Flying in Portland (Maine)

[Brian Bechard]

Just trying to get the hang of this thing. Ultimately I am looking to be able to nail a couple of shots for wedding videos and commercials/web videos for small businesses. This was shot on a GoPro Black+ and a DJI phantom 1.1.1

Any feedback is welcome. I think I'm close to getting what I want, but the IQ of the GoPro is a little troubling for me right now. I really would like to fly my 5DIII, but I don't have the intestinal fortitude or the budget for that right now.

Phantom practice on Vimeo

[Tim Lewis]

Brian It hunk this was really well done. I liked you mix of low altitude shots with objects in frame actually reaching the top of the frame. It made for a pleasant change from all the bird's eye view stuff often seen from copters these days.

Did you shoot 60p then slow to 30p?

[Brian Bechard]

Thanks Tim, I shot most of the footage at 60fps and then slowed it down in FCPX. Since I am still experimenting I did shoot some in 27K at 30fps and really didn't see much of a difference.

The GoPro is a great camera, but it is a little undermanned for this kind of shooting. Unfortunately the cost of flying heavier cameras is beyond my means at the moment. It's still fun as hell though!

[Jody Arnott]

Great video, really nice flying.

The only tip I have: give 1440p @48fps a try. It gives you the ability to reframe your shot, and also slow it down by almost 40-50% in post. 2.7k is nice for reframing, but I find that it's crucial to be able to slow down the footage to smooth out any bumps/jerkiness.

[James Manford]

Nice shots ... My eyes stuck to the copters shadow on the ground around the 50sec to the 1min mark of the video :D

[Chuck Spaulding]

Nice flying and AP.

There is a benefit to shooting 2.7K but it depends on how much post processing you want to do. I pretty much run all of my aerial through AE for warp stabilization, I Pan & Scan the 2.7K in a 1080P composite to get the framing I want and then remove the fisheye.

It makes it look much better bt its still a GoPro..

[Roger Gunkel]

Nice stable shots and flying :)

Roger

[Brian Bechard]

Chuck could you elaborate a little on your workflow? I haven't used AE's Warp Stabilizer much. I edit in FCPX and have been using the stabilization built in to the NLE.

Do you miss the ability to slo mo the footage using 60fps? I slow down almost all of my aerial footage as well as adding stabilization.

Some of the shots in the video are 2.7K and I can't tell much of a difference. I would like to utilize that extra resolution, but I'm worried about losing 60fps and my lack of AE skills is holding me back.

[Chuck Spaulding]

Originally Posted by Brian Bechard

Chuck could you elaborate a little on your workflow? I haven't used AE's Warp Stabilizer much. I edit in FCPX and have been using the stabilization built in to the NLE.

Do you miss the ability to slo mo the footage using 60fps? I slow down almost all of my aerial footage as well as adding stabilization.

Some of the shots in the video are 2.7K and I can't tell much of a difference. I would like to utilize that extra resolution, but I'm worried about losing 60fps and my lack of AE skills is holding me back.

Hi Brian, a little explanation is in order to help you decide which workflow might be best for you.

Shooting 60fps is a must without a gimbal but if your using a good gimbal that pretty much keeps the horizon level and steady then I find the reliance on 60fps isn't necessary. Also its important on how you convert your 60fps footage into 24 or 30fps. If your simply dropping the 60fps footage into a 30fps timeline, depending on the preference settings of your NLE its probably re-encoding the footage and using frame blending. That generally takes longer to render when you export and often has motion artifacts, you might notice a slight shudder in the footage etc. I use a utility called Cinema tools, it simply changes the header information of each frame so that it plays back at the frame rate you tell it to without encoding or frame blending so there is no render penalty on export and no motion artifacts.

I don't use FCPX, I'm still on FCP7x so when I talk about workflow its in context of FCP7, I'll try to point out where there might be some differences. I often use Smoothcam un FCP7 for shots that might be a bit shaky and sometimes it does a great job, however the real downside to smoothcam is that if I want to use ten seconds out of a one minute clip it has to analyze the entire clip which more often than not is a real waste of time. When using the Warp Stabilizer in AE you can restrict the process to, in this case, a ten second region of interest.

So once in AE create a new composite, 1920x1080 that matched the frame rate of the clip your placing in that composite, then drag the 2.7K clip onto the new composite timeline, obviously it will be much larger. Go to Effect>Distort>Warp Stabilizer VFX drag and drop it on your clip and it will begin analyzing. Like I mentioned earlier you can set the in and out with [command-B] sets the in and [command-n] sets the out. Also in the controls panel for the Warp Stabilizer lower the smoothness setting from 50% to 10-15%, in the drop down for the Borders control you can adjust the auto-scale, after analyzing the footage it will set it for you and you'll see what it set it to in the parenthesis. If you raise the Smoothness setting it will increase the amount of scaling, I try to keep that percentage under 105%, you can change the Smoothness setting at any time without having to re-analyze so experiment with the setting to suite your own taste.

Everything you do in image processing is a tradeoff between time and quality, the Warp Stabilizer is optical flow, it uses some number of frames before and after to calculate stability based on the smoothness and method settings. Its a very good stabilizer. The Smoothcam stabilizer in FCP7 is temporal but not Optical Flow, the Stabilizer in Motion is Optical Flow and good but REALLY slow, I don't know what version made it into FCPX. The good and bad of it is that because they're different one will work well on a clip that another doesn't, so its a bit of trial and error and you'll start to get a sense for what stabilizers work best for a particular type of motion. Like multirotors, there's a lot of variables so its not easy to say if its X use Y etc.

After I Stabilize I add the Optic Compensation filter to the clip to remove fisheye, I'm not at my computer and don't remember the specific controls but its pretty basic and works very well. Then finally now that the footage has been scaled because of stabilization and lens distortion I pan and scan by simply animating the translation parameters in the clip itself. So the clip is stabilized, optically corrected and then Pan & Scanned and after all that, with the exception of it being more stable clip it doesn't appear to be that different. I think the GoPro images are a bit soft at the edge so your losing that and the fisheye, but only you can decide if its worth the effort. Probably the right answer is to not use a GoPro but their so small, easy to fly and inexpensive so unless I have someone willing to pay considerably more for better quality footage I'll continue using them and trying to improve the footage as best I can.

I edit on a 24 frame timeline so for me 60fps, that's a 40% speed reduction, is too slow, shooting 2.7K at 30fps results in a 20% reduction that slows it enough to take the edge off but still fast enough to look almost real time.

I know I've made this sound much more difficult than it is but you do it a couple of times and its pretty easy to do. This is why I don't do tutorials..

Hope it helps.

[Brian Bechard]

Thanks for the knowledge Tim. FCPX does a great job with slow mo but I hate losing resolution on the stabilization. I'm going to try 2.7k today and play around with after effect as well. I think I'm leaving some image quality on the table using 1080p 60fps. My gimbal is an XPG by Xpro heli. It's brushless and does a pretty good job in the air. I just need that extra bit of smoothing out to really make it steady.

Don't knock your skills, you might have a future doing tutorials!