Glowforge replicator head!


#1

Just a thought, but hey, maybe in the future the Glowforge could have a removable head that does 3D laser scanning?

So you place an object on the bed. The GF scans it and saves it as an editable a depth map file. Replace the head with the normal laser head, drop in your material and duplicate the object!

Is that actually possible?


#2

No need to replace the head. The current one has an IR depth sensor that could do the scanning!


#3

I wonder if the resolution would be high enough to do useful things with.


#4

We live in interesting times! :slight_smile:


#5

Word is it’s accurate to 0.1mm. That’s Z. I know I’m not in the Cool Kids Club so my opinion is almost meaningless to some, but it’s hard for me to imagine X and Y would somehow have a different resolution while scanning than it does while cutting.


#6

I see your point. I’d think that’d have to be the case.


#7

It depends on the effective aperture size of the measurement. For example if it shines a 1mm spot you are not going to get 0.1mm resolution in X and Y even if you move them in increments of 0.1mm.

Then there is bandwidth. If it takes a second to take a measurement then it would take forever to scan a large area but if it takes 1ms it could do them on the fly.


#8

I don’t understand why that’d be the case. And I’m not looking for any further technical response here. I just don’t get it. If you have a way to explain it to me in child-like fashion, I’ll happily accept it! :slight_smile:


#9

To get 0.1mm resolution you want the area being depth sampled to be 0.1mm or less. If it was 1mm then you would get an answer every 0.1mm but it would be an average of 100 pixels. I.e. it would be a very low pass filter.


#10

This would only get a top-down view of whatever the object was.


#11

Thanks. I think I’m getting a better understanding.


#12

Yep. You’d have to rotate it to get the entire object.


#13

Yes but you can only laser engrave a top down version of it anyway.


#14

Good point about the spot size.

Imagine a Belgian waffle…


Let’s say you wanted to measure how deep the holes are, what would you do? You’d probably get something small, maybe a toothpick, and put it down into one of the squares until it touched the bottom, then you’d pinch the toothpick with your fingers at the top of the hole.

When you pull the toothpick out you could then use a ruler to measure from the point of the toothpick to your fingers to find how deep the hole is.

That would only work because the toothpick is able to fit inside the squares of the waffle. Now imagine you don’t have a toothpick, all you have is a broomstick. If you were to try to put the broomstick into one of the squares it wouldn’t fit, it would stop on the top of the waffle because the broomstick is too big.

If the measurement laser spot is bigger than the feature you’re trying to measure, it probably won’t work.

The holes in this window screen might not be measurable if the laser spot is a couple millimeters in diameter.


#15

I see what you’re saying there. And I’ve given it proper thought and now I get it. (It takes longer when I’m reading this between conference calls all morning!)

However, I feel I should suggest… Don’t use something so delicious-looking as a frickin’ Belgian waffle as a visual example! It definitely obfuscates your actual point! :wink:


#16

Hehe, sorry! I probably spent a good 10 minutes trying to come up with a household item that had small, but not tiny, holes in it that you couldn’t easily measure with other means.


#17

Good question. Oddly enough, waffle maker is the 1st thing that comes to mind! Maybe fly-swatter? Weekly pill box?


#18

We would then get back x,y,z coordinates (assuming that the head had a sufficient range). These values would then be used how?

Possible issues:

  1. The map is an array that would need to be converted into a usable form for the laser since it hasn’t yet got a “real” 3d capability to accept 3d files. i.e. and STL mesh or other form
  2. The laser can only accept a greyscale image. How would the min/max values be limited (clipped) for absolute black/white from min/max engraving.
  3. The laser may incapable of cutting to the depth indicated by the scan. Would multipass cutting (refocused) then be involved?

#19

The biggest question I would have about laser scanning with the current head is how fast it could handle the incoming information. We know very little about the onboard computing power. I do know the homing process takes a lot of steps and has to access the cloud.

Currently the homing process is like this:

  • Gantry and head are in back left corner
  • Machine powers on, moves gantry forward to center of bed
  • Camera takes a photo of gantry, head position, sends it to cloud
  • Motion is calculated based on estimates from photo, head moves closer to center (havent seen it land in the right spot on the first try yet).
  • Camera takes another photo to see where everything is, sends to cloud, calculates distance it thinks it needs to go to be centered
  • Continue above step until it finally gets to the place it needs to be. Usually takes a couple more moves/photos. Reminds me of how they handle the mars rover etc

So for it to actually scan something and build a full 3d pointcloud profile from it is going to be questionable in a few areas. the first is the dotsize of the laser. Its currently pretty large, but we dont know what they are able to gather from that. Is it small enough to get a scan with enough detail to be usable?

Also can the machine hold enough data/transfer it efficiently enough to move at a reasonable speed. Who knows.

I think it might be possible, but I feel like it would be a lot slower than other things that are on the market already that can scan from multiple angles. I think for it to really be useful for anything actually 3d (not 2.5d) it would have to be a new head.


#20

That’s not a Belgian waffle - it’s an American Belgian-style waffle. Just a regular waffle with deep holes.

Here’s a real Belgian waffle:

Notice the caramelization and the sheen from the melted sugar. Real Belgian (Leige) waffles have sugar chunks (pearl sugar - size of pencil erasers) folded into the dough (and it’s a dough, not a batter - more like brioche than pancakes). The sugar chunks melt and ooze through the surface to create a glaze. Tres yummy! @bhspalinger should be able to give us the true scoop on these puppies. I make them for family gatherings. We don’t eat Belgian-style ones anymore :slight_smile: