Let’s say I want to make a CAD (or CAD like) file that that is a faithful 3D representation of a smallish relatively complex real-world object. Something with a lot of curves and angles, like a small statue or an irregularly-shaped piece of rock, or a pair of scissors, or a lemon meringue pie, or a shoe.
How do I do it?
I seem to recall seeing a ‘Making Of’ feature (maybe LOTR?) in which some special effects people used something that looked like a big hand held bar code scanner to record the shape of a rather complex clay model.
Is this something an average person could afford to have done? Once you had the 3D file, how could you reproduce the object? Thanks.
We have one of the little 3d scanners like the one in the Leno video at a lab I work at. It’s a pretty nifty little thing. The video says it’s “less than 3,000” so I’m guessing 2,999? Still pretty cheap compared to what they used to cost.
3d printers, on the other hand, are still quite expensive to buy and run.
No, I just move the camera a bit between shots, and then turn them colors and merge them after I have them on the computer. My results are inconsistent at best, though, and it probably would work better if I had a special rig.
I’m a 3D nerd from way backWhen I started making anaglyphic stereograms, I was using colour filters on double exposures with a creaky old SLR. (This was in the eighties.)
Photoshop makes it a damn sight easier, eh what? If you just replace the red channel, you end up with something that looks almost like a normal photo, if the parallax shift isn’t too extreme.
Too bad you said “smallish”. At one time I was consulted about recreating the line drawings for a large wooden hull built in the early 20th century and I came up with the idea of placing several microphones at different reference locations and then go around banging the hull with a hammer. The time taken for the sound to reach each microphone would pinpoint the location of that point. The acquisition and processing of the information could be done by computer with relative ease. I will mention that both halves of the hull would have to be done as it was not absolutely symmetrical.
A long time ago I played with a piece of photogrammetry software that created a 3D model from a series of photos of an object on a turntable - the lighting stayed the same, and the object was rotated by a specified number of degrees per image. That worked ok for simple models. You can do something similar here - only two photos needed, and not great results, but …
I also played with a 6-axis 3D tracking arm, which dumped data out via serial, denoting the x,y,z of the stylus tip, with a high degree of accuracy. One of the engineering teams at the place I worked did a lot with machine vision - mainly focused on identifying cuts of meat (foreleg, hindleg, shoulder, rib, etc). Early 90’s, so pretty basic compared with what can be done now.
3D images are much easier to make now with a digital camera. You can make a rig out of a piece of wood cut in an L channel shape.
If you use one camera then you have to take a picture and then slide it over and take another. Anything that moves in the picture will be noticeable. I use 2 old 3 mb cameras side by side and take the pictures at the same time. The wood holder is screwed into a standard tri-pod (it should be level to your subject). When you screw the cameras down to the wood holder (I use wingbolts) you have to align them visually at an object so they are matched up.
To create an image you have to stack the pictures on top of each other (any photo software will do) and then crop the center section out so the ratio of length and width match the stereoviewer you’re using. It doesn’t have to be perfect because different brands of stereoviews used different ratios. Anything that is roughly square will work. Once you’ve cropped them (it has to be a centered crop) then you cut and paste the pictures back to a left and right image and print to the size you want.
Surveyors use an instrument called a “total station” which measures horizontal and vertical angle, and distance, from wherever it is to whatever you point it at. Now these are being targeted at people who do forensic examinations and whatnot, to characterize the chair Schmidt sat in when he was shot, or whatever.
If you are inside the 3D object and want its inner surface, this would be a snap.
If you are outside, you would set the instrument up in enough different places that it would be able to reach every spot you want measured. You’d also have to take some reference measurements to be able to reconstruct the relative placement of all your instrument setups.
These things are unspeakably cool. I have one, a very low end one, and it is amazing what it can do.