Fabricated Stop Motion Rig

I decided to take the very brave step of documenting my project to build a cheap stop motion rig as I am building it rather than waiting to the end when I know whether it was a success or not - nothing better than sharing your best disasters with everyone

Yesterday was spent welding up a bracket to firmly hold my existing tripod - car exhaust brackets grip the tripod stem. Last night I wondered if I should have just gone and bought a pan-tilt head (on its own) rather than have the rig carry the whole tripod but then I remembered I did want height adjust as well so perhaps I can convince myself I took the right path, it will probably look a tad ungainly but I am not sure there will be a practical downside.

Fabricated tripod bracket

Bracket On Tripod

The bracket can be used 'upside down' I deliberately designed it assymetric this way so that by simply reversing the bracket the camera can rise or fall by 15 cm - this just gives me a little more adjustment over and beyond the rise in the tripod itself.

I have two ready beds to hand, one is a folding aluminium rail support table for a chop saw, the other is a light lathe bed ( one person can pick it up ) I favour the lathe for a swing arm as the length of the boom arm will demand a steady base, the rail support for the saw will be fine for linear motion.

If it works for video footage then all well and good but I am designing primarily with stills - ie stop motion in mind.

The driver for all of this is lack of a high end video camera but a ready high end SLR, need to compose various tracking shots of high quality and pin sharp focus of hand made products.

"Keep it simple stupid thoughts"

I am planning to just use a steel rule or even the giveaway Ikea paper rulers for the linear bed.

For indexing on the radial arm I have experimented with simply standing steady right up to the camera end of the arm so that the end of the arm is nearly in my chest, I think I can reliably push it say 2 centimetres at a time, I dont want to over engineer some system of angle measurement only to find it does not make much difference.

I am not worrying  too much about silky smooth qualities, this is primarily for stop motion rather than video - accurate positioning yes but quality of movement is not a priority at the moment

Comments

  • I should add that I placed the post in 'users Gallery' because I hope to add footage taken using the rig at a later time.

  • edited February 9

     Today's progress.

    Mounted on saw bench

    The slider not yet painted, two screws to lock down

    Slider - two screws to tighten

    At the moment the camera and tripod would fall off if the two screws were loosened completely so I think a safety is in order.

    I was lucky enough to have a piece of thick sheet steel with an accurate 90 degree bend along one of the long edges for the slider, this acts as a guide to keep things parallel.

    A little 'technique' is required in use - its not silky smooth so a few improvements might be in order but first of all I want to take it for a test run and see how it performs in real use. I had  to polish up the steel surfaces that run on the saw bench - otherwise the steel will gall the aluminium bed of the bench.

     

     

  • I get an error when I try to load the images you linked

  •  @JMcAllister

    Sorry about that - they are hosted on Google photos - I wanted to upload the photos directly to this forum but it seems to accept only URLs rather than allow upload of the photo itself from local disk which would have been my first preference ( avoid broken links in years to come etc ).

    I have checked that I can see the photos from the thread posting on other devices not connected with my Google account - no obvious issues. I will look into it further.

    If anyone else is interested in this and cannot see the photos please shout out,

     

  • I'm interested and cannot see the photos either. I host mine on www.photobucket.com. Must be loads of other sites that won't go bust on you years from now. 

  • edited February 10

    Interesting!

    First 'rough' attempt taking actual footage.

    30 frames - little over a second.

    I will not be posting this very first attempt yet, it was done in a rush just as natural light was disapearing.

    Observations:

    1. Positioning the slider is a little awkward, friction means its difficult to get millimeter perfect position, I may add thin PTFE or teflon tracks on the underside of the slider.  Good technique could be developed - its just not a great 'feel'.  A  fixed indexing or measurement system is a must to ensure even steps.

    2. I suspect only the best precision mechanical equipment will get close to 'pixel perfect tracking' - precision comes at a price.

    3. I have used the HitFilm tracking and stablization to sort this out and it works very well.

    3. My Nikon D300s has a deficiency. With camera tethered to my laptop  I can focus, frame and deal with camera settings there. On the laptop screen I see the viewfinder image - all good.

    However...the essential grid option disapears as soon as I zoom the viewfinder image in the attempt to really nail my choosen reference point. I have to manually track guided by default viewfinder zoom.

    Intermediate Conclusions

    A sturdy well constructed but non-precision rig will not offer close to pixel perfect tracking, software will have to play a major role.

    Cameras which scale the viewfinder grid with a zoomed viewfinder image are preferable.

     The consequence of using software stabilization is the necessity to 'crop' - if stabilization moves one frame down in the Y direction then the top Y boundary will need cropping for the entire sequence.

    As an alternative I do have some CNC rails and buggies lying around from a shelved project. I could easily use these tough and very high precision items to achieve much better mechanical tracking with ball screw advance for easy slider movement and indexing. However these are expensive and I am not sure I want to throw them at this project and...

    HitFilm tracking and stablization is doing a really great job* I suspect that even on the best mechanical rig it would still be required.

    So what is the advantage in a high precision system, what benefits would I get if I did use the CNC bits and pieces?

    At this stage I think that mechanical precision will not out perform the software, a high precision system will be easier and nicer to use but may not outperform a rig like mine in the hands of a patient operator who has developed a 'feel' and does not mind working with something a little clunky.

    Its a case of weakest link here - the camera may be the weak because of the way that the grid is implemented in the viewfinder image if you cannot 'viewfinder track' accurately then whats the point of low error mechanics?

    As long as the software stabilization outperforms the mechanical then the pro's and cons are time, effort and ease of use and how much area needs to be cropped around the borders of the sequence frame .

    An infrequent user is probably fine with a well built but non-precision rig as long as they understand the extra work involved.

    *I can see that some enhancements to the Hitfilm tracking and stablisation might be very useful to anyone doing stop motion, I will put them in another post with screen shots and the possibly on to the wish list.

    ==========================================

     Non-essential fine detail - It is a good idea to harmonise the point you decide to do mechanical tracking in the viewfinder with the eventual point you will do software tracking with. I picked different points at different heights along the same vertical line. Each was selected for reasons of ease but if the camera moves closer to the object or subject then disparties arise.

    Think of a person growing taller but their head staying static in space ( so their feet move down  ) vs a person growing taller but with their feet remaining fixed so that their head goes up.

    You might expect that if the camera is moving in the horizontal plane then all points along the same vertical line are equivalent for tracking and this is broadly true as long as the apparent height of the object does not change, a camera moving radially around an object will not have this problem but a linear track means that the camera distance will not be constant, the object will grow and diminish in the image.

     

  • edited February 10

     Another attempt to get the photos visible.

    The Bracket

    bracket

    The Bracket On The Tripod

    Fixed To Saw Bench

    The Slider - two knobs to tighten. The scrap sheet I had lying around for the slider base fortuitously had an accurately bent 90 degree turnover (not visible in the shot below but visible in the one above) which registers against the edge of the aluminium saw track nearest the tripod. The person adjusting has to ensure that this is snug against the  saw bed for accurate positioning.

     

  • Hmm there seems to be a bug on this forum.

    It seems that if you re-edit a previous posting you can lose the posting you most recently added. I see a message along the lines that my posting is awaiting approval after multiple edits with the spurious loss of other posts. The post was regarding my first attempt at real footage - no pics. If anyone has a copy on their screen please keep a copy as it took ages to write.

  • edited February 10

    My first trial on Youtube.

    https://youtu.be/LyRXDjnonAw

    24 fps - taken at 4288 x 2848 - video exported at 1080p HD

    I set my Nikon D300s to export both jpg and RAW (nef) but for a quick first result I just took the jpg's direct from the camera with no processing.

    Please remember to set HD option for viewing in Youtube, best viewed frame by frame initially as its very quick.

    Issues

    • Variable lighting conditions, I really should have approximately corrected these by eye in camera, need to look at what methods are best for fine tuning brightness in a sequence of RAW images for the real thing.
    • 1 centimetre movement per frame, I need to try 0.5mm so I am definitely heading in the direction of buying a long thread and some more mods and winding the movement - too fiddly to do by hand.
    • Refocus on manual tracking area every frame.
    • Image stablization in HitFilm Express - the cross on the Mannequin was my mechanical tracking mark during the shoot.
    • Comparison of the footage on my laptop and the uploaded result in Youtube suggests compression - it is sharper and clearer on the local file.

    Overall I am very pleased, the issues and imperfections are down to me not the equipment.

    Immediate improvements that are required.

    • Slider needs to be more 'slidey' thinking PTFE strips underside.
    • Translation needs to be by winding a long thread.
    • Tripod adjustment should be done first - make sure tensions on adjustment just right before starting, I did suffer a little roll due to slack settings.

    Hindsight Observations

    Spot check points on proposed travel before starting, I wanted to go further but stopped due to disapearance of my physical tracking point on the mannequin - his hand obscured it. I was also using this point for the software stabilization tracking so no point continuing the shoot once I got to that point as I would not have been able to stablize after that.

    The stabilization in Hitfilm works very well for this example, I selected single point and template match.

     

     

     

  • First result

    https://youtu.be/LyRXDjnonAw

    Shot at 4288 x 2848 Nikon D300s SLR exported at 1028p Youtube setting.

    Camera exported RAW and jpg but for simplicity I just used the jpg with no adjustment out of the camera.

    Remember to view in HD and probably best to go frame by frame the first pass as its very quick.

    Issues

    • Changing lighting conditions, I really should have done an approximate adjustment by eye 'in camera' during the shoot, still would need to postprocess the RAW images for the interval but best to do initial adjust in camera ( exposure time not apperture ).
    • 1cm travel per frame is too much - need 0.5cm I will need to modify rig to use a long screw thread to advance slider.
    • Slider needs to slide more easily - thinking thin PTFE ( low friction plastic ) strips on saw bench.

    I think Youtube has compressed - it looks sharper on my local drive.

    HitFilm Express tracking and stabilization worked very well.

    Hindsight Wisdom

    • Check tripod adjustment tensions before starting make sure camera is very free to move but does not 'droop'.
    • Check full range of travel first - I had to stop earlier than planned as the mannequins hand covered my manual tracking cross (tripod tracking) - it was also the point used for the software tracking and stablization so no point taking more frames after this point.
    • Adjust exposure in camera by eye or judgement to minimise effect of changing light conditions, start off with exposures that are nearly the same before trying to refine in the software especially if planning to do a pre process on RAW images.

    Footnote - forum bug? - Again I lost the first version of this post - multiple edits on an existing published post does seem to result in complete loss of the post on occassion.

     

     

     

     

  •  Another known imperfection which does not concern me at this trial stage.

    The usual domestic distractions meant I took some frames more than once, I forgot where I was in the process.

    There is some stutter due to these duplicates which I should have weeded out but its really only a first trial so I am not bothered by this imperfection.

     

  • edited February 10

    https://www.youtube.com/watch?v=zTt9kQn7gUg

    Same frames but this time processed from the Nikon RAW files (.nef) using viewNX which has very rudimentary adjustment capability - I tried to even out the uneven lighting by eye, I think I can do better with practise but the main thing as already mentioned will be to try and adjust for changing light levels 'in camera'. 

    Shot at 4288 x 2848 Nikon D300s SLR exported at 2160p UHD Youtube setting in Hitfilm Express - select highest resolution available for viewing in Youtube.

    Forgot to remove duplicate frames again stutter remains.

     

  • looks cool, even with the flickery light

  •  Thanks - if I cannot remove all the flicker then I will call it an atmospheric effect.

    Read a few articles online from stop motion animators preferring some artifacts to keep it 'real'.

    I aim to repeat at half the step length for a smoother ride at a time when there are either no clouds or complete cloud cover in the belief that light levels will be more stable - it takes between 30 mins and one hour for 60 frames.

    After publishing the above I experimented with motion blur in Hitfilm, animators suggest it is essential in some cases.

    I felt it did smooth things out and give a greater sense of fluid motion but at the cost of sharpness and clarity which is one of my main drivers.

    My belief is that if I go to half the step length which is a more realitistic 'speed' for my requirement then the need for motion blur will diminish and I can use the minimum required.

     

     

     

     

  • Stabilization Software or Hardware?

    By hardware stabilization I mean a precision built rig perhaps using buggies and rails from the world of CNC* which was a possibiliy floating around in my mind but I believe it is not necessary.

    By software stabilization I mean the tracking and stabilization in Hitfilm Express.

    If you are on a lap/desktop try looping the above Youtube video so it is continuous, go full screen and place your mouse pointer on the cross of the manniquin's abdomen. I was amazed when I did this just how pixel accurate the Hitfilm stablization is in this case.

    I really do not believe that even with the most precise engineering a stop motion rig will rival the software stabilization. It seems to me that a rig should ideally offer precision in the direction of travel - equal steps and ease of use, it should not create so much error that significant border areas are lost due to the cropping required after software stablization but after that its really down to the software. I can see why pro's might want a robotic computer controlled version but for a linear track I am not convinced there would be much of a difference in the final result - the software is the main player here.

    My rig needs sophistication in terms of easy accurate step length and ease of use ( the slider could move easier ) but after attention should move over to things like lighting issues.

    ===============================================

    ( *CNC alternative - couple of buggies, 25mm solid steel rails and a ball screw advance mechanism with fittings - lovely ! Cost around 200 GBP for a 1.5 meter track. CNC would style would be nice - all designed plug and play fit - you  buy the necessary fittings - after that its drilling and screwing Buggies / Carriages   square rail  Ball screw advance . I actually have this stuff lying around after a CNC project that was not completed. Reason for not using it here - it is expensive, I dont want to break the set and I am not sure it would actually improve the final video that much - whatever the CNC delivers can be made up for by learned skill and the software ).

     

     

     

     

  • You definitely want your camera in manual mode to control the  exposure...maybe even get a light-meter to check exposure throughout, but best to control the lighting as well as much as possible, but don't let the camera make automatic adjustments

    Flicker might be coming from aperture - this occurs frequently in time lapse.  The aperture doesn't always stop down exactly the same place - not really noticeable in stills until you put them together into a movie.  Google time-lapse flicker to get some ideas on how to deal with that one and what you're comfortable with (software-pricey...I use LRTimelapse; manual aperture lens-pricey too; I also unscrew the lens a bit while holding down the depth-of-field preview which freezes the aperture in place...Canons can do this, not sure about Nikon - BUT BE CAREFUL - your lens is not locked to the camera, and you WILL forget that little point and drop the lens unless taking precautions like taping it so it won't spin)

    It's looking good - I really like the camera move!

  •  @dplester

    thanks!

    At the moment I am adjusting shutter speed not aperture control in manual mode.

    Whilst I have been working on this weather conditions mean that the light levels are varying - several stops between frames. I am rapidly coming to the conclusion that what I am trying to do is only realistic when light levels are near constant.

    If you have any observations regards the pro's and con's on adjusting Aperture or Exposure time I would be very interested.

  • Today I tried to improve slider so that it moves with a little more ease, and I need a scale to measure distance, looking for really simple solutions so this ended up being a 1mm thick plastic sheet to minimise friction and a tape measure stretched between clamps.

    Fair improvement all around, can now advance accurately at 5mm intervals with much less friction stick.

    Rather than glue the sheet to the slider it is just drilled out - the screws used to hold slider down (tightened each frame) also keep the plastic sheet in place.

    Plastic sheet for reducing slider friction.

    plastic sheet reducing slider friction

    Tape measure stretched between clamps

    Tape measure

    Attempt to get footage thwarted by issues with Nikon software - now resolved.

     

     

     

  • @Jonnie45 I think you're bang-on...For Timelapse, adjusting shutter speed and ISO are the preferred methods to control for changing light conditions to avoid  flicker (the variance between exposures is a much smaller percentage for those than it is for aperture variance, although, if you deliberately change aperture you might be able to get some really cool depth off field effects going on)

  • edited February 13

     Latest attempt.

    https://youtu.be/NnEhYEK-s7g

    Shot at 4288 x 2848 Nikon D300s SLR exported at 1028p Youtube setting.

    Here I worked hard on equalizing out the image brightness on a frame by frame basis working on the Nikon RAW images.

    Nikon software is often flakey reliability and usability - a lot of ups and downs, in the end I downloaded the latest version of Capture-NX-D to replace an older copy of viewNX.

    I was able to work well with Capture-NX-D my work flow was as follows.

    Arrange screen to give largest possible preview of the RAW image, the brightness histogram and the EV adjustment tool which can change the effective exposure in very small steps indeed.

    Find a screen layout which means that previews do not scroll, the next or previous frame will take the same position as the existing one as you travel up and down the imaginary timeline.

    Use arrow keys to change frames.

    Use mousescroll to adjust exposure.

    Its really worth getting the layout and working method just right.

    Thoughts On Quality / Realism

    I am very pleased with resolution and clarity, I could not achieve this level of focus tracking with the Nikon used in video mode.

    Exposure flicker is much better and I could improve further by picking a less variable day ( clouds and no wind ) and also just doing more work in Capture-NX-D, I think several passes of refinement could be used - if a flicker can be seen in the resultant video then make a note of approx frame number - go find and adjust adfinitum.

    The highlight flicker would disapear with even lighting. Two aspects of importance - how diffuse or even the ambient light is and the its intensity. 

    You can adjust for intensity changes but if the light is varying in an uneven way ( sun out = highlights shadows / clouds = even : no global compentation can match sunny frame against cloudy frame ).

    Although this latest version is an improvement I am having doubts as to whether I can push this to where I want it.

    I am sure I can overcome exposure flicker - pick the right day and spend ages working on the RAW, I am very pleased with resolution detail and focus BUT to me it still does not look like a smooth video - the illusion of film just is not really kicking in yet - I still see a series of jerky frames rather than a continuous movie.

    The driver for this is high quality 3 second pans of artisan products, high definition and detail with pin sharp focus througout - I feel I am approaching all these qualities except its not flowing as a movie.

    It might be variation in my step length along the bed.

    It might be lack of motion blur.

    Not sure at this stage.

  • A note on histogram.

    When I first started to even out exposure I was looking primarily at the frame itself, flipping frames back and forth looking for flicker and adjusting until I felt I had reduced fliker significantly.

    Later I paid more attention to the brightness histogram ( combined RGB not separate ).  I noticed that well matched frames gave very similar looking hills and valleys at the same position along the axis, the variation between frames was like watching an irregular line of ants marching along the silhouette of a hill - in otherwords fine variation in shape yet the eye is very good at spotting that when the hill is moving and when its just the ants moving, the latter being the optimised match - if you flick between histograms there is a sweet spot when you can just tell there is no large scale movement only ridgeline fine detail movement - the ants are marching but the hill is not.

     

  • edited February 13

    I tried looking at the screen (previous posted video) holding both hands to my face to view only a small area at a time.

    The jerkiness seems to be much more associated with the background.

    When I cup my hands with a small gap so as to visually isolate the manniquin I see a much smoother movement.

    Comparison between frames shows the background is moving in larger steps which is not surprising since the camera is moving in a line but staying centred on the manniquin abdomen.

    I don't think its as simple as just using motion blur to smooth out a faster moving background.

    I think errors in camera position and orientation, tiny imperfections in apparatus and method are corrected for in the vicinity of the manniquin by the camera tracking and later HitFilm tracking the cross on the abdomen...

    However... points not physically close to the abdomen cross point will see greater parallax related errors. If you wobble your head round but fix on a point on an object your eye tracking will correct in the vicinity of the point/object you fix on but the background will still wobble.

    In another thread the term depth of field cropped up  but now I am thinking we should talk in terms of 'depth of effective stabilization', in other words with a moving camera that has positional errors per frame then yes camera tracking and then later software tracking on a point can remedy those errors but just like depth of field there is a zone in the physical vicinity of the object where those corrections are most effective but outside that sweet zone they are progressively less effective.

    This partially contradicts earlier posts where I suggested that software tracking could resolve all - I was fixated on the object and treating background as out of focus window dressing without realising it could undermine the illusion of smooth movement.

    I may be able to side step the problems by looking at what kind of backgrounds I want to use in the 'perfected' process; if for instance I decide on a white defocused backdrop then the problem may just go away. I am looking for practical easy solutions here, flexibiliy is the way to go.

     

  • Not sure if it would help in your case, but GoPro Studio (free) used to have an "Exposure Match" (or similarly named) option in it.  Might be other software that also does it if you can't get it to accept your images. Don't know if the latest version - "Quik" - also does it, because I still use an older version, but loads of people have archived the older ones for downloading.

    Or you could try the Hitfilm Auto Effects; Color, Contrast and Levels, although, note: they are very slow when you specify an individual frame to compare against.

  • @Palacono

    Thanks for the tip, I took a look I could not easily find out but I am fairly confident that they would probably not support Nikon NEF raw format.

    My need for highest frame by frame quality probably means I am stuck with Nikon software, it does produce great results just not the best software to work with.

    I adjust exposure in RAW or at the very least in TIFF - regular 8 bit RGB will probably fall apart, not just loss of detail but also colours between frames may start to drift causing new flickering artifacts.

Sign in to comment