I'm looking for an accurate motion control in order to get clean background plates for the rig clean up.
I can't go for an expensive crane so I'm looking around the option available on dragonframe website.
I tried the Noxon but I had some issues. No "0" point and hard to make it right with a firmware pb. Specifically the focus module. And after two test it broke down...
Then last week I went for an Edelkrone system (latest Slider Module with head plus and focus module) but their is a framing shift when we come back to the initial keyframe and the focus never comes back exactly to the exact value (for example -66 aimed > it comes back at -65,857 for example) which enable a proper use of Dragone frame.
Do you know by experience which system can be the more accurate ? eMotimo ? Ditogear ? or else...
Just so you know, I'm using a APS-C Fuji XT3 with a mechanical lens : 7artisans 35mm 0.95 (Rubber Focus ring on the lens = Tilta Focus ring)
Many thanks for your advises
I have been building my own moco rig, the third I will have completed to date. This one has many 3D printed parts and I have recently been testing it. So far it has been quite capable of the sort of accuracy needed for stop motion. The tests I have been carrying out are: several identical passes, then overlaying them in Resolve. And also deleting a couple of frames every 5 or 10 to make the rig go back and forth. I did find a problem with my rotation axis, which I was able to sort out (the joy of 3D printing replacement parts!).
As you say, DF does not necessarily get the exact numbers when it zeroes back. But the difference is, I suspect, less than a pixel, and likely less than any errors introduced by backlash etc. I think I would advise you to go by visual checks - so if there is a noticeable shift on subsequent passes, try to work out which axis is playing up, and where the backlash originates from. You may be able to correct for this. For example, are you using backlash rollback, which ensures that when an axis returns to zero it goes a little way beyond and therefore always approaches a move from the same direction, effectively taking out the backlash before starting?
I would be surprised if the commercially available equipment is inaccurate when set up properly and with the necessary corrections as above. After all, that is what they are charging a lot of money for!
Tks a lot for your expertise Simon !
I'm impressed by your knowledge about this.
I need to learn about the backlash rollback. This was mentioned also by Brian from eMotimo.
Obviously their is a visual shift in the frame so tomorrow I'll run some tests to see which axis is less accurate.
The backlash rollback you talk about can be set in Dragonframe setup ?
Tks a lot again and good luck for the finalization of your rig
The backlash rollback is to be found in the Axis Set-up window, near the bottom. It is also mentioned in one of the DF tutorials. I have not been using it, but I believe it only needs to be very small, just enough so the axis is already rolling in the right direction. This prevents any sudden jump at the beginning of a shot, where the motor moves but all that happens is slack is taken up. When the motor finally gets the axis to move, it is already making larger moves each frame, and the axis is several frames further along from the start, hence the sudden jump.
Make sure to check everything in the system is tight. This is very obvious but still easy to overlook. Pulleys must not have any rotation on the shaft, belts must be tight and gears well meshed. And everything rigid. The bearings of the dolly should be firm enough so there is no play in any direction. Any small error in setting up will likely result in visual shifts, so it is good to observe the unit as it is doing a move test to see if anything is moving not as it should.
I have found that the high precision geared motors from OMC are really the only ones suitable for this work. The 10:1 has about half the backlash of the 20:1, which is already quite small. So I am using the 10:1 type for rotation and pan axes. Something else to consider is that micro-stepping decreases torque, by a lot. So I am using 1/8th or 1/16th where possible rather than 1/32. However, if you are using a commercial unit, all this should have been considered by the manufacturer.
For your tests, do a move using just one axis at a time, to see if you can identify the bad one. It is most likely the slider axis, but you need to be sure. Also, keep the max jog speed not too fast, as the inertia in the dolly may cause it to move too far if it goes quickly.
As for the focus control, I wouldn't worry about it not resetting to the exact number place, as a tiny error in the depth of field is unlikely to be noticeable. My focus motor is an ungeared NEMA 17 with a lot of step-down in the gearing, so it works very slowly indeed.
Thanks a lot for your time Simon !
I run a 10 frame sequence (without backlash rollback) this morning with and without an object to get an empty background plate. I also put physical marks on the gear (Slide/Pan/Tilt) and it looks like it keeps the same position on both start and end keyframe.
But I had a small shift that I could diminish by reframing the empty plate in opacity on the main action (frame by frame because the shift is not the same on all the frames)
I'll check again the all stability and add a backlash rollback. The Backlash rollback need to be set at 4 frames for example ?
Tks a lot again
I think the rollback is in mm rather than frames, but yes a couple of frames or a mm would be about right.
It is more concerning that you say the shift is not the same on all the frames. This suggests to me that something is not quite tight enough in the mechanical set-up. But then you say the start and end points are precise on the gear. The camera is firmly attached? Is the focus adjustment on the lens reliable? I say this because when focus is adjusted on some lenses, the glass moves a bit inside the barrel and can sometimes make a difference, i.e. not return to the exact same spot.
It's now working much better after mydiscussioin with Edelkrone. They adjusted the firmware on the stop motion module which resolve the accuracy of the step.
For the movement after running a test of a full stop motion movement (24 frame) it's almost ok
I have a very slight pixel offset on few random frames but nothing that I can't adjust in photoshop
Thks for your help
Hello. I'm new here but have been getting back into stop motion after a thirty year diversion into the live action world as a camera operator. For the past year I've been experimenting with motion control, and after a lot of research I bought the eMotimo SR4 and slider. I've been astonished at how accurate something so relatively inexpensive can be. I use it with Dragonframe and have managed to achieve pixel perfect repeatability on many occasions.
There are tricks to it, however. The backlash compensation in Dragonframe has not worked for me. It's possible I simply don't understand it well enough. However, I find that if I go back (whether to correct an animation mistake mid-shot or to do another pass) all I have to do is go to a keyframe several frames before where I want to pick things up. Then, when I move the camera from there to the keyframe where the shot starts, everything is moving in the right direction and the backlash has been taken out of the system. Does that make sense?
Here's a link to a very short test of the system. What you'll see is two separate passes superimposed over each other, each set to 50% opacity. It's very solid.
All the problems I've had with getting perfectly repeatable results have come from either me bumping something, or from the setup changing between passes. Something as simple as the light that was bouncing off the support rig in one pass, not being there in the clean pass, have made rig removal more challenging than I expected. But I'm learning.
Anyway, that's my long-winded rambling endorsement for the eMotimo system. Hope some of this was helpful.