All singing all dancing camera slider.

[cjdawson]

My latest project, inspired after the SGL start party, is to make an all singing all dancing camera slider for time lapse photography.   I've been thinking about this quite a bit, and it's going to be a major piece of work.

So far, I've spent quite a bit of cash betting pulling parts together, and I still have a load more work to do on that front.

 

I can't really show many pictures at the moment as all that I have to show a the moment is a shoebox full of bits.    Here's the spec of what I'm going to be building...

 

Here's my brief.

Design and construct a camera slider for a DSLR (must be compatible with both a Canon 30D and a Canon 70D - my cameras) suitable for time lapse photography, both astro and non astro.   The slider must be capable of working in the field.  It must be capable of working on the flat or at an angle.

 

I've been thinking alot about the design and I'm nailing down the features as well as designing the build.  I'm at the stage where I've not solved all the problems, but am confident enough to have ordered alot of parts.

 

Here's the current feature list of my slider.

 

1. Slide rails made from 3mm thick 30mm aluminum pipe. 2 meters long.  It will be a twin rail system.

2. Slider will be able to operate face up or face down.

3. The camera will be mounted on a rotating ball head.

4. The slider will be motorised in both the sliding motion and the rotation.

5. The slider will have a built in intervalometer

6. The slider will have safety switches to ensure that the dolly does not try to drive off the ends of the rails.

7. There will be capacity for some expansion at a later stage.

 

 

How's this all going to work?

Let's start with a Raspberry Pi 3.   They're fun.  I'm going to use Windows IoT for this.  Could probably do something with Linux, but I'm not a confident playing with the linux based programming languages.  Besides, I'm a c# developer so using windows for this is second nature to me.  I'm sure lots of people will be saying booo hiss to this, but I don't care.  I've got the tools and the knowledge, so that decision is made.  Anyways, apart from that the rest is going to be transferable to linux, so never say never ;-)

 

The slider is going to be belt driven using a 9mm HTD belt with a pully mounted directly to a Nema 17 1.8° stepper motor.  I'm going to drive the motor using the half stepping technique (might switch to full stepping if needed) which means 400 steps per shaft rotation.  

My pully has a diameter of roughly 21mm.  So that works out to about 66mm per rotation, and about 30 rotations.  30 rotations at 200 steps per rotation works out to 0.3mm per step and at 400 steps per rotation is about 0.15mm per step.

Making a 45 second time lapse with a frame every 30 seconds, at 25fps.  Means 1125 frames.  As we are talking about moving the camera over a mm per frame, I think there is plenty of head room here, even at 200 steps per rotation.  The slider will be capable of 6000 steps of tavel at 200 steps per rotation and 12000 steps of travel at 400 steps per rotation.

 

Why am I talking about 200 steps and 400 steps?  There a three ways of driving a stepper motor, wave driving, full stepping and half stepping.  For this build, I'm likely to be using either full stepping (for the torque) or half stepping (less torque, but more accuracy)

 

The rotating camera mount, is a camera ball head, mounted on a spindle.  The spindle will driving using a Nema 17 0.9°  that means 400, or 800 steps per rotation.  This time the spindle will be driven through a planetary gear mechanism.  The big gear has 30 teeth, and the small has 15.  The 15 tooth will be mounted to the motor.   Meaning that the 400 and 800 steps per rotation become 800 and 1600.  For a 1125 frame time lapse, it's a bit close, but it'll do for now.  Might look to upgrade to a motor with a gearbox later if required.

 

The motors will each be drive using a Dual H-Bridge, This will allow me to power the coils individually, or indeed turn them all off to use magnetic breaking (a no current break that makes using of the stepper motors ability to generate current)  This will let me save power over a 4-5 hour or longer time lapse.

 

I've got a couple of micro switches with bearings on them.  These will act as simple end stop detectors, which will let me stop the motor before it gets too close to the end.

 

The slider tray is going to run on two rails made from 30mm tube.  The slider will be made mostly from Aluminuim 3mm thick.  I think that should be thick enough for this project.  The wheels are made from 22mm roller bearings, using M8 bolts as an axel.  On each rail there will be 6 bearings.  two on the top, two below, and two on either the inside or outside of the rail  (I haven't decided on that detail yet.)  This is similar to a roller coaster.  Doing this means that I can have the dolly riding on top of the tracks, or indeed have it riding underneath.  This will allow more versatility than simply having the dolly riding on top of the rails.

 

At each end of the track there is going to be some recangular aluminum box tube.  This will provide the anchor points for the timing belt.  The contact points for the switches. Points to be able to attach to camera tripods, And the structural point for the two rails. If the bearings have to run inside the tracks, this will be the only point that will keep the two tubes in line.

 

The dolly itself will contain the H-Bridge circuits, it will also contain the cable release.  A 12 volt dew heater controller (I'm putting it here as that will not be PI driven).  I'm also thinking about adding some kind of light meter.

 

So, what about the software?  As if the hardware wasn't a project enough

The software is going to try to make things easy.

1. Setup the slide motion.  Using the move-shoot-move technique.  Over the course of the time lapse, both the rotation and slide motions will be run from the PI.

2. The time lapse part will be controlled by the PI as well. 

3. I want the ability to ramp the exposure time over the course of the time lapse.   This is where the light meter will come into play.   I want to be able to register the ambient light that is likely to be in the frame, and keep tabs on it  The idea is that over the course of the shoot, the exposure time can be adjusted to compensate for light levels  (think daytime to night night transition, or vice versa)

4. I want the software to calculate the longest exposure time using the 500 rule.  I'll tell it my focal length and it just works it out.  (500 / Focal length = longest exposure in seconds before star trailing)

 

So how am I going to control the PI?

 

The PI won't be on the slider, it will be in a seperate control box.  I've not designed this part yet.  The PI will be using a 7" touchscreen.  It's the one from the Raspberry PI foundation.  This device is great, it's a capacitive touchsreen, and it powered from USB. Using a 12v - 5v 3amp converter I'll have more than enough power to be able to run both the screen and the PI.

 

That's pretty much as far as I have gotten so far.  

 

The problem that I'm currently thinking about is how to do the wiring.  Currently, I'm looking at 4 wires per stepper, 3 wires for the cable release 12V power capable of a couple of amps. 2 wires for the light sensor. 5 wires for usb port and if possible a few spare for later upgrades.

The Usb port will be for a possible later upgrade that would allow the PI to change the ISO on the camera.

In total that means that I'll need 4*2 (motors) + 3 (cable release)  + 5 (usb) + 2 (light sensor) + ?(spares)

18 cores, plus 2 power.   I'm thinking that maybe the simplest solution here is to use a 25pin D-Sub connector with a seperate power line although looking at the specs it looks like 3A per pin is acceptable.

 

Things that I have not yet properly solved.

 

1. How the light measurement will work.

2. How to mount the electronics in boxes (I know that I want to have the PI with screen in a handheld unit with a lid that can be closed)

3. How bit a battery I'm going to need to power all of this.

[Alien 13]

Look forward to any results I am planning at making something similar that could possibly double up as a macro rig standing on its end for indoor use.

Alan

[cjdawson]

I'll be keeping the thread updated as I get parts, Yes there will be pictures coming too.   I'm contemplating doing something with the software to make it distributable too.  But I'm not sure about how that will be possible at the moment.  Might just make it an open source project on git hub or something.

[cjdawson]

1 hour ago, Alien 13 said:

Look forward to any results I am planning at making something similar that could possibly double up as a macro rig standing on its end for indoor use.

Alan

Got any pictures of what  a macro rig is?  I might be able to include that functionality into my built.  I want it to be capable of as many things as possible so that I don't have to modify it too much after I'm done.

 

If you are talking about being able to use it for that focus thingy (I forget it's name) where you take loads of photos with a very shallow depth of field and combine the images when they are in focus.  It'll do that

[cjdawson]

Just remember, the technique that I was thinking of it called focus stacking.   But I'm thinking that are you talking about the slider standing on it's end for indoor use, I'm not that this is the technique that you mean.

[Alien 13]

41 minutes ago, cjdawson said:

Just remember, the technique that I was thinking of it called focus stacking.   But I'm thinking that are you talking about the slider standing on it's end for indoor use, I'm not that this is the technique that you mean.

Yes its something like that, its funny but trying to search for something without knowing its correct name is quite difficult the only image I could find is this one 

but it gives the general Idea.

Alan

[cjdawson]

Gotcha.  I think for macro, it would be possible, especially if my maths are correct and that I'll be able to achieve 0.15mm per step.

[cjdawson]

Most of the aluminium parts arrived today.  This thing is going to be huge.  I've got enough parts to be able to build up the frame of the track.  3mm thick aluminium might have been a bit thick.  1mm would probably have been enough.  Oh well.  This thing is going to be able to take some abuse

I might have to think about adding some detachable wheels to one end of the track so that I can drag this thing to location.  Can see several upgrades to help with portability.  Starting to think that 2 meters might be a bit long.  Will test over the weekend and see if I can get the stuff into my car.

[John78]

2m will be too long, 1m should suffice - timelapse is slow.

 

personally I wouldn't bother making a track or dolly myself I'd just buy one of these.... For £60 you just can't go wrong...

 

http://ooznest.co.uk/3D-Printer-CNC-Kits-Bundles/Mini-V-Linear-Actuator-Kit

 

can get up to 1.5m long should you want.

 

A simple LDR should work for your light meter average it over many seconds.

 

The very best sunset/sunrise transition timelapses as I understand it are shot with canon point and shoot cameras running CHDK because it can control ISO, shutter, aperture and ND filter all automatically based data collected over the previous shots so it's perfect straight off the camera, with no external hardware, no intervalometer, it uses all the inbuilt tech gizmos, metering etc.... bramping in comparison is poor and other methods require hideous amounts of post processing.

 

Iinteresting project though, the hard bit is the software I guess a RPi could put the frame off the camera or liveview and analyse it to figure out appropriate settings to avoid exposure or DoF jumps but that's well beyond my programming skills!  I assume your intervalometer and slide controller are in the same package?  You need to msm.

 

This project had a load of potential before it got discontinued....

 

http://photoduino.com/documentation/hardware/photoduino-shield-3-0/index.html

 

 

[John78]

IGUS make a perfect camera slider also...  As used in these...

 

http://www.digislider.co.uk/collections/digislider-igus-kits

[cjdawson]

The aluminium parts arrived today, all except for the Angle piece that I ordered after revising my design.  The actuator might have been helpful, however I don't like that the motor is mounted at one end, would complicate the wiring when I want to have a motor mounted on the slider for the rotation aspect.

The rails that I ordered are 2.1 meters long.  This would give a huge amount of movement and I'm sure the timelapses from it would have loads of lateral movement.   However, I started thinking about the practicality of it once I saw the length of the pipe.  That length is going to be too long to be able to go into my car or trailer.   My trailer is 1.45 meters long, so I'm thinking that it will be ok to make my slider 1.5 meters long and put it on the roof bars on the trailer.    That will give me 1.4 meters of travel.  Not as much as I was hoping for, but it should be good, why oh why am I already starting to picture a version 2 slider that is driven differently and can run on a track that is extendible.  The idea there is poles that screw together to to add length, would be part more portable, but the drive mechanism would have to be competely different to what I'm building at the moment.

 

Anyways, this weekend.  I'm going to cut the 2.1 meter rails down to 1.5 meters  (I might even go shorter, but will decide later).  As the dolly itself will be 25cm wide, I really do not want to cut the tubes shorter than 1.35cm as that would be give me 1 meter of travel on the dolly.  135cm - (25cm (dolly) - 10cm (box ends)) = 100cm   The longer I can get the pips the better.  With the current 2.1 meter. I can do 1.75 meters of travel.

I might end up trying to alter the design slightly so that I can get more travel from the dolly.  I might be able to reduce the width of the cart for a start.  When I order the plate, it was with the idea that everything was going to be mounted on the cart, including a 7" touchscreen.  Since then, I've decided that I want to remove the PI and the touchscreen from the trolly and have it in a seperate control unit (easier to control)  So might end up cutting down the plate to a smaller size - once I've got everything laid out properly.   In the meantime, I'm not planning on rushing anything.

[cjdawson]

I'm making a bit of progress on this project.  Got the pully mounted on one of the stepper motors.  Had to drill out the pilot hole and add a small hole and tap an M3 thread.

 

Also, I've decided that I'm going to build up the whole rail system using the long pole, so the total length of this thing is going to be 2.1 meters.  that will give me about 2 meters of travel once the rails have been mounted on the rectangular box section.   Spend the afternoon drilling through holes that I'll use to mount the rails to the box section, and I've also spent time polishing up the rails.  Next job will be to clean them with soap and water, then lacquer them.

[cjdawson]

Switching tack again.   Decided this weekend to work a little on the Electronics.    Used a breadboard to mock up a circuit for the ambient light sensor.  This is using an MPC3008 chip, 8 channel 10 bit ADC.  Rigged up a Phototransister to it so that I can read the values.

I had the photo transister hooked up to a 10k resister.  This give me a result that when I shine a torch on the sensor, the reading goes up to about 600.  But without that illumilation, it's showing about 25.   I've got some more resisters on order, so with any luck, I'll be able to make it more sensitive at lower light levels.  I'm hoping that outside daylight like we had yesterday would yield a value of around 900 on the sensor, so that pitch blacknesss would yeild 0.     The ADC is capable of a range form 0-1023, I'm using a photo transister in series with a resister, The photo transister has one lead connected to 3.3v, and the other lead connected to the resister.  The resister is then connected to ground.

The adc is connected to the point between the two components.

 

Hopefully this can be made sensitive enough to low light levels to be useful.  (My plan is to have this help to decide the exposure time)

[Jonk]

Colin, were you referring to the conversation we had when you mentioned SGLXI inspiration?

Good to see someone else attempting their own - mine is nowhere near ready to call complete, but a couple of things I have learnt so far:

1. Battery power - for a long 8+ hours you'll need a substantial battery, as stepper motors can take 2A if supplied by a decent driver (big easy driver isthe one I use), so a 10Ah battery for example may not be enough. You also need to use the same battery to power the camera via a DC adaptor, as the camera battery may only last 4-5 hours tops.

2. I used a drive belt before, then found out the hard way that if power to a stepper motor is cut, the weight of the camera will simply drop and fall, the motor does not lock, so I changed to a ballscrew design to fit to my rail. I'll see how it works as I may need to add a worm drive to this, in order to prevent gravity doing its thing.

3. Add a dew heater to your camera, and power it from a controller, again from the same battery. My SGLXI test was partially ruined when dew hit, at least it was only a test!

Good luck, will be interesting to see the outcome.

For refererence, I used 2 x MiniEngine 2 Arduino kits and packaged them into a plastic box with the LCD displays on show. I'm currently testing a 12V Lithium Ion battery, but may need to use a Leisure battery externally if this proves to run out before 8 hours at full load (2 cameras, 3 steppers, dew heater etc).

 

[cjdawson]

15 hours ago, Jonk said:

Colin, were you referring to the conversation we had when you mentioned SGLXI inspiration?

Good to see someone else attempting their own - mine is nowhere near ready to call complete, but a couple of things I have learnt so far:

1. Battery power - for a long 8+ hours you'll need a substantial battery, as stepper motors can take 2A if supplied by a decent driver (big easy driver isthe one I use), so a 10Ah battery for example may not be enough. You also need to use the same battery to power the camera via a DC adaptor, as the camera battery may only last 4-5 hours tops.

2. I used a drive belt before, then found out the hard way that if power to a stepper motor is cut, the weight of the camera will simply drop and fall, the motor does not lock, so I changed to a ballscrew design to fit to my rail. I'll see how it works as I may need to add a worm drive to this, in order to prevent gravity doing its thing.

3. Add a dew heater to your camera, and power it from a controller, again from the same battery. My SGLXI test was partially ruined when dew hit, at least it was only a test!

Good luck, will be interesting to see the outcome.

For refererence, I used 2 x MiniEngine 2 Arduino kits and packaged them into a plastic box with the LCD displays on show. I'm currently testing a 12V Lithium Ion battery, but may need to use a Leisure battery externally if this proves to run out before 8 hours at full load (2 cameras, 3 steppers, dew heater etc).

 

1. For battery power, I've currently got 2*70Ah leisure batteries.  One I use to power the scope, the other is currently free for use.  My plan is to plug the power in the battery box that I use for my telescope, as even when running my scope it's nowhere near the limit.

I'm already planning on pickup up some DC Adapters for my camera's.  Being able to remove a battery from the equation is always a good thing in my opinion.  I want to do this for both my telescope and for the camera rail.   I was planning on building a power support directly into my telescope.  I'll most likely do this in the rail project first. 

2. I understand exactly where you are comming from on the belt front.  I know it's a risk that the power might get cut from the stepper.  For my first attempts, I'm going to keep the rails fairly flat, so even if it does fail, it shouldn't be to big a deal.

3. Got the dew heater covered already.  I did my first time lapse test on that same night.  Here's my result - also ended in dew ;-)

 

I remembered that you were talking about using magic lantern for controlling the exposure times.  I've had a look into it and have discovered that both my DSLR's are not supported by it.

On reflection, I really don't like the idea of hacking the firmware anyway, and have a couple of possible solutions.

1. Built a shutter release into my project (I'm going to do this as it's dead easy to do, I have the parts just haven't put them together yet)

2. Hook into the Canon camera's using the SDK from the cannon website.  https://www.didp.canon-europa.com/ This should let me use one of the USB Ports on the PI to communicate directly with the camera giving a huge amount of control, will be a bigger task though.

 

[cjdawson]

Just had a minor success.  I've got the circuit for the ambient light sensor working.   I'm using a phototransister in series with a 1M resister.  This is a basic voltage divider circuit.  The 1M resister is used to limit the current and provide a minimum voltage value.     Between the resister and phototransister I have taken a line that is fed into a channel on a MCP3008 chip.  This chip provides a 10bit ADC, so gives me readings between 0 and 1023.  1023 represents 3.3V.

With my circuit, I've got the sensor giving out values between 0 and about 850 when I shine my big maglight torch at it.  Without the torch, my living room which isn't strongly lit is giving a reading of aroun 690.  And with the light off, it drops to about 75.  If I then turn off both of my monitors I get readings between 0 and 30.  I think that readings below about 70 are full of noise.   I might have to see if there is something that I can do to smooth out the readings.

All in all, this is looking good enough to be able to use as a light level meter to help exposure ramping.

[Jonk]

Just for a laugh, here's my test from SGLXI.

2 things to note - the dew, and the fact the video is 'bouncing'. I think i know why this is happening.

It was only a left-right movement for the time being (at a roughly 45 degree angle), the 360 degree and up-down brackets haven't been made yet, I'm hoping to have 3 axis of movement in total when complete.

https://www.flickr.com/photos/64466793@N08/26712124280/

 

[cjdawson]

 

1 hour ago, Jonk said:

Just for a laugh, here's my test from SGLXI.

2 things to note - the dew, and the fact the video is 'bouncing'. I think i know why this is happening.

It was only a left-right movement for the time being (at a roughly 45 degree angle), the 360 degree and up-down brackets haven't been made yet, I'm hoping to have 3 axis of movement in total when complete.

https://www.flickr.com/photos/64466793@N08/26712124280/

 

What lens were you using?    I'm looking for a much better lens for wide-field that the kit lens that I've been using.

[Jonk]

I have a Canon 6D and samyang 24mm 1.4 and 14mm 2.8 lenses.  That one was the 24mm. There are many articles online about it, lonelyspeck.com is a good resource.

[cjdawson]

The 24mm is exactly what I'm looking for.  Now I see why you have so many more stars in your view that I managed.  f/1.4 vs f/4.5  Think I can see myself saving pennies for a new piece of high quality glass.

[Jonk]

Also the 6D has made a huge difference. I was able to get the camera and lens on ebay for less than £1k. Pricey I know, but worth it!

[John78]

When you get your bramping working using the LDR be sure to remember the last x shots exposure length and apply a running average of the light level readings over the last few shots - it'll stop the exposure flicker and save you a massive post processing activity.

Here's an un-intersting sunset from a while ago, its just a raw stack of the frames straight off the camera but its smooth....

 

Good work so far though

 

 

[cjdawson]

My plan for the implementation of the ramping is that the light level will be measured and placed into a pigeon hole. Once that happens, the light level will be boxed, so if it varies, it will have to change enough to be into the next adjacent box before the light level changes.  The "box" however will not have the same value needed to switch between the longer and shorter exposure.  The idea is that once the light level switches from one box to another, then there will need to be large change for the box to switch again.  It won't be possible to be "on the border" of two boxes, as once you have crossed from a shorter to a longer exposure, things will have to brighten significantly to get back to that shorter exposure.   This staggering should be enough to stop the flickering that you describe

[Jonk]

If you're going to use an automatic lens, you will get flicker due to the aperture blades not being equal for every shot. What you can do there is use an aperture ring (not ideal) or better still, a manual lens, where the aperture doesn't change.

If you can't get a manual lens, you can always smooth out the flickering by using frame by frame evaluation - a program called LRtimelapse can do this, or you can do it manually.

[cjdawson]

My plan is to run either my 30D or 70D in bulb mode.  As everything will be fully manual, apeture blades shouldn't be moving between shots.  On version 1, I'll be changing the exposure time only.  On a later version, I might be able to get the PI to connect to the camera via USB so that I can also control the ISO.   In all cases, the shot will be controlled by the PI rather than the camera.