Here are photos of various pursuit camera setups using my testing technique invention:
Pursuit cameras can be as fancy as this professional SLR + heavyweight tripods
(RTings)
(HDTVtest.co.uk)
Or as shoe-string as this homebrew Blur Busters prototype setup (the first real one after the
successful sliding tupperware experiment). No camera rail needed, just a sliding tupperware container, lubricated with scotch tape on all sliding surfaces!
(Blur Busters Early prototype #1)
(Blur Busters Early prototype #2)
In fact, this prototype #2 setup -- the first real Blur Busters pursuit camera setup -- and even the homebrew setup above on vise-mounted wooden blocks, was the one that successfully created 1/10th pixel pursuit accuracy in
this LightBoost pursuit camera photograph of this
TestUFO motion test.
(Accuracy approaching tens-of-micrometers for a hand-propelled pursuit camera photo!)
This is a 93ppi gaming display (24" 1080p), using LightBoost blur reduction (
strobe backlight mode) at its lowest persistence mode, minimum strobe length (under 1ms strobe flashes, resulting in less than 1ms MPRT persistence). 93ppi translates to ~0.2mm pixel pitch. But look at how the tickmarks align visually perfectly (visually, it clearly shows less than 10% of a pixelwidth error -- each row of tickmarks are not even visibly offset by a tenth of a pixel at all!). Which means this photo, if 1/10th pixel error -- is a 0.02mm pursuit accuracy for this particular photograph. But let's be conserative and consider it a 25% pursuit error (1/4 pixel) = 0.05 millimeter pursuit error.
Despite the crude setup -- that's a sub-0.05 millimeter (
50 micrometer)
hand-propelled pursuit camera photograph! (93dpi display, 0.26 meter/second horizontal sliding speed). It took approximately 12 attempts (of a well-practice) before I was able to get such an accurate hand-pursuited camera photograph, with such a perfectly aligned temporal tickmark track. Even some 5-figured priced setups weren't able to get this pursuit accuracy.
Most reviewers are getting less accuracy than this, but it's not always critical to have extreme precision -- aiming for roughly 1/2 pixel precision (approx 1/4th millimeter). It takes a lot of practice, and the right amount of friction on the pursuit bar. I also experimented with various kinds of lubricant (unfortunately, those often make the pursuit slider very dirty), but with the right amount of friction, the camera can pretty much slide under its own momentum with nearly zero varying speed, helping the temporal tickmark accuracy greatly.
There's only so much accuracy you can get out of a hand-propelled camera, but even motion on Retina displays (4K monitors) are still definitely quite accurately hand-pursuitable -- after lots of practice with a well-optimized rail. The bigger the display, the lower the ppi, and the lower the pursuit accuracy needed -- but it's harder to slide the camera at a constant speed for longer period. Some reviewers, such as RTINGS, are now using much longer (4-feet-plus) camera rails to photograph large HDTVs.
Using a camera's automatic shooting or video feature helps a LOT, so you don't have to focus on button pressing -- video even works (pick clearest stillframe) as long as you can adjust exposure length per video frame -- you only focus on the sliding the camera from left to right edge. Depending on the speed of your camera's auto-shoot feature, you can take more than 10 pursuit photos per trial (one camera slide). For video, you can capture several dozen frames at 15fps or 30fps! Just make sure you use the least compression (highest quality stillframes). That can be a big time saver, if you use a zoomed image viewer to quickly hone into the accurately-pursuited photographs.
In retrospect, I find it extremely important to greatly stiffen the pursuit camera bar, to prevent flex/bounce. You know your pursuit is extremely smooth when you see screendoor row gaps being preserved pixel rows (LCD screendoor being preserved in the vertical direction, due to lack of vibrations during pursuit). The "scan line" effect is a great antivibration-verification, because you know you preserved the LCD screendoor effect.
List of Successful Pursuit Camera Mounting Methods
- Rail on wood blocks are excellent. If you're willing to do extreme homebrew, mount your slider on a thick piece of stiff thing and just mount ot on the same desk. The wooden-block prototype actually proved slightly more accurate than two average tripods. Use cheap vises to keep the wooden blocks mounted to edge of desk.
- Rail on tripods work if you use very heavy tripods combined with a very stiff camera rail (or a stuff panel underneath camera rail). And definitely anti-skid your tripod feet (rubber-feet, etc) so tripod slide doesn't interfere with pursuit accuracy.
- Vise-mounted rail (without wooden blocks) works well too. Basically a camera rail vise-mounted to the edge of a desk. Use very heavy, thick, stiff vises (whether commercial or homebrew vises, adjustable vises, fixed vises, etc) if you're hovering the rail in mid-air on vises. Alternative, cheap vises work well with heavy wooden blocks -- for homebrew variable-height camera rails. Alternatively, using the stiffness of the wooden blocks prevents the need for expensive stiff vises -- allowing the use of cheap $10 vises as pictured above.
- Sliding slippery block on desk. No camera rail needed! In earlier experimentation, teflon/polymer-bottomed block on a slippery melamine desk surface actually works surprising well! Using pieces of teflon-coated wood or metal as a guideway to slide against. Slippery teflon tape your friend too, on all sliding surfaces. It was unexpectedly almost as accurate as the Blur Busters wood-blocks prototype setup.
- Shoestring Tupperware Technique (photo) Zero-cost homebrew! I tested this. Upsidedown stiff tupperware or rubbermaid plastic container (sliding block), scotch tape (teflon substitute for all sliding surfaces) & a scotch-tape-covered flat edge (edge of a computer keyboard, edge of wood, etc) as a guide rail, as pictured. A point-shoot camera duct-taped to the stiff edge of an upsidedown plastic container, sliding on a desk or kitchen counter. Completely made from household items, it achieved verifiable <0.5 pixel accuracy (approx 1 out of 10-to-20 attempts) at 960 pixels per second on 24" screens! Takes more attempts to do accurate pursuit images, and scotch tape needs replacing after about 50 pursuits (becomes less slippery), but it actually worked surprisingly usably for pursuit photos of 24" 1080p computer monitors, and apparently this MacGyver setup functions better than expected.
- Hand-pursuit through mid-air Zero-cost extreme homebrew! Apparently, it's accurate enough for some reviewers (sub-pixel accuracy is possible at 960pps!). Rapid auto-shoot feature or video (with custom exposure lengths per frame, and a high-quality codec) is useful, since you might succeed H/V subpixel accuracy only in 1-in-100 photos. With auto-shoot or video (pick clearest stillframe), the job becomes becomes much easier.
EDIT: See new Easy Pursuit Camera Thread