360 Servo pendulum mount. Part 4: getting panoramic cover

OLYMPUS DIGITAL CAMERAWith the rig aloft turning in 25 degree steps as fast as the shutter release will allow I tied off and hoped for the best.

OLYMPUS DIGITAL CAMERAEven in a gusty Bft5 the 9’Levitation can’t lift my fully loaded bike. Wind speed 27102014as the wind approached its peak I decided I’d had enough of flying the delta on its ear with the camera dangling over the city and carefully walked it down.

Automatic stitching. To to automate panorama building I used the standard pano free-ware tools ‘AutoStitch’ and Microsoft ‘ICE’, neither of them allow much editing or control other than sifting the image libraries and retrying with or without suspect frames, I have seen some great results using these tools but they are designed to work with a fixed panning camera rather than the more mobile KAP camera. First off I used the Olympus EP1 as it can get a full 360 sweep of images without fail on the bench. Getting the rig moving swiftly from shot to shot is only half the story, even at 15 shots a circle in 40s the movement between shots is enough to confuse ICE auto-stitcher on 4 frames, the swing of the camera on the 2nd frame is all too disruptive:

PA271696_stitch 2k

pano8…AutoStitch makes a different kind of mess.

Manual stitch. Manual fitting got 11 frames together:

Assembly-1of which the centre 3rd formed an acceptable montage:

City of Cambridge 60mAGLDetail is what makes these things for me and the distance from the camera to subject puts the EP1 imagery on the limit, near objects are fine but the middle distance starts to look soft on close inspection:

EP1 manual fit pano1Better reso camera. For the next attempt I stepped up the camera resolution as, although the shot to shot rate was good with the EP1, the Eos M produces much better quality although its AF performance plagues its automatic 360 cover.

IMG_0084

The direct to servo mount relies on the metal gear of a TowerPro MG 995, I wouldn’t risk anything else!

Some of the frames were superb but the AF/AE let down the full circle cover:

OLYMPUS DIGITAL CAMERAthe camera panned faster than it can shoot for half of the sweep (I got very few shots of the ‘into the sun’ sector). I selected 2 consecutive frames for this stitch:

City of Cambridge 28102014and there was a good selection of single frames too:

_MG_9629-2despite the sluggish auto functions the Eos M capture is impressive.

pano10the AutoStitch result has a fish eye effect and despite a good blend along most of the join it suffers from some unpleasant blurring of detail:

pano10 croppano10 crop2ICE made a different kind of pixel pie of the join:

_MG_9602_stitch cropPan speed. I’m determined to keep the camera panning as fast as possible to get the max shot count in the shortest time with the minimum movement of the rotation centre. Using clickPanPro direct to an un-geared servo gets a shot to shot rate of around 2s.

Camera tilt. The easiest stitch is with the camera vertical but the lack of foreground detail is disappointing, especially as the low altitude makes for wide perspectives. The natural movement of the camera gives hints of the foreground suggesting tilt is a good thing but more tilt means severe multiple perspectives which are tough to resolve into a single view.  The more tilt the harder the stitch, so finding the magic angle is the next challenge.

Pendulum suspension. It’s advantage its robustness and ease of operation, its disadvantage is that motion damping is least effective on the kite line axis. Lateral swing is damped but ‘in line’ swing can happen faster than the pan step. Most of the motion in KAP is absorbed by the whole line-rig-kite system. A combination of flex in the kite, wind pressure on the line and the ‘hinge’ formed by the rig mass forms a slack point in the ‘U’ of the line which allows the conflicting forces to balance out, when wind speeds pick up the line jumps straight and movement at the camera is excessive: the pendulum swings violently until the system settles. Surprisingly the pendulum works well at 70cm with a short (15cm) pivot bar; big swings are rare (even in Bft 5) and the camera ‘floats’ through the air suspended at the slack point in the line.

Big swingMost of the motion seems to be caused by tugging on the line at haul down.

It’s disappointing to find the automatic stitching fails so badly, perhaps with a gentler wind I’ll get better shot to shot consistency. KAP panoramas are not easy but the results are rewarding!

B

About billboyheritagesurvey

Heritage worker
Gallery | This entry was posted in KAP and tagged , , . Bookmark the permalink.

One Response to 360 Servo pendulum mount. Part 4: getting panoramic cover

  1. Pingback: Around the Archaeology Blog-o-sphere Digest #9 | Doug's Archaeology

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s