Latest on my F3 Motion Control Rig

[Aaron Newsome]

A few of you have seen some progress along the way. I thought I'd give a more detailed update on the rig.

What originally started as a cheap wireless follow focus project has now turned into a 6 motor full blown programmable motion control slider rig.

Originally, I just needed a cheap way to use my iPhone or iPad as a focus controller. Everything out there was too expensive so I built my own and wrote the software to make it all work. After seeing how easy it was to do focus and iris, I retrofitted my Kessler head to do motion control (added encoders for absolute positioninig) and added a linear motor.

Then I ended up replacing the Kessler with a pan tilt and roll head from servocity.com. I pulled out the DC motors and used stepper motors for all 6, that is:

Pan, Tilt, Roll, Linear, Focus & Iris.

The software, I wrote myself. Currently it is pretty full featured but lots more to come. I can use any laptop or an iPad/iPhone to run the rig. It has a concept of presets for programming moves. Sequences for recording and playing back moves. All motor speed, ramping and braking are stored with each programmed moves. Moves are precise to the millimeter and frame accurate for playback.

Anyway, here's part 6 of some video I made about the rig last weekend. It's showing playback of a sequence.

http://dl.dropbox.com/u/5506543/Moti...280x720%29.mp4

Right click and download it if you want to check out the rig in action.

[Chris Medico]

That is flat out awesome.

How does the rig work if you have to do faster moves? Do the motors make enough RPM to to follow faster action?

[Shaughan Flynn]

That is just awesome. Are you able to ease in/out on the servo moves?

If you are disinclined to answer this next question - I totally understand: How much $ have you put into it so far?

And, finally, will you make your app and or it's source available?

Great stuff!

[Alister Chapman]

Great stuff Aaron. Which controllers are you using?

[Aaron Newsome]

@Chris - the motors are all very powerful and can move very fast, probably faster than you'd ever want to swing a camera. Most of the work I've been doing with the rig are smooth moves not fast ones. However, if I used the rig for a rock concert and the lead singer is jumping around like a maniac, I'd probably have to crank the speed up. For that I'd want to hook up a physical joystick for action that fast. Not my ipad, or computer.

@Shaughan - for the easing in/out, I call this speed ramping and braking in my software. This single feature puzzled me more than anything so far. I finally settled on an excellent PID library that I found written in C. I rewrote it in Java since my software is written in Java. Speed ramping works REALLY REALLY well for making "cinematic" moves. The camera starts at zero speed and a small period of time, will ramp up to target speed. How long it takes to get up to speed is the "ramp". Although the curve remains somewhat constant currently.

As far as the cost, the parts are not that expensive. Most of my money was spent buying the wrong things and having them not work. If I only add up what is currently being used on the rig, it's probably around $3k. The speed rail rig/platform and pan/tilt/roll head were kinda pricey. I could definitely figure out how to do these cheaper.

@Allister - The single board computer that is riding on my rig, and all the stepper controllers are Phidgets. yes they are VERY expensive compared to something like the Arduino controllers I've seen for $15 or $20. The Phidgets controllers are $75 and the stepper motors are around $40 each so they're kinda pricey too.

[Aaron Newsome]

When my friend showed me his rig, he demonstrated the nickel test and I was pretty impressed. My rig originally failed the nickel test because the starts and stops had too much backlash and the starts were pretty abrupt.

After I worked out the speed ramping, my rig passes the nickel test with flying colors. This is a real test. Right click and download the video to see the test:

http://dl.dropbox.com/u/5506543/Moti...ickel-Test.mp4

[Chris Medico]

Nice!!

I would love to build one if you ever make the plans available.

[Aaron Newsome]

I'll be sure and post the parts list on here. As soon as I work out a cheap platform alternative with wheels, as well as a cheap sturdy replacement for the servocity pan/tilt/roll head. The servocity pan/tilt/roll head is about $1600 by itself.

If you can afford that, then yes you can probably build one of these.

Currently I'm trying to drum up some interest on the software side though. I'm not much of a programmer so my code is really not optimal. Plus there are some motion control tricks that I can't figure out by myself so I'm thinking if I can drum up some interest, opensource the software and then get help from better programmers than me.

Trust me when I tell ya, I've got some motion control ideas that are currently not implemented in ANY rigs that regular people can own. That's my goal, to get some of this stuff implemented.

I have zero interest in timelapse photography, but I know a lot of people do so I would be looking to put that in my software too. Currently my software is all about real time control, and sequence playback of moves for motion picture camera. With timelapse it's all about move a little, snap a pic, repeat.

[Aaron Newsome]

Also, here's me running the rig with a WiFi iPad:

http://dl.dropbox.com/u/5506543/Moti...280x720%29.mp4

[Matteo Ricchetti]

Hi Aaron, very good work,
I made many rigs in the past, but I left stepper motors versus Pid controller with optical encoder, my probelm is findind good gear boxes without backlash.
Can you please tell me wich kind of gearboxes have you found?

If of any interest here you can find some of my stuff

Slider autopan on Vimeo

Canon Ef lens remote control on Vimeo

this was made with stepper motor (2002 -mini dv camera)

Rotocamera - YouTube
Matteo

[Aaron Newsome]

Hi Matteo. I previously used DC motors and encoders but stopped using them. I now use six stepper motors instead. I also use a pid controller in my software. This how I get the smooth starts and stops with the steppers.

My steppers have the gearboxes built into them so the setup is pretty simple. The linear stepper is a 5:1 and the pan roll and tilt are all 26:1. I'd probably be ok with all 26:1 though. One of the things I learned while buying all the wrong stuff at least once for everything.

I'll have to check out your videos when I'm not on my iPad. Nice to meet another rig maker. Maybe we could exchange some ideas. I've never really looked at any other rigs or software when I decided to build mine.

[Aaron Newsome]

I was able to check out both Vimeo videos. Very nice stuff indeed.

[Aaron Newsome]

Here's an update on my rig. It's the very first look at my "auto tracking" feature for the motion control rig. It works like this:

Point the camera at something and focus it.
Click the "Lock Target" button in my software
Move the camera on the slider in any direction, at any speed

That's it. The camera will stay pointed at the object (the pan and tilt) and the object will stay in focus (the focus stepper). Doesn't matter where the object is, or how extreme the pan or tilt is, my software will try and solve it. Well, I take that back. Really extreme angles will not be solvable, for example if the object moves directly underneath the camera platform, there is no way to see it (unless of course, it's overslung).

Anyway, this is the first of what I consider "really cool stuff". I don't want to let too many cats out of the bag, for what's coming next. But I do have two pretty big ideas for this linear slider rig before I move on to rigs that can position the camera on the z and y axis'.

Anyway, here's the auto-track in action:

http://dl.dropbox.com/u/5506543/Moti...l%20Update.mp4

if you want to see the camera rig actually doing this move, try this one:

http://dl.dropbox.com/u/5506543/Moti...%20Move%29.mov

Also, I've already figured out a way to do this tracking even better. I currently constantly solve for where the camera should point, and then tell the camera to move there. It works but it's not terrible efficient, and generates a lot of "events". I have a better way to do this, the solving math was the hard part for me.