I also bought DynaPel's SteadyHand some time ago, but I get better results using deshaker in VirtualDub. I have also written a little script for Vegas for deshaking using deshaker. Using this script you select events on the timeline. The script will render every selected event to file, run deshaker on it, and add the deshaked file as take to the event. The script is currently part of a bigger project, but if there is any interest, I can split the script apart an publish it.
I'd definitely be interested in a copy of that script, if you find DeShaker to be worthwhile.
I've got some handheld helicopter footage which isn't too bad to start, but could use some "smoothing out".
One of the problems with HDV and image stabilization is that the HDV compression system sure doesn't like the shakes. Once the damage is done in the camera nothing can be done to undo it.
I'm not saying you cannot stabilise the video using tradition methods, just that there's now another layer of issues to consider.
I'd also add that shaky 16:9 is visually worse than 4:3.
A gyro stabiliser would seem prudent when shooting HDV from copters and boats.
Deshaker is a VirtualDub plugin. It won't work with native m2t clips, only Cineform avis. To use it, you need to buy Connect HD so that you have access to the Cineform writer outside of Vegas.
Once you've done all that, be prepared for some long waits on both passes. It just takes forever, at least on my lowly P4 3.06. The end result however is great. If you are going to SD, you don't even take a resolution hit since there is still extra resolution after "deshaking".
The end quality of Deshaker is a little better than the Dynapel program. It takes longer and you have to manually do two passes, but Deshaker has two things that make the end result look better than Steadyhand. First, it handles interlaced footage better. It actually separates the fields and corrects each field separately. Second, it fills in the missing edges with data from previous frames. This doesn't always work, but it works often enough to be quite useful. It can also zoom like Steadyhand or do a combination of zooming and filling in edges with previous frames.
As was mentioned, there is also the mpeg component to HDV and you can often see artifacts when the camera is shaky. Deshaker can't do anything about this, but for gentle shakes it still works well.
My only complaint is how long it takes, and the fact that you can't batch file it due to the two passes and single log file. Because of this, I go out of my way to get as steady footage as I can. I always use my Spiderbrace instead of going handheld now.
Could you give me a short run-through of the workflow?
Just the generalities of the conversion process... what is Virtual Dub and how does DeShaker fit into it?
What do you take out? And how do you get it back to HDV?
Vic,
I've used DeShaker on handheld HDV tracking footage shot with an FX1E and I was very pleased with the result ("surprised" might be better in the circumstances). I had to do walking/handheld and (as expected) the "rushes" were not good (mild statement). After DeShaker I was asked how I was able to dolly in the circumstances, which caught me by surprise because the result still showed motion variability. I guess the questioner simply was recognising that the shots were too good to be handheld. I've also processed handheld shots that had slight movement and they come out as good as if on sticks. So your helicopter shots will be a cinch. DeShaker gets confused trying to adapt to significant changes in the scene and some wonderful image distortions can be generated when something big and different enters the frame (eg. a door frame).
>Deshaker is a VirtualDub plugin. It won't work with native m2t clips, only Cineform avis. To use it, you need to buy Connect HD...
I sucessfully used Deshaker on my 1080i HDV footage and I do not own Connect HD. I saved the full-resolution (1440x1080) output with the HuffYUV codec (4:2:2, lossless compression). I spent some time at lower resolution tweeking many of the parameters for best results prior to the full-resolution pass. The segment of raw footage I started with was severely shaky. With some trial and error I was able to get very good results, I would say bordering on the remarkable. Note, my final output was standard-def and I did not view the results on a HD monitor.
The nice thing about Connect HD is that it is a high quality, frame accurate render of the m2t compression - which is what you need in order to deshake. Deshaking compares individual frames to compile a smoothed framestrip. Plus, when you bring the footage back into Vegas you can save it back out of VDub into the Cineform Codec 2.7, it's a highly accurate HDV image.
Aside from the very effective HDLink capture utility, that alone is worth the price of admission.
Thanks all - this is really great - not to digress, but I can so clearly remember when embarking on a journey like this was a painful solo venture.
Now Darryl - two questions
first I've already captured a bunch of M2t footage. Is there a Connect HD conversion, or do I have to recapture? and is VDub another part of the Connect HD?
If you have the Connect HD 3, you can batch convert the m2t clips, but once you see how long Deshaker is going to take, you might not want to do them all :-(
Also, there are a couple of issues that took me a while to figure out. You have to enter an audio offset so that your picture and audio line up after you're done.
Also, you lose about a half second at the tail end of each clip. You can get around this by adding a half second of black to the ends for those clips where the action goes all the way to the end.
Quite frankly, the whole thing is so time consuming that I try to avoid it whenever possible.
Now can you "pick and choose" the areas you'd like to Connect HD > convert > DeShake ?
That is, could I pre-cut a "shaky" edit, then select by time code or otherwise, the areas I want to stablilize?
========================================
BTW - I'm definitely going into this jungle, and I'll be reporting on all of it, as it comes to pass.
But I've said it before - "Experience beats Theory Everytime" - so thanks to all contributing to this most interesting thread.
I'll tell you - today I spent an hour above South Florida in a small four seater Raven R44 helicopter. Shot my tests with a handhelp Sony Z1. No doors on a golf cart with Rotors.
The footage is great, but too shakey for professional presentation. T
Thus these questions...
It's all about money, of course. This heli was the absolute cheapest way to go - "only" $500 an hour. The next step up is $875 per hr., with a built-in 5 gyro lens pod, controlled from the inside of the copter via TV screen and joysticks. Defintely worth the extra bucks, BUT SD. A similar HD setup is $2,500 per hour. Now if there was EVER a NEED for HD footage, it's in magnificent aerial scenics.
So where do you draw the line? $500 bucks ain't cheap, but it's "do-able". $2,500 bucks takes a nice bite out of anybody's budget (or more realistically - profit). Since this is the age of "Digital Enlightenment" I started my tests, and gambled the $500 bucks.
I tell all this, because I'm often asked my workflow and thought processes in production - Money Spent vs. Value Received. Plus I have the advantage of lots of actual experience in the "high price spread" - Jet Rangers with Tyler rigs in 35mm. But I have devoted my "new life" to DV/HDV production techniques. It's been very rewarding so far. Todays hour in "the old days" would have been easily $5,000 or more. What freedom - if we can just make it work.
So - this journal will be continued - with commentary from a unique cadre of specialized experts. You see - these days things are changing So Rapidly - that anyone with virtually ANY experience, is an expert. (No slight intended - I am totally grateful for your input, and mean what I said above).
These are great times to be alive - filled with potential, excitement and discovery.
Victor,
get a Kenlabs gyro system, the difference in the price of the copters will pay for it. Better still rent one, if you were down here I could probably get you the gyros for free.
Another tip from someone who does a lot of helicopter work (not me).
Get a bungy cord to hang the camera off. With the right setup it can work wonders, do it wrong and it can make matters worse, definately you need to experiment, try adding weight to the camera.
Actually we had a discussion on helicopter mounts about 12 months ago. The problem with helicopters is that blade lift varies cyclically as each blade changes from advancing through the air flow to retreating, so with a 2 bladed rotor this cyclic variation in lift (and hence bobbing up and down) is quite noticable. What you need is for the camera to stay in the average position of the fuselage, which is helped by lots of inertia and some soft support. Which is what Bob has said in fewer words.
Vic, you'll be converting all your footage to DI anyway. Use John Meyer's "how to" and try it on one clip. Remember it requires 2 passes (the first analyses, the second processes) and don't worry about the audio being shifted and the 30 frames leader and trailer (you need to work through those once comfortable with the process). But of course, not needing to do extra processing is the ideal solution.
Now here's a thing.
My helicopter owning mate tells me the 4 blade / 2 blade thing is not true. I didn't really grasp his reasons as to why but he was pretty definate about it. I think what he was trying to say was a 2 blader might be better than a 5 blader, it depends on much more than how many blades.
Well the more blades the more evenly lift is distributed, so I find that surprising. For a thought experiment start by consdering a helicopter with a single blade.
Trust me, I don't get it either but the guy owns a copter (Bell Longranger) and he tells me way less vibration than the Eurocopter Squirrel which is 3 blades. The Squirrel shakes so much the dash is shock mounted plus it's a beast of a thing to hover / land.
Maybe where he's coming from (other than experience) is that all else being equal yes more blades would be better but there's other source of vibration. The design of the gearbox, how the engines are mounted, what sort of engine(s) they are etc. Plus aerodynamics would have an influence as well.
If I get a flight in his machine I guess I'll be able to report back.
Doing some 'community service' work might be rewarded with an opportunity to shoot some Sydney aerials. Funny how things can come together.
Bob,
your buddy is wrong as a general statement, and regarding for camera work. It may be that a high end 2 blader has less vibration than a lower end 3, or maybe how tweaked the machine is, but as a general rule, that's just not right, IMO. Having shot in 2, 3, and 4 blade machines, there is no comparison, and with a gyro, it's still better in a 4 vs anything else. This one of the reasons the Artbeats product is so much better than most anyone else', for instance. I've experienced this myself, but a long discussion with Phil Bates about what/why/how much his heli rentals are brought this out from his hundreds of hours of shooting handheld, mounted,and gyro'd with Beta SX, Beta SP, 35mm, 16mm, and HDCAM formats with a wide variety of lenses.
Additionally, one of the helicopter companies that exhibits every year at NAB has video showing why they are so much "better" with their 3 and 4 blade vs 2.
It's true that the heli can be tuned to have a smoother rotation at "x" speed vs another speed, and that a 2 can be made as smooth as a 4, if you don't mind lower speeds.
We're working on a project now that is almost exclusively all heli-shot due to the nature of the subject. I'll show some of it at the Sydney VASST if you'll be making it to one of the days.
You can Google Deshaker and get a large number of hits regarding downloads and usage.
Vegas 7 already comes with a Cineform codec, but I believe the Cineform Connect package includes an improved version - which will autoload should you go that route. More Specific info here - Connect
BTW, I'm running VirtualDub 1.6.16, mostly cause I'm too lazy to update to 1.6.17.
I am currently in progress of extracting my deshaking scripts from the bigger c# project and also convert them to jscript for more compatibility with different versions of vegas. I will also write some documentation and create a webpage for it. Expect this to be finished either this evening or tomorrow. I will post once the scripts are published.
A deshaking script for Vegas would be great except that there would be no way to deshake the two interlaced fields separately as is done with VirtualDub and Deshaker. This is something Deshaker does exceptionaly well.
VirtualDub is free.
HuffYUV is free. Get v2.1.1, not v2.2.0.
No need to spend money to try it out. Both support HDV resolution. John Meyer's tutorial is a great resource. There was even a VirtualDub script posted in this forum if you get daring enough. You can find it ==> here <==. Note that this script is for SD. But using the manual VirtualDub process John Meyer describes works great. I've had some problems with still footage getting messed up, so I generally put both original and deshaken on separate timelines and pick and choose which one to use.