I think it was Magic Canvas commenting on my going on 150 tries and not liking what I was getting 
7 Likes
It must be like that Calvin thing you see on the back of pickup trucks.
8 Likes
That would be an awesome puzzle too
3 Likes
I guess this is sort of laser related, but wow. There are hacks and then there are hacks.
Using the internals of a DVD drive as a microscope is wild.
9 Likes
That would make a great idea for the hopper! The Glowforge already uses a light laser to measure height. How hard would it be to make that measurement over a grid and output it as a grayscale image?
You could do an Easter rabbit. that would be great in the hopper!
5 Likes
Hah, this is a funny idea. Sadly, pretty sure that this was in essence what they were trying to do with continuous autofocus, and as we know that never happened, The glowforge measures height via parallax effects on the red laser dot. I’m not sure what trouble they ran into on the GF side that prevented that feature[1], but in terms of an optical microscope: The resolution won’t be great [both x/y and also height, we could imagine how to math it up as some combination of trigonometry, camera resolution size, and laser dot size], and there are lots of challenges in handing sheer drops along the Y axis, the laser (and therefore height measurement) will “skip”.
Did they ever explain why not? I always suspected it was a combination of physical limitations of the resolution of the autofocus system, software limits in terms of pathing data format and complexity, and technical limitations of the hardware/server side. ↩︎
3 Likes
I wasn’t thinking microscope, just 3d recording. Drop-offs would be a blank but could be filled in as last known height before the drop-off. It might not be all that useful but would be click bait for selling the premium and a fun programming project for someone. It might even lead to better lasering of warped wood!
3 Likes
A thought experiment: let’s assume they just pull one measurement per square inch on a grid. Given how long a single autofocus takes I have a feeling that scanning an entire sheet of material would be very slow indeed.
Nerdy math that's off topic since technically this is for nonlaser things :)
For a full sheet we’re talking about 10x18-ish measurements (since we can’t measure at the extreme edge) which means 180 photos that need to be processed and then heights calculated. The data transfer will start to be significant, and the compute time even more so. We see one autofocus taking … I don’t know I haven’t timed it, but optimistically 10-15 secs for one seems fair.
15 secs times 180 = 2700 seconds, or 45 minutes, just to process one sheet’s height measurements.
Even if we assume I’m off by 50%, that’s 27 minutes. I’m not sure the math works out to make this worth it, and that’s ignoring the cloud computing costs that GF would have to eat.
So without completely upending how all of this works I don’t see this happening. I doubt the image processing can be done locally on the GF hardware, that’s a pretty heavy process, and one of Glowforge’s strengths lies in that they can tweak how all this works in the cloud.
You could limit the scan area to the material it detects or a user defined area, but now we’re talking about a whole other set of rabbit holes to run down – cross-platform UI design, ML image processing, etc. Glowforge doesn’t pay me to do it, so this is as far as I’m going to go. Besides, they really tried to make it work and failed, it’s a special kind of hubris to think that I have some magic formula that they didn’t try – they have way more in depth knowledge of how their systems work.
I’m curious how xtool thinks they’re going to pull off the automatic height scanning. A dedicated hardware distance scanner attached to the head is how I’d do it, just a matter of finding the right rangefinder that works with sufficient resolution and positional accuracy at the distances we’re talking about here. Seems like someone’s already solved this, wonder if @shop knows?
2 Likes
i don’t.
i do know that the universal PLS6 in my office has both a manual focus tool (basically a puck on a stick with a ledge on it that you raise/lower the bed until the ledge hits a spot on the lens housing) and an autofocus tool. we’ve never used the universal autofocus because the rep who installed our machine said, “just use the manual puck. it’s a lot more accurate.”
now, to be fair, that was jan 2018. it’s been 5 years, they may have improved it. the GF focus button didn’t exist yet either. we were still entering manual heights for any non-proofgrade then.
2 Likes
How much of that is startup? A more reasonable measure is calibration and it would not take that long to do the whole sheet as it need only compare existing calibration numbers.
A much smaller piece say 6"x6" could be the same number of measuring points but more concentrated. The original suggestion would be more like a continuous recording like a 240 LPI engrave at 100 speed long but not ridiculous and no need to watch as it will not become fire.
1 Like
Not really… that’s still one photo. Grated the score process takes some time, but the compute and image processing is probably a significant issue.
1 Like
I thought it was a white dwarf sucking the energy from its partner.
3 Likes
11 Likes
I have seen several videos on electroplating 3D prints (though, not this one, yet). It seems like a lot of work but, people are getting great results.
5 Likes
that would look really cool engraved
1 Like
Have you done anything with this one? (I’m not yet caught up with posts so if you posted about it and I haven’t seen it, forgive me.)
7 Likes
Nope, it was just a compare-and-contrast between the various image generators out there.
In terms of lasering something, i think @shogun won that contest hands down. His lighthouse box lid was a really affective engrave, I saw it in person last weekend. It has a clarity of subject and a crispness that I think is great.
5 Likes
3 Likes
Now might be a good time to mention this extension:
Chrome
Firefox
It removes the obnoxious thumbnails from youtube and makes it far less “ugh”. This video looks like this on my feed:
Without it it feels like everyone on youtube is some Crazy Eddie-style billboard. Their videos are INSAAAANE.
EDIT: OK that led me to wiki, which was wilder than I thought…
I grew up with these ads all the time, and never knew about all the fraud stuff. I also never knew that Crazy Eddie was local to the northeast US. Interesting.
7 Likes