Backend Changes for File Rendering?

Community Support
1

I have a file with far too many colors because it requires quite a few distinct steps. I know that these take a long time to load.

I have loaded this file before and it took maybe 10-20 minutes to upload and render. After having uploaded it, I was able to re-open that file and I never timed it, but I didn’t feel it was a bad wait time, so it had to have been loaded in under 5 minutes.

But now, I made some changes to that same file and attempted to load it. 3 hours later the rendering your file hasn’t cleared. During that time I broke the file in to 7 smaller files with parts of the full set and loaded each of them individually. Each of those took 10 minutes or so to load, taking longer with each one and hanging up on the 6th (I still haven’t loaded the 7th).

In the time I loaded those six, the original file (still open in a different browser, I believe the large file was in Chrome and the smaller files were in Firefox) was still attempting to load and had not managed to do so.

I refreshed the page on the 6 smaller chunks and that took somewhere between 2 and 3 hours to render the design and get me back to seeing all 6 loaded so I could attempt to load the seventh (sadly I had given up on it and clicked to come to the forums as I opened the laptop, the page loaded the design just AFTER I clicked on Community – which by the way STILL does not default to opening in a new tab which I had suggested long ago and is a trivial and USEFUL adjustment).

So… I have a file that I used to be able to open which is already uploaded which can now not open even when allowed 1.5 hours to load using a freshly rebooted computer and an incognito browser (so as to avoid any cache or system resource issues at all). And I have a near-duplicate of that file which I cannot load fresh even when allowing 3+ hours to upload and render. In both cases the files show fine in the GFUI preview, but do not render for me to put settings and send them to print.

Has something changed in how files are processed which makes excessive steps no longer viable? I really and truly do need all 267 steps in the file.

2

So I can’t comment on what might be slowing you down, but I’m curious about this. 267 seems like … a lot. Mind telling why you need so many?

3

The short answer is, what you’re doing is so far beyond the expected use of this machine, it’s not going to be supported. I also would argue it’s not reasonable to expect that to change.

The UI gets flakey once you get above a dozen or so steps, but that’s not a hinderance for virtually all users.

4

Try closing them all, on all browsers, rebooting your router and machine, and loading up one renamed copy fresh from your desktop, to the Dashboard. (Library.)

(And if that doesn’t do it…the file itself might have been corrupted by the changes you made. I’d check that over carefully for duplicate paths and unintended overlays.)

Oh, and there’s no reason to let it load for 3 hours…if it doesn’t make it in about 20-25 minutes, it’s stuck processing. Close and reboot and try again.

2 Likes
5

It is a massive test file for the full range of Pro machine power/speed combinations. Many materials cannot use the full range because the slow speed high power will burn through, or the high speed low power won’t mark. But some hardwoods (purple heart) and some metals (titanium) can show distinction across the entire range.

While I could test the full range over the course of 27 file loads and button presses, being able to run a single file in a single press was great while it lasted.

My issue is that the machine WAS able to do this, and now cannot. So the machine did change, AGAINST my use case.

File isn’t corrupt, and the thing which is different is that the copy of the file already on my dashboard, already used in my machines before, already re-opened multiple times on multiple computers… THAT cannot even open now. Unless the version stored now in the cloud has also gone corrupt (which would be a different issue, but still an issue), the problem is in how things are being processed/rendered is changed.

2 Likes
6

Or that file got corrupted in the server/cloud somehow. Closing it, deleting it from the Library and reloading it to the Library might work. (We just saw that with another file recently…I believe that’s what the OP had to do to clear the issue.)

Did you keep the original on your computer I hope?

7

I have the original on computer and in emails. I can give a shot loading the original file on the original computer which had uploaded.

8

It might be necessary to delete the one on the Dashboard first. But good luck!

9

OK let’s assume you actually need this, why not take advantage of the vary power system to get the power levels you need and then you only need to have one operation per speed?

You could make a single stepped gradient for each speed and set vary power at 100 (or full but that’s problematic for recreating what number it means) power setting, and you’re golden.

Just use the hex codes to precisely get your blocks of color, given that vary power uses the greyscale value. Remember, a hex code for a true grey is equal parts R G and B, [e.g. #111111, #222222 or #A1A1A1] so you can extrapolate the power # in a true grey to be (any channel)/255*100.

So, 10% power (decimal 25.5, rounded down to 25 and converted to hex as 19) ought to be #191919,
20% (decimal 51, converted to hex 33) ought to be #333333, and so forth. Make that gradient and copy it… 10? 20? times and set speeds accordingly.

Anyway, it’s theoretical, but I think it ought to work.

4 Likes
10

Sure, I get that, but any changes they make are not going to be tested against a use case like this, regardless of whether it worked in the past.

I can relate to what you’re trying to do, but having started down that path myself previously, I quickly came to the conclusion it wasn’t going to be of much benefit over other test methods that have been shared here many times before. A handful of tests allow you to quickly narrow down settings that work well.

11

Potential. I have thought about that as an approach. But the whole point of testing is to know what settings will result in the desired outcome for future products. Vary power may not give me enough of a span at a specific power to say “Yeah, that is the color/depth I want for this project” and I would also have to guess at the actual power by measuring distances out.

I am hopeful that if I go with Vary Power I am able to get multiple gradients in a single file to run them at a multiple speeds. But it seems likely that is a sticking point, and I need to test at a LOT of different speeds.

12

No, you don’t.

There is more variability across a single piece of the same material (let alone different sheets) than settings you would need to test. You choose a working range, usually no more than 100 difference in speeds, and a couple of points in-between. Then narrow it down one or two more iterations.

13

No you wouldn’t.

Just make a standard image with however many grades (essentially blocks of greys… 10? 20? however many you want to do) you want and then label them with hershey text scores. Make a grid and place that text at the top as a guide… then you just read down.

I will say though that you are splitting some mighty fine hairs and that this sort of power grid is usually not super necessary. Maybe your needs are different here but there’s a reason why most of us don’t get this detailed on the settings, there’s a cost/benefit curve to this kind of analysis and what you’re describing is usually on the wrong side of it.

People have posted highly detailed grid tests like this before, hunt around, it’s out there. I feel like maybe @rbtdanforth went down this particular rabbit hole once?

3 Likes
14

In fact all you need is the most extreme depth with vary power. White will not leave a mark and black will go that extreme depth. The gradient will go to everything in between. Even engraving top to bottom of the total half inch range I would defy you to define two different shades next to each (123vs124) I have tried to work with images with not enough shades available and even if you use the contrast you get the “rice terraces” effect as shades are 30 or 40 numbers apart but in practice you adjust the contrast to be at each extreme without to many levels at at either black or white. If you can see the details they will show up.

If you want to get good master files you can make one set for each material like this one…

Then with a max of two or three materials representing a range of basswood to walnut all 3/8 thick do a range of speeds from 4000 to 1000 with a few LPI from 1355 to 340 full power and single depth and perhaps the same with the dithered choice.

From such a set you could easily windage the ideal settings for just about anything.

1 Like
15

I appreciate people trying to help. I REALLY do. But I do not appreciate people telling me to bugger off and that what I want is wrong. I work in physics. I do lots of VERY strange things. For many use cases, you are likely quite right. But I am often doing things well outside the normal.

If you have comments helpful about doing something I am trying to do, I would love to hear them. But please stop telling me to just not bother. I could do the not bother without coming to the forums. I am here to figure out if anybody else has noticed issues loading things which previously could load, and if they figured out a way to deal with it.

I am not saying “please don’t answer me if you aren’t Glowforge staff” like so many people do in here. I posted instead of emailing because I do want community input. But what you have posted repeatedly doesn’t read out any assistance, just “stop doing that”

Anyway…

I took the suggestion of trying out Vary power, and the results I got are showing further issues.

image
image612×816 167 KB

This is a shot of the full sample I ran. I am not so great at approximating distances or something, because the numbers are the speeds I ran, and I tried to space them out when it stopped showing visible differences in the engrave, but I had a LOAD of extra space at the end. There should be 22 different areas, that is how many equal size gradient bars I placed in my file. But when I entered settings, it only gave me 19 fields to enter speeds for. In the interface whenever I selected one setting it highlighted an equal size bar, and I organized them to be in proper order from top to bottom of the piece.

However, the extra space here does show possibly room for 3 more total bars worth of power.

image
image612×816 206 KB

The low speeds all behave as I would expect, clear difference in output/results.
image
image612×816 198 KB

Pushing just barely above “normal” speeds I can still see distinction between the output. I do expect on baltic birch plywood to stop seeing clear differences, even though I increased the speed jump per line.
image
image612×816 184 KB

Approaching 3000 speed I stop being clear on where there is a change in speed. But I would expect 4000 speed to have almost zero marking capability even at full power.
image
image612×816 229 KB

And here at the end because of the mystery 3 bars I never gave settings for, who knows what I am actually looking at. But there is no difference from 3000 to 4000. That may be true for Baltic Birch as a whole, but there is the issue I cannot put in a picture…

While this was running I noticed the lack of change back around 3,000. I watched the laser and it was moving well beyond the engraving range to slow down and speed up. But when I timed the laser head moving back and forth, there was no difference in the time to do 5 passes back and forth between what I assumed should be the 3300 and the 4000 speed lines (I hadn’t yet realized that I have 3 “bonus” lines somewhere)

I ran a new test on paper, doing only 3,000 and 4,000 speeds. Once again I timed the head, starting as it began to move left, and counting until it came back 10 times. At 3,000 speed this took 25.21 seconds, and at 4,000 speed 26.06 seconds.

I will assume I lost count in there, if the 3,000 speed had only done 9 return trips, then the time should have been 28,01 seconds instead.

image
image612×816 134 KB

Fortunately this does show a clear difference in results from 4,000 and 3,000 power. So I was incorrect in my belief about what would happen with BB Ply at the upper speeds.

I may be able to make this work by placing discrete color rectangles and saving them as PNG like I had to do for the gradients to work.

On woods and the like where you are removing material this will be true. But titanium and other such materials can behave very oddly at different speed/power combinations:

Figuring out from that test grid what you get for 68 power and 75 speed would be a complete guess.

3 Likes
16

It is certainly the case that a few materials have temperatures that cause special chemical reactions that there would be many special conditions as how much heating it would cause, Titanium and Niobium go through special color phases at specific temps but it would be temperature and chemical reaction rather than engraving, Similarly when marble id heated beyond a certain point it breaks down to Ca O and CO2 that then react with moisture in the air but again it is the heating and chemical action with the air. but those are specific special cases not connected to the normal use of the laser.

17

@jacobturner, did reloading the original file get the loading problem resolved?

18

Honestly I am always kind of suspicious about what is going on behind the scenes. The company tries to save us from knowing too much, but I think they go much too far in this effort.

Your processing observation is interesting and I hope you can resolve the issue to your satisfaction.

Please keep doing strange things.

3 Likes
19

No luck on that one. Neither opening what was already uploaded nor trying a fresh upload of the originally uploaded file on the original computer and network which had uploaded it.

20

I found this thread because my usual engrave test pattern (which I found on this forum some time ago) has been loading for minutes now. I can’t remember ever waiting more than 10s or so at “Rendering” with this file It’s 111 steps (10-100+Full power at 100-1000 speed, plus a legend).

There has definitely been a performance regression in the back-end. And I don’t see any announcements of new features that would justify such a huge perf hit.