Timescapes - Digital Timelapse Discussion

TLA: Time-Lapse Automaton Prototype Stage 1 Complete
Page 1 of 1

Author:  shutterdrone-old1 [ Thu Aug 28, 2008 8:09 pm ]
Post subject:  TLA: Time-Lapse Automaton Prototype Stage 1 Complete

For the DIY'ers, I've finally completed stage 1 of my prototype motion control system. Here's a quick look at the setup (still have to add the tripod mounts at bottom):


There're more photos on my blog ( http://roamingdrone.wordpress.com ), and I'll try to get a post in tomorrow with a lot of details about how the software and hardware work together.

I'll also post a series of notes on problems one will likely encounter doing some of the more complex activities (e.g. using i2c/TWI between two arduinos on a 10' cable, managing timing, etc.) in such a task.

This weekend I should have some test videos up, but so far, I'm very impressed with the smoothness of motion (in some cases, it's too smooth/slow, I may switch the gearing on the truck motor from 3.33:1 to 1:3.33.) - I just don't want to post messy videos of my home *grin*. I need to add the tripod mounts, which should only take a few minutes and then I'll burn up some shutter life taking videos of the garden outside showing different speed settings.

It's worth noting that it only works in bulb mode right now, as I have to know when the shutter is closed before moving the motors (this allows for 'perfect' motion control with no change of vibration or movement blurring a photo), but I'll get into the details on how I'm going to enable three different modes later (camera-controlled shutter with fixed shutter time, camera-controlled shutter with 'auto-exposure' compensation for light changes, and fully managing exposure time over light changes from the controller.)

Any feedback would be greatly appreciated!


Author:  timescapes [ Mon Sep 01, 2008 4:08 pm ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Looking forward to the vids!

Author:  shutterdrone-old1 [ Mon Sep 01, 2008 5:24 pm ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

timescapes wrote:
Looking forward to the vids!

Should have some test shots up soon - did some motion tests this weekend - first run, had some software issues that had to be worked out, and second run yielded some decent results (should have a 'test reel' together tomorrow or so), but really blew some of my assumptions. Like, proper gearing for horizontal motion, and the 'smoothest' panning motion, seems best to move every other shot - moving every shot, or at any rate slower than every two shots seems to result in jerky motion, could just be the quick and dirty rendering of the video though, heh.


Author:  shutterdrone-old1 [ Tue Sep 02, 2008 7:53 am ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Ok, so here's the first test video: http://roamingdrone.wordpress.com/2008/ ... est-video/ that blog posts talks about some design assumptions that didn't pan out, and some flaws too (really need to get an anti-backlash nut if I'm going to use a worm-drive... :oops: )

I don't know, maybe I'm being too critical, but I don't think it's nearly as smooth as it should be, and I'm pretty sure that near the end when it switches to every second shot (still trying to find the part where it spent more time in a 2-shot cycle, the camera rolled over shot numbering on me, and messed up my notes) it's completely un-usable, but any feedback would be appreciated.


Author:  timescapes [ Tue Sep 02, 2008 8:20 pm ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Yeah, that is unacceptable. But guess what? All of timelapse shooting is a matter of trial and error. So just keep working on it and perfecting it. The amount of movement is too much per frame, which results in that jerkiness. My mumford head does the same thing if I set it to pan more than .125 degrees per frame, on a 10mm (APS-C) lens.

I think using a cable instead of the threaded worm drive will help. I also looked into some linear stepper rails a while back....


But I have no mechanical or engineering background, so it's tough for me to build this kind of stuff.


Author:  shutterdrone-old1 [ Wed Sep 03, 2008 6:43 am ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Well, those linear steppers like that are horrendously expensive. Most I've seen come into play around $1,000 USD and go up from there. Just using guide blocks and rails like those available at McMaster Carr can run into the several hundreds (guide blocks alone being > $100 USD) and you still need an actuator.

I fashioned myself an anti-backlash nut out of a second flange nut with a screw and a few nuts between to create an adjustable-tension setup, and it removed most of the backlash out of the drive-end, making lateral movement possible. (It was not possible before as the backlash amount was greater than the minor movements I was trying to make.)

However, imperfections in the drive train end-positions (i.e. being exactly 90 degrees and exactly same height and exactly same horizontal position) and the less-expensive roller bearings ($4 each vs about $10 for precision bearings) cause a minor up-and-down fluctuation when the drive-screw rotates. I had figured to compensate for this by using a slide mechanism in the connection between cart and drive-nut, but it turns out that there are imperfections in the cart as well. Horizontal motion is now smooth, except for the wobble from these imperfections - oops! =)

I saw the cable idea from the time-machine site, but to be honest, I stopped liking the idea of a fixed track length (be it by cable, or otherwise) pretty early on in the development of my prototype. I think that with the problems I'm encountering here, if I can't get the wobble out of the current drive done in the next day or so, I'm going to begin my next stage prototype, which does not require fixed-length tracks. It will be self-propelled, and use sections that can be added or extended, possibly even while in-motion, letting one do long runs with fewer track sections. This one has a much greater chance of success because requires far less perfection in the overall drive-train.

My goal is to have a complete DIY guide to building such a system, including which components, places to source them, and the code for the microcontrollers all being open-source. The current look is that for those with less money, but a lot of sweat-equity available, you could build one for around $500 and provide motion in all three axis. I just want to work out all of the issues first, so I don't have others going down blind paths with me and burning the sort of money I have on this project *grin*


Author:  timescapes [ Wed Sep 03, 2008 8:42 am ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Self-propelled with no limits on track length is very appealing if you can pull it off.

Author:  Antz [ Wed Sep 03, 2008 1:59 pm ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Well, for something cheap you could stick a Milapse head on a cart.
For driving the cart, the motors used on Orion telescope drive kit are not that much, and the standard controller could be adapted for a variable speed control easily enough.
It would not let you stop all motion between shots very easily, but you would have full pan tilt & tracking control.

Author:  matt b [ Wed Sep 03, 2008 5:19 pm ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Landscape Dolly 1.jpg [60.64 KiB]
Downloaded 1321 times
Be aware that exposed drive mechanisms eg. linear drivers, screw threads, etc. can give you problems if used on location from dirt, grit , rain and so on. I assume you will want to shoot outdoors in varying weather conditions.

A couple of years ago I fabricated my "Landscape Dolly' a wooden platform type style- tripod leg locators with a tie-down and an optional industry standard mounting adaptor. It is a two-person set-up, but I have on occasion put it together myself (they don't call us time lapse shooters for nothing - it takes time!)

It is driven by a ridiculously simple bicycle chain fixed at either end of the track. The motor drives a regular bicycle sprocket wheel which is hidden under the platform. The track is old Elemack pipe track that you can buy cheaply as most grips use precision type track these days.

This is obviously 'overkill' for a dSLR and was built to handle traditional 35mm cine cameras.
I'm currently building a dSLR friendly system (lightweight+portable) and will keep you posted on its progress.

Author:  shutterdrone-old1 [ Thu Sep 04, 2008 6:56 am ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Well, I got some better results by removing the wheels - it turned out the screw heads on them were catching when the cart would go out of alignment. I fashioned some friction sliders out of 8 l-brackets I had laying around. Of course, it'd work a lot better using some grooved idler rollers, but I had none of those handy at the moment.

Now, of course, Vegas is being all screwy on me, but the video it produced already showed some 'catching' issues. Ahh well. I'm starting to CAD up the design for the next prototype, other than the housing/enclosure there shouldn't be much I'll need to buy for it, so it should go quickly if I make my designs correctly.

Anyhow, regarding how it would be self-propelled: I would use an alignment/gripping system common to most dolly carts, e.g. the 'skateboard' wheels design of angled casters on either side on track sections, and the difference between many existing systems is that it would use a friction wheel to propel it on a third track section. A stepper will drive the friction wheel, and depending on what existing friction wheels (e.g. wide, soft rubber) it would be capable of moving in very small increments, or very large ones. (If I can only find larger wheels, I would gear the steppers down further... Increasing the cost. I have some friction wheels out of an old printer, but they have an unusual bore size which I do not readily have any rods that they will work with... Reducing their utility.) It would, of course, use all of the existing code which I've written - requiring no new effort there.

The track sections would be highly inexpensive using common parts, but would required either brazing or welding to attach the three sections to each other. I'm currently investigating, as part of the design of the new prototype, the cost to have them produced for me. (As I'm likely going to to have the body parts of the new device made for me as well, to reduce my effort, and increase the likelihood of everything working together =)

Regarding exposed drive-train elements, yes, I am aware of these issues. Which one of the reasons I early-on decided that the worm-drive mechanism wouldn't be a great choice for the long-run and decided to quickly start on a new design after I had completed this. (I wanted to play for a while with it while I worked on a new one =) As much as possible, I will be using Delrin gearing to increase the longevity of the drive-trains (delrin wears less than non-hardened metal gears, and no gear will have so much torque on it to require metal gearing. Hardened metal gears are absurdly expensive, and most would have to be custom-machined.)

I also realize that a person could hack up all sorts of ways of achieving similar goals, but my point is not to compete with them on a "amount of effort required" or "minimal dollars required" front, but to provide the blueprints (and maybe even kits, if there's enough interest) for building a system that solves many of the problems you might encounter. One design point I haven't talked about yet, but I've got all the hardware for in my hands, is making it capable of being remote controlled from up to 300' away, using cabling, for very little cost (we're talking about $5 extra cost, plus the cost of cabling). Wireless is completely possible, but it changes the microcontroller needs, and would have to consider other microcontrollers, increasing my personal time investment.

These designs, and all of the code will be freely available and open-sourced. The software will be GPL, and the hardware designs (PCB designs and CAD drawings of hardware) would be freely available for personal use, and welcome to change. (I would be hesistant to license those parts in such a way that someone could just take them a fabricator and produce a complete product for sale without even so much as mentioning me.)


Author:  shutterdrone-old1 [ Thu Sep 04, 2008 10:55 am ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Actually, after doing some deep searching - I'm pretty sure I'm going to go with this setup as the base motion system:

http://www.vxb.com/Merchant2/merchant.m ... ionSystems

That's the cheapest I've found by far for a 55" setup. (On the order of hundreds of dollars.) Admittedly, it still requires a friction plane somewhere for driving it, but that could easily be crafted with an out-rigger of sorts using some brackets and flat aluminum plating that the friction(drive)-wheel rests on.

Like my current setup, it could use tripods for a base, allowing it to easily be height-adjusted, etc. (2 tripods should hold that setup) - depending on how clean the ends are, another section could be added by butting it up and attaching w/ a plate or some such.


Author:  Antz [ Thu Sep 04, 2008 12:15 pm ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Have you considered a rig where the camera is mounted on the rail (with the gliders anchored to a base unit), then moving the rail forward?
It could make great push forward shots not having the track in the foreground.

Author:  shutterdrone-old1 [ Thu Sep 04, 2008 12:55 pm ]
Post subject:  Re: TLA: Time-Lapse Automaton Prototype Stage 1 Complete

Antz wrote:
Have you considered a rig where the camera is mounted on the rail (with the gliders anchored to a base unit), then moving the rail forward?
It could make great push forward shots not having the track in the foreground.

Ahh, hmm.. Interesting. I hadn't considered setting it up "upside down" like that. Considering that I am planning on making the truck unit separate from the pan/tilt unit (so I can take just the pan/tilt unit when needed -- the controllers and such would be in the pan/tilt unit, and just wiring run to the truck unit) it would be possible to take the p/t unit off, mount it on the bottom, and then attach the truck unit like normal to the track, but on the "top" attach it to some tripods. However, you'd have to make sure you counter-balanced it all to keep it from flipping over when the glide rod/camera/pt unit get far enough out from the center of balance =)

I had considered adding a 4th axis though, to rotate the entire track assembly, this still has balance issues, but would allow movements where one "goes partially around a target" (e.g. a truck/dolly arc) without actually moving the footprint, or having to have a curved track. It would, however, require good pre-programming skills to make sure your pan movements were in-sync with truck and rotate to make the effect pull off seamlessly.


Page 1 of 1 All times are UTC - 8 hours
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group