faster shutter rate, the camera would simply wait until it was time to start the next frame.

I have som ND film (from the old theater days). I'll see if I can install it "behind" the lens of the Drift. It may help in certain cases but not all. It may bugger the whole thing up so no one try this at home.
 
Last edited:
Thanks

Sam Buchanan said:
Based on my experience with a Drift camera on my Legal Eagle XL (LOTS of airframe vibration!), I think we are getting "shutter speed" and "frame rate" confused. The two terms are not referring to the same camera mechanics.

I have my Drift camera set to 60 fps because it reduces shutter roll (jello) effects. But the camera shutter speed can vary while still operating at 60 fps. Here are real-world examples:

http://www.youtube.com/watch?v=f9iforGIIQ0

This vid was shot in bright daylight at 60 fps. Notice the prop artifacts caused by the scanning of the shutter.

Here is a video shot in low evening light at 60 fps, notice how the prop is now blurring instead of "chopping":

http://www.youtube.com/watch?v=ndSqKW94WCQ

The reason the prop blur is enhanced in low light is because the camera decreased shutter speed in order to achieve proper exposure. The frame rate was 60 fps in both cases, it didn't change, but shutter speed did change.

Density filters have been recommended in order to increase prop blur by slowing the shutter speed. But high frame rate is the best method for reducing rolling due to airframe vibration. Using a neutral density filter will not impact frame rate, that is a setting in the camera menu.

Hope this helps,
Click to expand...

Good information, and great videos! Thanks!!!
 
Don said:
Sam,

You're right I was using the terms interchangeably...I think. Thanks for picking up on this and trying to straighten me out. Something was bugging me from my SLR and film days but I had not put my finger on it. Let see if I understand the difference.

When I set the frame rate to 60 FPS, the I will always get 60 FPS regardless of the lighting. What the camera will change (in order to get the right exposure) is the "shutter speed." I'm further guessing that the shutter speed is really the speed that the progressive scan that makes up one frame is completed. I'm going to stick my neck out and say it sounds to me like, if you're shooting at 60 frames per second, the scan rate (or shutter speed) had to be 1/60th of a second (and maybe a bit faster to make up for lag) or faster. If the light was bright the scan rate (i.e., shutter speed) could go up several or many times the previous 1/60th of a second rate.

In bright light, I can imagine the scan rate may be 1/1000 sec or faster to get the right exposure but, the frame rate will still be 60 FPS. In dim light the scan rate (or shutter speed) would drop to 1/60 sec (or a bit faster) and fast moving stuff would be blurred.

Do I have the difference between frame rate and shutter speed correct?
Click to expand...

No, I can't see frame rate exceeding 60 fps. The software we use to produce videos is designed (per menu selection) for a specific frame rate (30 fps, 60 fps, etc).

I think the frame rate is locked per the setting in the camera menu. But shutter speed can vary and still not impact frame rate--a high shutter speed will just mean a particular scan line was exposed for a briefer period of time than a scan line with a slower shutter speed.

Shutter speed and frame rate are not related (except maybe on the slow end) if my understanding of camera mechanics is correct (but still willing to learn after 32 years in the photography biz...).
 
Last edited:
Sam Buchanan said:
No, I can't see frame rate exceeding 60 fps. The software we use to produce videos is designed (per menu selection) for a specific frame rate (30 fps, 60 fps, etc).

I think the frame rate is locked per the setting in the camera menu. But shutter speed can vary and still not impact frame rate--a high shutter speed will just mean a particular scan line was exposed for a briefer period of time than a scan line with a slower shutter speed.

Shutter speed and frame rate are not related (except maybe on the slow end) if my understanding of camera mechanics is correct (but still willing to learn after 32 years in the photography biz...).
Click to expand...

I think we're meaning the same thing now with the terms frame rate and shutter speed/scan speed. FWIW, the Drift Ghost S will do 60 FPS @1080p, 120 FPS @720p and 240 FPS @WVGA. Not to debate you, but it doesn't seem like 60 FPS is a limit these days. What I don't know is if that's a spec to sell cameras or if that's something valuable to have.

Thanks for helping me get the concepts straight. It will be helpful as I explore what the camera can do.
 
Don said:
I think we're meaning the same thing now with the terms frame rate and shutter speed/scan speed. FWIW, the Drift Ghost S will do 60 FPS @1080p, 120 FPS @720p and 240 FPS @WVGA. Not to debate you, but it doesn't seem like 60 FPS is a limit these days. What I don't know is if that's a spec to sell cameras or if that's something valuable to have.

Thanks for helping me get the concepts straight. It will be helpful as I explore what the camera can do.
Click to expand...

Yep, I need to read up on this some more. The high frame rate is due to the lower res image, not as many scans needed for one frame, not as a mechanism to determine exposure.

But I don't know how our production software handles the different rates. When outputting a finished video, we specify what frame rate we want. What happens between camera file and finished file I know not. :)
 
For flying, I haven't see the 120fps will be as useful. It was motivated by the target audience for these POV cameras - the extreme sports community. You can imagine the "cool" factor of slowing down snowboarder in powder.

The latest iPhone also supports 120fps. I use it often for the slow-motion effect. (video examples of slow-motion)

I have not found a use for slow-motion while flying but I've not testing 120fps as a mechanism for reducing jelly roll.

There are a few more DIY options for added a neutral density filter to the GoPros. For the Drift cameras the best option will probably be to add an ND filter to a water proof housing.

001_0327-330x245.jpg
-or-
drift-waterproof-case-blog-lens.jpg
 
Last edited:
As another data point, I've noticed how the late evening Legal Eagle vids shot in really low light are softer than those shot with brighter light. This is no doubt due to the slow shutter speed needed to achieve proper exposure and airframe vibs that usually go unnoticed being apparent at the slow shutter speed.

So I assume the very high frame rates some cameras boast will be useful only with high light levels.

Another observation--as far as I know, YouTube dials the frame rate back to their default setting regardless of the quality of the uploaded file.
 
Last edited:
Glen, the idea of adding the ND filter to a case makes a lot of sense to me. As for the higher frame rates, I wonder if there wouldn't be specific instances when you're trying to slow something down (not just general flying) that it would be handy.

For example, sometimes I have the nose-wheel shimmy but it isn't usual. So far, the only consistent variable seems to be whether I have a PAX or not. There are also two different motions I've seen. One is the side to side wobble (with a PAX) and the other is more of a bouncing oscillation that I've seen taking off solo on a paved runway. I wonder if descent slo-mo might help me sort out whether there's a problem with my technique or something else. I'm not trying to get off topic and discuss nose wheel shimmy, just saying there might be a use for slo-mo for other than dramatic effects.
 
Question here from a complete videography idiot... Does the new Garmin camera use this same type of image sensor? Is it as likely to experience this same type of jello/warping/whatever image distortion?

At some point I'll buy a camera with the express intent of using it to capture first engine runs, first flights and who-knows-what further down the line. I'd rather fork out money for something that's going to produce a decent image in our little niche aviation application, so all advice is gratefully received.
 
Canadian_JOY said:
Question here from a complete videography idiot... Does the new Garmin camera use this same type of image sensor? Is it as likely to experience this same type of jello/warping/whatever image distortion?
Click to expand...

I'm no expert but the Garmin VIRB uses a CMOS sensor. The choice in video cameras seems to be between a CMOS sensor, with a rolling shutter and a CCD sensor with a global shutter. They both have their strengths and weaknesses. Both are capable of producing high quality digital movies.

Until recently, CCD were high priced and produced the best quality images. Apparently, in recent times CMOS sensors are catching up on quality and CCD's are dropping in price.

From what I've seen from cameras in this class - mainly Drift and GoPro cameras, the quality can be amazing. If you learn about your camera, I think Glen has pretty well shown that the jello shimmy can be avoided and Sam and others have shown how to avoid whacky looking propellers while using a camera with a CMOS sensor. Apparently CCD sensors have an issue with "smearing" and I don't know what causes it or how to avoid it.

My best guess, armed with the little knowledge I have so far is, exploring the strengths and limits of your equipment, so you can play off the strengths and avoid the weaknesses, is worth more than just buying a camera with different limitations.

You might see if you can borrow a camera (Drift, GoPro, or VIRB) and simulate an engine start of first flight and see what 'artifacts' you find in your video and then try and avoid the problem. My guess is on engine start, your prop is going to look strange unless you shoot it with a ND filter or do it in low light. If you shoot the first flight from the ground, I don't think you'll have many issues...but likely it will be time well spent testing it out.
 
Canadian_JOY said:
Question here from a complete videography idiot... Does the new Garmin camera use this same type of image sensor? Is it as likely to experience this same type of jello/warping/whatever image distortion?

At some point I'll buy a camera with the express intent of using it to capture first engine runs, first flights and who-knows-what further down the line. I'd rather fork out money for something that's going to produce a decent image in our little niche aviation application, so all advice is gratefully received.
Click to expand...

The jello/warping can be eliminated with a very rigid mount that is secured to something that doesn't vibrate. Unfortunately, most of the RAM mounts that have the camera cantilevered at the end of a ball-mounted arm will enhance vibrations as the camera jiggles at the end of the arm. Compound the problem with a rubber suction cup attachment and you really have a mess.

I tried several custom-built mounts on my Legal Eagle before finding one that really works nicely with insignificant vibration. Here is a sophisticated looking mount but I never could damp all the vibration from the two-cylinder VW engine:

camera-mount.jpg


The present mount has the camera mounted rigidly to the airframe without hanging on an arm:

camera-mount-tail.jpg


This one works great.

camera-mount-tail4.jpg


These mounts are different from what you need on an RV but you get the idea. Fortunately, an RV doesn't have the large amplitude vibs of my little plane so we may have greater latitude in designing a mount.
 
Last edited:
Thanks for the feedback on the Garmin VIRB, as well as all the helpful pointers. I've got long enough between now and first engine start that I might be able to figure some of this camera "stuff" out!
 
The deeper I dig, the more I learn. There's a ton of information out there but most of it is pretty scattered. If anyone's interested in a fairly compact article that compares CMOS rolling shutter artifacts with CCD global shutter artifacts, here's a link. If you're looking for camera reviews, it isn't here. The author looks at the technology and does a good job of making sense of how it translates into actual videos real people take. http://www.dvxuser.com/jason/CMOS-CCD/
 
Prop Blur Filter

One of VAF sponsors (Aircraft Spruce) sells a filter that claims to remove the annoying propeller distortion.

11-12333a.jpg


11-12333b.jpg


Anyone have any experience with it?
 
11-12333a.jpg


From ACS's web site:

"The GoPro's image sensor is controlled by the amount of light it senses. In normal lighting conditions the sensor runs faster than the propeller causing distortion and odd lines in your video. Our filter modulates the amount of light entering the sensor and keeps the speed of the sensor slower than the propeller. Since the amount of light changes constantly and the filter does not we designed the filter to completely remove the propeller in 90% of the conditions experienced in VFR flight. For best results we recommend avoiding shots straight into the sun."

Me thinks this is a plain ol' neutral density filter (probably a 3X if I had to guess). :)
 
You must log in or register to reply here.
Back
Top