Can you smooth out bumpy car footage?

C

c.g._eads

Veteran
Let's say your car footage is a little bumpy; Can you smooth it out in After Effects? I mean I know you can get rid of a camera jerk - but that's because you have lots of points of reference. If I have my camera trained on the actors from either the back or the front - there's no real point of reference since the outside scenery passing by is constantly changing.
 
If you have a program called Mocha, by Imagineer systems, you can do a full motion track and stabilize the footage pretty well pretty quickly. That's what I use for big stabilization projects. As for AE, I'll look into it for you.

Do you have some sample footage I can take a look at for you? I'll be around doing nothing for a while. So I'm tackling a lot of different projects.

-Matt
 
Last edited:
Not if you shot on an HV20 or other rolling-shutter CMOS camera. You can stabilize CCD footage if you use something like Dynapel's software, but you can't stabilize CMOS rolling shutter footage.
 
You should be able to use AE's built in motion tracker if you're just looking to fix unstabilized footage. I'm not sure how a rolling shutter would affect a motion track. Or are you referring to some other type of fix?
 
You cannot stabilize rolling-shutter footage because the frame doesn't move uniformly. The top half of the frame might be moving downwards and the bottom half might be moving upwards. It all depends on where the rolling shutter was in the frame when the bump happened.
 
Barry, how did you know I was using an Hv20? Are you telepathic?

Matt, don't have any footage. But I know at best it's going to be a little bumpy when I shoot it. I'm just trying to prepare in advance.
 
I hear Barry can small a HV20 at 500 paces. Upwind. :lipsrseal


You could use a Merlin to greatly stabilize your footage, though there's a nack to shooting Steadicam in a vehical, let me know if you go this route and I can pass on some tricks.

- Mikko
 
Do you have AE? I can see if I can motion track it and stableize it for you but I doubt it. As TehBarry said, its almost impossible with a rolling shutter CMOS cam.
 
Wow, I've never known that...

Makes me sad as I look at my HV20 and ask what the hell was I thinking when I bought you...

Again Barry, a great help as always. Though I wish you weren't right.
 
Mattykins said:
Wow, I've never known that...

Makes me sad as I look at my HV20 and ask what the hell was I thinking when I bought you...

Again Barry, a great help as always. Though I wish you weren't right.
Click to expand...

you were thinking hd 24p that looks killer for under $1000. of course will have it's limits but still a great cam.
 
I personally think it's funny, this whole emergence of "rolling shudder". It's become this terrible pair of words. Now, the HV20, which was once a beautiful piece of equipment, is always looked at as the handicapped kid - just because of those two words: "rolling shudder". You could do this to any camera out there. Pick on one little deficiency, and give it a name like, "vibrating overstretch", and all of a sudden the camera that you love feels completely inadequate. I promise you that I would have never ever noticed the "rolling shudder" had nobody ever told me. I still can't recognize. But I know it's there. Cause my camera is terrible. I hate the HV20.
 
Yeah, I remember reading about this when Stu Maschwitz pointed it out on his blog, while I think he mentioned he picked up on it over the SynthEyes forum where Russ Anderson (the owner/creator of SynthEyes) pointed it out when he used one on a heli shoot to use for tracking.

Like Barry said, it's a matter of each frame being slightly different from top to bottom, which throws off a tracker's calculation because the pixel/subpixel positioning. Seeing how the RED is a CMOS camera, I'm wondering how they approached this. I have seen successful tracking tests with the RED and I will be getting some raw footage to test on soon myself.

All this aside, the HV20 is still one of the best bang for the buck HD cameras out there, and the shutter problems are really only visible on fast moving footage.
 
Barry_Green said:
Not if you shot on an HV20 or other rolling-shutter CMOS camera. You can stabilize CCD footage if you use something like Dynapel's software, but you can't stabilize CMOS rolling shutter footage.
Click to expand...

You sure can stabilize CCD footage, BUT, it won't look good if you shot the the footage using a slower shutter and the footage has motion blurring in it. Its hard to explain what this looks like, but it can be pretty ugly and unnatural looking. If you know you will need to post stabilize, a sleight increase in shutter speed can go a long way...

Another factor is how much you want to resize the image. Correcting for really shaky footage means having to move the frame around in all sorts of ways, how much depends on the nature of the footage. The only way to keep that from showing up is to enlarge the stabilized image and crop.

IMHO, spend a little more time/money and get it right on set.

Just my 2 cents.
 
Not ragging on the HV20. It can make brilliant footage under the right circumstances, and it's $689. $689! No digits missing there... it's only three digits of price. But there's no free lunch either. If you want a camut era that performs well under all circumstances, don't try to skimp by with a $689 consumer cam! That's all. You rarely get what you pay for, byou most certainly don't get what you don't pay for.

Outdoor, bright light, steady camera, the HV20 is brilliant. Indoors in low light, handheld, in bumpy/vibration circumstances, the HV20 is hideous. Know its limitations, know what it can do well, and don't ask it to do things it doesn't do well.
 
Oh well, she has served me well. I never noticed the flaws until I started messing with the pulldown. Never had to do a motion track with it though. So I am assuming you can't track it at all then...

That's unfortunate. Good thing almost everything I have to motion track is telecined 16mm. Poor HV20. I was going to test your tracking capabilities this week....well I will still run my tests, and see how it goes. Can't hurt trying.
 
You must log in or register to reply here.
Back
Top