How Do I get rid of "jello" when stabilizing hand held shots?

[Phil Bates]

Thank you , everyone for your comments on this thread. Just FYI, this shot was meant as a test only, so no offense taken on the "crappy footage" comment. I am planning a severe weather aerial storm chase, and expect that I will have to deal with footage at least this rough. So, I took a test flight in a Cessna last week just to see what it felt like. My Epic currently has the Optitek Nikon Mount but it may be worth purchasing the Canon mount and go with an IS lens. That is what I am thinking as I ponder everyone's comments.

Also, when I processed this in AE, I chose Position Scale, Rotation, not Subspace Warp (there is no pinning of the edges). I also chose Stabelize/Crop, not Autoscale.

Paul, thank you for going to the trouble to run another stabilization pass! I am impressed that most of my issues disappeared. Did you use The Foundry's Rolling Shutter plugin? Or did you just tweak the AE stabilizer? I did notice the luminance flashing at the top of the frame, what's that about? (not complaining, just want to understand).

Thanks again,
Phil

 

[Phil Bates]

...Yep, the software starts with pinning the edges of the frame as default. If you do not give it the space to move by applying the correct type of scaling you are trashed. I suspect that if the original settings were set to "Automatic", there were not identifiable markers for the software to recognize - one tree looks like another. Result: Edges stable, center trying to find it's relationship - looking like jello. So the scaling needed to be set manually - as was proposed....

Brian, just to be clear, the original settings were not set to "automatic", so there is no subspace warping. Understandable that it looks that way. So I think we've determined the "jello" is coming from the nature of the distortion present in a wide angle lens or the rolling shutter (or both).

Thank You!
Phil

 

[Brian F Kobylarz]

Brian, just to be clear, the original settings were not set to "automatic", so there is no subspace warping. Understandable that it looks that way. So I think we've determined the "jello" is coming from the nature of the distortion present in a wide angle lens or the rolling shutter (or both).

Thank You!
Phil

Phil,

I think some of us were confused - post 3 and post 7 referred to trying / using AE warp with the jello effect resulting.

For what it is worth, I recently shot two sets of aerials of the latest VIRGINIA class submarine on sea trials.
First run with my Scarlet / Canon Ti mount / Canon 24-105mm IS lens.
Shot most of it fully wide, some zoomed in.
Stabilization was fair - certainly better than HH without IS, but the atmosphere was cooperating.
Did not detect any "jello effect" in any shot - but applied no software stabilization in post.

The second run I used my EPIC with 25mm RPP - dovetail mounted onto a Tyler pogo stick with Kenyon gyro.
Much better results than using the Canon IS lens. Without any hesitation, if you cannot spring for a Cineflex or similar, the Kenyon solution is better.

However, the Kenyon adds a lot of weight - so we have the pogo stick with internal spring - not a real solution for shooting out of a fixed wing. With the Jet Ranger, we take the door off which gives me a decent operating environment. A Cessna has a much smaller opening to shoot through.

Used that footage last week in DC (why I was not at NAB) on 42" displays - I detected no jello effect or any distortion in either camera setup - rolling shutter, wide lens - not a hint of any issue.
Again, applied no stabilization in post - everything I did to stabilize was at the point of original capture.

 

[paulherrin]

Thank you , everyone for your comments on this thread. Just FYI, this shot was meant as a test only, so no offense taken on the "crappy footage" comment. I am planning a severe weather aerial storm chase, and expect that I will have to deal with footage at least this rough. So, I took a test flight in a Cessna last week just to see what it felt like. My Epic currently has the Optitek Nikon Mount but it may be worth purchasing the Canon mount and go with an IS lens. That is what I am thinking as I ponder everyone's comments.

Also, when I processed this in AE, I chose Position Scale, Rotation, not Subspace Warp (there is no pinning of the edges). I also chose Stabelize/Crop, not Autoscale.

Paul, thank you for going to the trouble to run another stabilization pass! I am impressed that most of my issues disappeared. Did you use The Foundry's Rolling Shutter plugin? Or did you just tweak the AE stabilizer? I did notice the luminance flashing at the top of the frame, what's that about? (not complaining, just want to understand).

Thanks again,
Phil

no problem

not a double pass - i took the first five seconds of your footage. the only thing i applied was a single warp stabilizer with the settings i posted (plus some color correction, just for fun). i don't think it actually made much difference going from automatic reduction to enhanced for rolling shutter, but i figured it wouldn't hurt. cs6 should offer more precise control. i highly recommend watching the tutorials from adobe tv (there might be others, too) and just glancing at the documentation. that'll put you in business.

what's going on with the top edge has to do with the synthesize edge frame settings... check out the documentation/adobetv links about that - it's really cool, but can be a little finicky. without that, there would hardly have been any sky in the shot and i wanted to try to keep it in. i could have adjusted the settings to get rid of that stuff, i just didn't want to re-render cause i'm on my laptop and it took awhile (cause my color grading, mostly - the stabilizer was pretty fast and played back in realtime for me, on a 5yo laptop)

as for shooting - it was good you were using a high shutter speed, of course. that's a must. the kenlabs gyros are really handy, but kind of heavy. test out the IS (just run around with it or whatever) - i've always had good results, but you should test yours to make sure before you go with it. i've found that it does help quite a bit.

 

[Phil Bates]

Thanks Paul, I'll give it a try and let you know how it goes.

P

 

[Lauri Kettunen]

Hi Phil, I struggled precisely with the same problem of stabilizing aerials myself about a year ago. The challenge of stabilizing follows from the fact that the front glass of lens is far from being parallel with the ground level. For this reason a small change in tilt results in a rather big change in the projection onto the sensor. For an example, say one stands in the middle of a highway and shoots along the road shoots from a low angle. Changing the tilt and the hight a bit creates a rather big change in the projection of the highway onto the sensor.

As far as I remember, to stabilize such images I selected two points from the upper part of the image and then used tracking in AE to stabilize the image.

 

[Phil Bates]

Thank you Lauri. I have often trusted the most distant objects as anchor points for stabilizing. I like your explanation of why this is important.

Phil

 

[paulherrin]

the very best solutions, as bjorn mentioned, will involve 3d tracking to solve the camera. compensating for lens distortion is key in that process, although not absolutely necessary to get halfway decent results. i'm really eager to try out syntheyes... that looks pretty nifty.

 

[Phil Bates]

Well, I purchased a Canon mount for the Epic, and a Canon 24-105 IS lens. The results are similar, lots of jello, even at the wide end of the lens. I even see jello before doing any post stabilization work.

So... My question is this: If I see jello before the post work, is it "baked in", and not fixable with software? If so, I am having doubts about using the IS.

Thanks,
Phil

 

[kylebainter]

Well, I purchased a Canon mount for the Epic, and a Canon 24-105 IS lens. The results are similar, lots of jello, even at the wide end of the lens. I even see jello before doing any post stabilization work.

So... My question is this: If I see jello before the post work, is it "baked in", and not fixable with software? If so, I am having doubts about using the IS.

Thanks,
Phil

Phil,

I agree with you IS observations. I have shot both IS and non IS, and IS is great for certain things, but I think it is certain movements/frequencies that it cannot help. I think the best way to stabilize something like that is a cinesaddle or one of the portable tyler stabilizers.

Now kind of off topic, has anyone else noticed much more "jello" with the epic/scarlet than their Red One? I personally have, and I was wondering if it was just me.

 

[Uli Plank]

I observe less with the Epic and about the same with Scarlet.

 

[Mick van Rossum NSC]

I observe less with the Epic and about the same with Scarlet.

Did a test, mounted an Epic on a Scarlet and did some fast panning; the Scarlet has a bit more Jello indeed, but Epic was better than Red One. (non scientific test)

 

[Lee Dashiell]

I have never seen a rolling shutter/jello issue with RED One or Epic. I'm not aware of there ever being an issue with RED.

 

[kylebainter]

I have never seen a rolling shutter/jello issue with RED One or Epic. I'm not aware of there ever being an issue with RED.

I've never seen any on my RED One MX even shooting out a heli on my lap. But I have gotten Jello on the Epic. I'm wondering if it is because of the smaller size, being not as heavy, etc....

 

[Mike P.]

I've never seen any on my RED One MX even shooting out a heli on my lap. But I have gotten Jello on the Epic. I'm wondering if it is because of the smaller size, being not as heavy, etc....

Sensor is also bigger when you read the entire thing (as in, 5k vs 4k or 4.5k)

 

[Michael Brennan]

Hi Phil, I struggled precisely with the same problem of stabilizing aerials myself about a year ago. The challenge of stabilizing follows from the fact that the front glass of lens is far from being parallel with the ground level. For this reason a small change in tilt results in a rather big change in the projection onto the sensor. For an example, say one stands in the middle of a highway and shoots along the road shoots from a low angle. Changing the tilt and the hight a bit creates a rather big change in the projection of the highway onto the sensor.

Laurie,
If I understand correctly, the example you give is correct because the subject is close to the camera.
But I dont think it applies when the closest subject is 500 to 1000 feet from the lens.

Mike Brennan

 

[Michael Brennan]

Thank you , everyone for your comments on this thread. Just FYI, this shot was meant as a test only, so no offense taken on the "crappy footage" comment. I am planning a severe weather aerial storm chase, and expect that I will have to deal with footage at least this rough. So, I took a test flight in a Cessna last week just to see what it felt like. My Epic currently has the Optitek Nikon Mount but it may be worth purchasing the Canon mount and go with an IS lens. That is what I am thinking as I ponder everyone's comments.

Also, when I processed this in AE, I chose Position Scale, Rotation, not Subspace Warp (there is no pinning of the edges). I also chose Stabelize/Crop, not Autoscale.

Paul, thank you for going to the trouble to run another stabilization pass! I am impressed that most of my issues disappeared. Did you use The Foundry's Rolling Shutter plugin? Or did you just tweak the AE stabilizer? I did notice the luminance flashing at the top of the frame, what's that about? (not complaining, just want to understand).

Thanks again,
Phil

Phil,
The body is excellent at absorbing high frequency vibrations from the engine, so most of the shake is caused by ones body bouncing, one buffet can create a couple of body/arm bounces, if you are hand holding the movement of the arms trying to control extra G imparts unwanted tilts and pans, that far exceeds the rate and scope of the original buffet.
It is surprising how little wobble occurs when the camera is bolted to an airframe. But the issue with bolting a camera to an airframe is attenuating the high frequency vibrations caused by the engine.
Cinesaddle is good at reducing high frequency and eliminates secondary wobble so what is left is the aircraft wobble. A strut mounted camera in addition to hand held could be worth considering.

The difficulty in shooting near severe storms is the random localised downbursts where the rate of descent of the falling air is often greater than the rate of climb of the aircraft.





Mike Brennan

 

[Lauri Kettunen]

Laurie,
If I understand correctly, the example you give is correct because the subject is close to the camera.
But I dont think it applies when the closest subject is 500 to 1000 feet from the lens.

Mike, the distance is not a true factor here. Although you have a point that objects become smaller with increasing distance and this may affect how the stabilized image appears. But I've seen the jello in aerials even the objects have been a mile away.

 

[Dmitry Burenok]

Kenyon is good when you doing easy easy hovering shots - if it's pivoting or hard banking - it becomes useless. Software stabilizing and de-wobbling is definitely way to go.