Bug on slices display?

[Loïc Fejoz]

Hi team,

Using IceSL is a real pleasure! So much easier to write than OpenSCAD.

I will show you later what I have done and provides few remarks.

Yet, on the screenshot attached, you will see what I believed is a bug. Indeed one can see both the slices and apparently the plain shape. Even worst, they seem to be misaligned!

Merry christmas.
Loïc

[sylefeb]

Hi Loic,

This is indeed suspicious, the layers should always align with the object. Can you please send us your lua script in PM? We'll look into it asap.

Thanks -
Sylvain

[Loïc Fejoz]

Hi Sylvain,

The model is now on github (without documentation yet) :
https://github.com/loic-fejoz/loic-fejoz-fabmoments/blob/master/couillard/couillard.lua

The display bug also depends on the point of view...

I believe there is another bug but maybe it interferes with the printer configuration.
Indeed the first layer of the counter-weights are printed as if it is the interior of the object and not full-filled layer.

I thought of a few improvement on the language but it is probably better to talk directly together. Will you or someone of your team be there on Wednesday 7th January ?

PS: I told you about the tinwhistle. There are now photo of the original and printed results : https://github.com/loic-fejoz/loic-fejoz-fabmoments/tree/master/tinwhistle .

Best regards,
Loïc

[sylefeb]

Hi Loic,

I just fixed a bug regarding layer display, which I think matches the issue you described in your original post. Maybe you can confirm this was it: If the layers are displayed and the window is resized (e.g. enlarged in width) then the layers do not correctly follow the object.
This will be fixed in the next update. It has no consequence on printing (but in general I advise to not print if anything looks incorrect).

I'll look into the other issue you mentioned.

Tinwhistle: great print!

Best,
Sylvain

[sylefeb]

Regarding the fill-up of the counter weights it seems fine here, see attached images.
It might be that you are not using enough covers? I typically use 6 (there is a known issue that using 1 is not always reliable and might miss some boundaries ...)

Pro-tip: press 's' to get a screenshot in IceSL ;)

[Loïc Fejoz]

Hi,

For the display bug, you are right. Resizing is the only way to reproduce this problem.

For the counterweight, I didn't tell but, first use the print3d() configuration (uncomment on last line) then slice. In the screen shot attached, I have even commented the other pieces.
As you can see, there is the exact same problem, and still, I am using 6 covers and 2 shells.

I have published a photo of some attempts to print it (in 3mm and 5mm width).
https://github.com/loic-fejoz/loic-fejoz-fabmoments/tree/master/couillard

There are some troubles but I haven't set the min time per perimeter, nor the process thin feature. I will give it another try.

Thanks for the screen shot tips.

[sylefeb]

Hi Loic,

For the thin vertical poles you definitely need 'min time per layer', especially with PLA.
You should not need thin features on this model? This is only when the model has a feature size smaller that the nozzle (0.4 mm) which is generally to be avoided.

Regarding covers: I was able to replicate the problem here - I confirm that is a bug. We'll look into it. Thanks!

Sylvain

[sylefeb]

Hi Loic,

Nice catch, this bug was due to small numerical issues at the boundary of the scene bounding box. I fixed it, it will be fine in the next update.