SO the first test of 2 axis plus focus via emotimo and dragonframe is complete. Not perfect, but not half bad either. Check the video to see the results, followed by a BTS.
SO the first test of 2 axis plus focus via emotimo and dragonframe is complete. Not perfect, but not half bad either. Check the video to see the results, followed by a BTS.
Ok so its been forever since I put a video BTS up. This is a bit long, but if you are into this stuff you will probably find it interesting. It covers the start of a shoot, plus 8 days into it. It highlights the Dragonframe software and shows some pretty killer functionality. Of course you cant just buy dragonframe and get shit to work, you have to build some stuff. I spend a little time going over the Driver/Power supply box and what its used for, as well as some time on the Pan Tilt from Servo City.
Hope you enjoy
Its been a busy but awesome weekend. 🙂
First off, I am really excited about this thing. A while back Kyle had printed up some lens collars for me. The idea was to mount the lens to the 15mm system instead of the camera body.
Focusing is a bitch to do. One of the main reasons is there is always SOME degree of slop or play in a lens mount. when tugging the focus ring back and forth the lens will shift ever so slightly but it is MORE than enough to show up on a timelapse. I have several macro lenses that I timelapse with and the focus ring is very stuff. So to combat the lens shift the lens itself is mounted to the 15mm system, and the camera attaches to the back. That worked pretty well, but still was not perfect.
When playing around with the 6D, eMotimo, lens apparatus, dragonframe, I had set it up as such. In order to lock in a very tight fit i used some copper wire and a eye-screw and wired it down so the lens apparatus had a very tight lock. But seriously, it looks stupid. But it worked like a champ for my focusing testing.
I ended up sending Kyle a sketch to see if he could modify the design and reprint. And Kyle that glorious bastard did it in Safety-Orange, which I think looks absolutely badass.
He had actually added the 15mm support on the wrong side, but after installation it will work out fine. The Lens apparatus has an amazing level of contact with the lens now. There is no perceivable give between the gear and the lens ring.
Using a fairly short 15mm rod. Everything is a nice compact form.
And to top it off, Kyle put a window in the lens ring so you can see the focusing scale.
Another great thing about this is is that since all the hardware is on the lens itself, I don’t have to deal with the added height of 15mm rod system under the camera. I really would not want to deal with that extra height on this rig. This really made my day!
Beware this is getting geeky.
Managed to put the power box together. First off lets explain what this is and why I need it.
For this setup to work you have to power the motors, and you have to be able to chop that current into various patterns to get the desired response from the motors. If you just apply power they motors will get damaged. These are not like DC motors that just spin when you apply power, rather you instruct them as to what you want its position to be. For this to happen you have to use whats called a “driver”, which is basically hardware designed to manipulate stepper motors.
So thats what this box will be used for. It comprises of a pair of 36 volt 16 amp power supplies, which power the 10 Stepper Drivers. Each driver can control a single motor.
So the Laptop is used to build the movements out in some pretty awesome stop-motion software from Dragonframe. . That plugs via USB into the Dragonframe DMC-16. That piece of hardware is what is responsible for lighting control and motion. On the motion side it can send signals to up to 16 stepper drivers, and one motor per driver. It uses RJ45 (ethernet) cables to connect to this new power device to send the step(move)/direction signals to the Drivers.
Laptop –> DMC-16 –> Driver –>Motor.
It went surprisingly well. I had picked up a server chassis to use to house the power supplies and drivers. It was a bit bigger than I expected but it has great ventilation and a solid build for about 80 bucks.
I had picked up some wall plays for RJ45 connectors and they would fit fine right behind the slots, a few cuts later and a little file work…
Next I needed to be able to mount the jacks for the motors. Using some 2 inch aluminum stock Measured it up, marked it out, and had to drive to Ace Hardware because I did not have a 5/8th bit. Damn thing cost me 20 bucks. That was Trip 4 I think.
PRO TIP. Put painters tape on the aluminum and make your markings on the tape to keep the plate looking nice.
The plate measured, cut and drilled. After I got all the plugs attached in I used this fantastic glue called GOOP. I seriously love this glue, it sticks to damn near anything and is like a super sticky rubber cement. It never hardens brittle and has some crazy grip. I used it on both RJ45 panels and the aluminum panel.
Next I went in and drilled and tapped the sub-bottom after marking various spots so I could mount the stepper drivers and the power supplies with screws. That bottom panel comes off making it easy to modify it as needed
Once everything was installed and bolted in it was time to start dealing with the wiring. That was a real pain in the ass. This shitty 4 wire telephone cable I got from Ace Hardware it did not want to strip and the copper was pretty brittle. It was a real bitch to deal with. I probably spend close to 4 hours just wiring.
The voltage regulator is basically just used as a super simple fan speed control. It just takes a 36v feed off one of the power supplies and reduces it to 7 volts to keep the fans running.
You cant really see it in the pic, but each of the drivers is numbered, and I split them across two power supplies. Even numbers on one, Odd on the other. If needed down the road the thing is rigged up for 12 drivers. I am only using 10, but if i need to expand later I can virtually just pop them in.
The back end looking pretty good to me. Not huge on the plastic ethernet panels, but they will certainly work and are very secure.
Finally done wiring the damn thing and it was time to turn it on. I checked the voltage and everything looked in spec, so I wired up the first row of drivers, turned it on, and nothing exploded. I powered up the other side and got to testing.
During testing I discovered I could drive a motor off all off them but 3, 5, and 7. Checked the wiring and that shitty house telephone cable had a couple leads that broke. I had to re-terminate them but everything checked out after that. A few min later and all 10 are working as expected
Now that everything is tested, I closed it back up and started experimenting with the stepper to see what sorts of speeds i could reach. Everything seems to be working well!
Here it is in all its glory, closed up and ready to make some awesome shit happen. I really could not be happier with how well this went together.
Next its time to address the systems frame, figure out to mount the PTR, how to get the z axis going.
One thing at a time.
First things first.
I am now on Twitter. Lets see how this pans out. Go follow me.
Yes, there is already a Biolapse on twitter. Turns out there is a musician that goes by that nomenclature. So I had to go with Biolapse Films.
I have a new gmail account for Biolapse as well. BiolapseFilms@gmail.com
I want to distance Biolapse from The Chronos Project. One is a non functioning hardware company, but Biolapse is all about filming plants and is very much alive and active.
Now, on to OTTO.
These last few weeks have been a bit frustrating. The prior week I had been on-call with my day job, and it is very invasive and consumes a lot of my time. This past week I have been hiking in the morning, working a later shift, and had not had much time to work on the project. But there is some progress since the last update.
So for those who are wondering WTF I am talking about, lets take a quick discussion into stepper motors.
Stepper motors are friggin awesome. With a DC motor, you apply current and it starts to spin. Stepper motors do not work the same way. Rather than just spin with current, they take a single “Step”
Without getting in the nitty gritty, most step motors have 1.8 degree steps. What this means, if you want the motor to make one full rotation, you send it 200 pulses and the shaft will rotate 360 degrees. Send it 100, it will only move 180 degrees. If you want it to spin, just keep sending the pulses and it will keep spinning. If you want the motor to move to an exact point, then send it the number of pulses to get to that point. (this is a very gross simplification of a pretty complex system)
So how does this all play out?
The Dragonframe software on the laptop controls the scene. You use that to position the camera and work out the camera movements. It will then send these movements to the DMC-16 Dragonframe controller. This controller sends the step pulses to up to 16 motor drivers. The drivers take those pulses, and will power the coils within the motors to get them to behave as expected.
With this, you can get a very predicable outcome, and get repeatable moves. You can program the movements, and replay the routine over and over, and the motion will always be identical (as long as you take care in building the rig, don’t have anything binding up, etc).
So the next step is to get these all wired. I have a pair of 36v 16amp power supplies that I will use to power the drivers. I picked up an empty server chassis to house these in. I am hoping to get this wired up monday.
The Pan Tilt system is in!
I manged to get the Pan/Tilt/Roll hardware from ServoCity. It is the PT-2100 w/ roll attachment. It really is pretty badass. It can handle up to 40lb loads, far more than I will do, but I wanted to get something solid, and solid it is. The bearings are fantastic, the motion is smooth. The only issue is it might be a little bit big for what I need, which means I may have to slim it down a tiny bit. I am not worried about that though. Ill address that later if it becomes and actual problem.
The unit came with no electronics, no motors, which saved me 350. I was just going to replace the motors and had no use for the controller, so that worked out nice.
The 50:1 geared steppers arrived, of course, I cant play with any of this shit until I get the drivers and power source wired up monday.
I have some design considerations I am still mulling through. The big question is the Z axis. I am not quite sure how that will all go together just yet.
This is the biggest cause of concern for me. The design has already taken a couple directions. But I am fairly sure I am on the right track now. I will be building this out of 8020 aluminum extrusion.
The overall frame design should be good. I will be using some fully supported bearings with 1500mm ball screws. So this is basically going to be a giant CNC machine. The X axis is figured out. I just need to make a few final measurements. The Y axis, i have to chose between two layouts (not going to bore you with the details) and then the Z axis which raises the camera up and down… well… I’m sort of stumped. I will figure it out though.
I am Hoping to get all the frame materials ordered in the next few days. I want to order now, but any mistakes could prove to be costly and waste money. So better to check, measure, design, double-check, double-measure, confirm design. THEN order exactly the parts I need, and only the parts I need, all pre-cut so all I have to do is assemble.
Now, time to get to building.
The new equipment has started to arrive 🙂
And it has become clear that this will be a lot of work on my end, as there are no good off the shelf systems capable of what I am doing. Dammit. 😉
Canon 6D arrived. Honestly I’m not too excited about it. I have used so many cameras at this point, that a pretty generic DSLR has a hard time exciting me. The build quality is solid, no articulating screen which is incredibly handy for this type of work. The image quality will no doubt be fantastic as all Canons have wonderful IQ. However, it just feels and looks boring. Luckily I got an open box special from Adorama. I almost shit myself when it arrived and I opened the box and it was nearly empty! Turns out it was part of a bundled package with a f/4 kit lens and the camera body was sold as a refurb. Snapped a shot and used a program to determine the shutter count. 1. Nice! Brand new never used 6D for 1200!
Latop arrived. I picked up an ASUS Aspire laptop which came surprisingly devoid of bloatware. I think it had solitaire on it, but that is about it. As expected with an inexpensive laptop the track pad sucks, the screen is not phenomenal, but it does have a decent i5 processor, 8 gig ram, and 256 SSD with the ability to upgrade in a second hard drive and extra ram with very little effort. Im not going to bother with the ram, but I did just order a 500gb HDD for $35 bucks. With the OS and dragonframe it only has about 200gb left over. That just seems a little bit cramped to me. I am fairly happy with the laptop, it will make an excellent brain for the Otto.
Dragonframe DMC-16 arrive and I am super excited about that. It allows 200khz pulsing, which is 10 times faster than the arduino based DF setup will allow. It is a solid built piece of hardware. The only complaint is the USB connection on the front. I imagine this will always be tied to the computer, I wish that was in the back but no big deal. It allows for 16 channels of motion control, but I’m starting out with just 10 for now. I dont know if I will need the extra 6 or not. But i like the fact I can scale things up a bit more if needed. Right now I am aiming for X,Y,Z,P,T,R,F which leaves 3 extra steppers I can control.
I spent some time playing around with Dragonframe on the new laptop. I am still waiting on parts before I can finish power supply and driver setup(more on that later). So for now the best I can do is play with DF and the eMotimo which integrated up pretty nicely. What i had not realized, is that Dragonframe can interface up to quite a few systems at once! Not only can it connect to the DMC-16, it can connect to the eMotimo at the same time. In fact, it can connect to quite a few devices simultaneously.
This means not only will i have the ability to control 7 axis on Otto, but also 3 spare motors for turntables and such. And on top of that, I can wire the emotimo’s aux port onto a chronos rail and be able to control all 3 axis there as well. Not shabby at all! an unexpected suprise!
Dragonframe also gives some pretty sweet capability for lighting control. I suspect that the BCM (my Biolapse Control Module) will still control the day/night sequences for the grow lights, and possibly the LED fill lights, but the DMX controlled lights will all be via Dragonframe. Not sure what DMX stands for, in fact, until a month ago I just thought he was a rapper. I picked up 4 cheap DMX LED lights and I can assign each of them a set of channels to control the color and brightness. I can setup the routine in dragonframe to adjust color and brightness as the scene progresses. So not only motion control, but full lighting control with multiple lights. I am super excited to play around with that! 🙂 I do have concerns about their performance on very dim levels, but I will address that another time.
So I spent some time testing out Dragonframe today since I am stuck at the house. Honestly I have to say I really enjoy it. There is a bit of a learning curve, but in the end it is a CNC controller with some awesome lighting and camera controlling capabilities.
I rigged it up with the eMotimo, and managed to tie in a Lens apparatus in the eMotimo’s AUX port. It really works like a champ! I am fairly impressed. There was a bit of fussing on the Lens apparatus to get it on there nice and secure. One of the problems is I am using a Nikon lens on a Canon body. That can create a bit of play on the lens in the mounts and adapter. The 15mm rigs work pretty good for mating the lens apparatus to the camera, but I want to eliminate any play in the lens that may be caused by the back and forth motion of the lens apparatus. A while back Kyle printed out a few lens collars for me so I can mount the 15mm rod system directly to the lens, and not under the camera body. This does a pretty good job of getting rid of any “lens tug” that one may encounter.
So the lens collar kyle printed was great, but the entire purpose was to get rid of any lens play. However, the lens collar itself has a point where it flexes. So to get rid of that flex I use used some wire to tie everything together. Works, but looks sort of stupid. I am going to see if he can send me the design for the collars and if he will print a new one out if i make some modifications. But this works in a pinch.
Side view. You can see the Grey band around the lens, that is the 3d printed lens collar, which has a 1/4-20 thread beneath it. I have a low profile 15mm rod system attached to it, so the camera is mounted to the eMotimo, the lens system is completely on the lens. Wight is not bad either. its just plastic and aluminum.
Once that was setup I put out a couple targets to test focus with. Dragon frames. A Star wars metal droid sculpture, a lego guy, and a metal venus fly trap.
The keyframing system is fairly straight forward. A Keyframe is basically a target. In this case I setup about half a dozen keyframes. To accomplish this you basically use the arrows next to PAN TILT and FOCUS to adjust the placement and angle of the camera and adjust the focus where you want it. You don’t need to put keyframes on all 3 axis every time or else the end result is the camera moving, stopping, moving stopping. its best to use only use a keyframe on an axis that is stopping or shifting direction.
I started by moving the camera to the droids head, then down just left of the lego dude, then on the lego dude, next on the base of the metal plant, then on the plants mouth, then the tounge, shifting the focus each time.
Dragonframe makes focus checking SUPER easy with the focus check function. You can do a “focus check” which zooms in on a small point and you can make very fine tuned adjustments. Seen below I’m zoomed in on the face of the robot.
To get a better idea, the face of the robot above is a zoomed in view of the image seen below. That sort of functionality makes focus a breeze!
The robot is the starting point. You can see the waving lines below, each one represents a different axis. The little dots are the points on the timeline where the keyframe is placed.
Once I was happy with how things looked I ran the routine. Right off the bat I ran into some issues. You can see on this crappy footage below
This is NOT video, this is a collection of still images. The system moves, takes a picture, moves, takes a picture. SO on so forth. I collect these into Adobe Premiere Pro and render them into a video sequence.
What IS good about it, is it NAILED the focus targets!!! The 15mm rig mounted direct on the lens with the wire supporting it worked very well. I don’t want to get into the mechanics of backlash (which Dragonframe does have the ability to compensate for, but your results may vary depending on the system. Focus is hard to compensate for) but I am EXTREMELY happy with how well this worked all things considering this was my very first attempt. THis actually put the first 360 clicks on the new Canon 6D.
What is NOT good, is it is jumpy as hell. I double checked the number of frames to make sure none of them were dropped. I was pretty perplexed, and the best I could determine is there was too much movement and not enough frames. This could be done easily with a video camera because it shoots while moving and you get a bit of motion blur that bridges the frames giving it a smooth output.
I upped the number of frames to 700, and dragged the keyframes to drag things out a bit thinking this might smooth things out a bit.
It certainly slowed it down but I still see the motion jumping around. I rendered it with about a dozen different formats trying to see what the problem was.
I even tried applying a subspace warp (good god I love the name of that) which made a drastic improvement, but I dont like Bandaids i want an actual fix.
Subspace warp can be a life saver, but the problem is it shifts the frame around a bit so you need to have some croppable areas or you get black artifacts where it moves the edge of the frame in view. You can see this on the bottom of the frame just as it moves down from the metal droid.
Finally I managed to pull my head out of my ass and saw the following in premiere pro.
I literally spend like 2 hours trying to figure out why it was so jumpy.
So in this case, the sequence was built at 25 frames per second, but I am trying to turn it into a 29 frame per sequence output. This means it has to figure out where to get the 4 extra frames, and where to fit them in. This causes that jumping phenomenon.
I adjusted the sequence to 29fps, deleted the frames and re-loaded them back in, then rendered in 29fps again.
FAR BETTER!!!! Some of the motion is still too fast for 29fps. I am wondering if a program like twixter may help? Or just be more aware of how I set things up to prevent those really fast moves.
Most of the shit I shoot is pretty small. Which means there is a lot of macro photography going on here. When you do that sort of work. the DOF (Depth of Field) gets very slim. For those non-photographers, the DOF is the area that is in focus. When working with small objects and focusing very close to the lens, the Depth of field can get very tiny, there may only be about a centimeter to work with, everything in front and behind that sweet spot is out of focus. Some of the work i have done in the past the area in focus is no more than a millimeter deep. So the videos you have seen where the focus is nailed may not seem a big deal. But below is a sample where i shot it with NO focus adjustment. And you will see why focus control is so critical.
It looks like shit doesn’t it?
Focus control can be an extremely fickle mistress. Most timelapse photographers find it too difficult to deal with and don’t bother. When shooting tortured trees in front of star fields, or breathtaking vistas of a mountain or something you don’t need to worry about it. For my line of work, focus is critical to learn how to work with.
There are some methods you can use depending on the camera where focus ramping may be plausible within the camera. But your results may very and sometimes it can be hard to predict the outcome. A couple years ago Kyle and I funded a kickstarter for the Lens Apparatus. Aside from one produced by a friend James at StepoutMoco all the other options are stupid expensive. $1000+ just for a damn motor.
Ours works pretty darn well but we are no longer making them. The demand just dropped as nobody else does this sort of work. So if you want one, get with James. I wish he had a better online presence. I had to dig for about 10 minutes to find his product. His newest version looks fantastic he is really getting awesome at his designs. I honestly think it looks better than our Lens Apparatus did, and it uses a stepper motor that makes it perfect for Dragonframe.
It was funny. Last night I was wanting to integrate focus with this rig and couldn’t find a Lens apparatus. I don’t even have enough parts to build one. Finally found one today when i was cleaning out an old drawer. Its an older version, but it works. 🙂
A quick shout-out to the guys at eMotimo! This device never stops impressing me. Kyle and I tried to build a good Pan/Tilt system, and it is NOT easy. Brian Burling knocked it out of the park with his system.
It works very well with dragonframe. The only reason I do not plan to integrate it with Otto is it is important to me to have a camera moco rig where the camera sensor is dead center of the pan and tilt axis, and I really want a “roll” axis as well.
I would be interested in their new Spectrum, but I am sinking a lot of money into some high quality stepper drivers and want to control everything through the DMC-16. This means i would be spending a couple grand basically to gut all the electronics out and hack it into my rig. That seems silly to me. I dont know how i could integrate a Roll axis either, so I plan to stick with a PT-2100 from Servocity if they ever get them back in stock.
All said though, the TB3 is just a bad-ass piece of gear.