Allow extrapolation of autleveler scan data again

[lorenzresearch]

Description

Dear devs,

I am a happy user of UGS for years now but when I upgraded from 2.1.12 to 2.1.18 I ran into a problem that is a show-stopper for me.

In one of the releases between 2.1.12 and 2.1.18 the decision was made that autoleveler scan data should not be applied outside the scanned area. I understand, and agree to some extent, that this is for safety reasons. From a mathematical point of view this is not needed as long as the distance is small compared to the grid spacing. The bilinear interpolation works just fine for extrapolation. The routine that finds the closest grid points clamps it to the limits of the array already.

However, this excludes completely the use case of chamfering a workpiece that has already the final dimensions. I can only scan the surface close to the edges. Chamfering means the tools is moving just outside that scan-able surface. This makes Chamfering impossible now.

I have looked into the source code and I have found the explicit exclusion of coordinates outside the surface scan. I attempted to fix and compile the source for my own use but I am not willing to fix the dependency hell that is caused by requiring specific Java versions etc.

I have now written a script that extrapolates the surface data. (Side remark: I noticed that a constant grid spacing is required that is implied from the spacing of the bottom left four points. Still, the grid data is written with XY coordinates. From a numerical point of view this redundancy is very dangerous! You should either use XY coordinates or define a spacing in the header, not both)

I would like to see the use of extrapolated surface data coming back to UGS. Yes, enabling it could be hidden in "expert settings" and behind a lot of "This could be dangerous!" warnings. But currently I cannot use UGS for a rather standard CNC use case without external processing of the surface data. And that's a pity.

Thanks for listening/reading,
Eric

[banana-legs]

I just trashed milling a circuit board by not spotting this change, and broke the 0.3mm square-end milling bit in the process as it dived in far too deep for the bit. The circuit is just a shade smaller than the copper-clad blank and so I have to probe 5mm in from one of the board edges to avoid the contact clip on the board, and then rely on just a little bit of extrapolation to get the tracks on the edges of the board to mill ok. Rolling back to 2.1.16 seems to have a working autoleveler.

  Evan

[lorenzresearch]

@banana-legs another good example for where this new behaviour creates serious problems. I do PCB milling, too, but luckily I have become aware of the issue before I upgraded UGS on the PCB machine. And luckily the reference point (0,0) was always the highest point for the CNC I use for chamfering. If other areas would be higher, not applying an extrapolated scan would have trashed material, tools, or the whole machine for me, too.

Let me revise my statement above: I do understand the intent behind the change but not extrapolating the surface scan can create dangerous situations! (And not showing a warning about it is not really responsible)

I'll stick with 2.1.16 until this is fixed

[breiler]

This change is not intentional. I am currently on the other side of the world from my computer so I can't look into this at the moment.

[banana-legs]

@breiler no problem and I understand you are busy and give up your time for this excellent code; many thanks!. I agree that problems can occur with extrapolating too far, so I am with @lorenzresearch in that having a warning that the milling will be in an extrapolated region, and then an option somewhere to use extrapolated values would be ideal if possible.

I was unlucky in that my lowest point happened to be 0,0 and when the z was clamped to zero at the far side of the board, it created a step that was too big for the tiny milling bit. I am fine using 2.1.16 too for now.

     Evan

[lorenzresearch]

@breiler thanks for recognizing the issue and looking into it when you find the time!