Age | Commit message (Collapse) | Author |
|
|
|
Need to make proper constructor to initialize
the `data_` from a range
|
|
The inheritance scheme was proofing to give a lot subtle errors
when using and compiling different test scenario's. Which proofs
the best practice of not inheriting from std containers.
Moving to a compound pattern, seems to be the better choice if
we don't want to use free-functions but dedicated containers,
for Points and Polygons.
This does require a lot of boilerplate code to follow to
the practice of the least surprises.
|
|
Just me playing around, might make it in, might not.
|
|
|
|
|
|
|
|
Both implementations should be equivalent
|
|
[CURA-9486] Increased seam randomness
|
|
Instead of two very similar loops, one for searching forwards and one for searching backwards, put all differences between the two into functions, so the main loop can be rewritten as 2 iterations over one sub-loop.
part of CURA-9486
|
|
|
|
Ultimaker/CURA-7271_slow_lines_individually_minimum_layer_time
Cura 7271 slow lines individually minimum layer time
|
|
part of CURA-9778 | PP-71
|
|
part of CURA-9778 | PP-71
|
|
part of CURA-9778 | PP-71
|
|
CURA-9074 fix tower support
|
|
suggestion from code-review
CURA-9549
Co-authored-by: Joey de l'Arago <j.delarago@ultimaker.com>
|
|
Only update score if it is improved more than $\epsilon$. As the start index is now predetermined by picking the vertex closest to some target location we favour this starting point. This starting point is now favored over any other candidate with a very similar score. This resolves the issue where the seam was spread more or less randomly over circular shapes when seam-preference was set to sharpest corner.
CURA-9486
|
|
This reverts commit 1af8653c
|
|
Brim overhaul: New features and Bug fixes
|
|
part of CURA-9066
|
|
... but should probably be an actual setting now at some point.
part of CURA-9066
|
|
part of CURA-9066
|
|
part of CURA-9066
|
|
Formatted README
|
|
part of CURA-9066
|
|
Makes it more clear that you can also just get it for all extruders.
pat of CURA-9066
|
|
Replace redundant booleans with EPlatformAdhesion, replace 'manual' variant (save both Polygons* and int) with std::variant<Polygons*, int>.
part of CURA-9066
|
|
Also make formula more clear. (And slightly better -- if we ever had a ridiculus amount of extruders, it would become incorrect.)
part of CURA-9066
|
|
The variable 'is_last' is a perfectly 'normal' part of the object, not a cached value, mutext, or whatever that's part of the 'meta-data' of the object instead of what it 'is'. It makes way more sense to remove the const qualifier(s) in this instance.
part of CURA-9066
|
|
The seam placement in 4.13 was desirable. We weren't able to copy this behaviour over to 5.x as the change in polygon gave unpredictable results. This was due to micro segments in the outline resulting in deviant angle calculations. Now that we have a more robust angle calculation we can revert to this desired behavior again.
CURA-9486
|
|
As we removed the tie-breaker the order of the input points does not matter anymore; starting at any position in the polygon will result in the same vertex being selected for the seam. It is thus not required to calculate the start point for the polygon, and it suffices to simply start at index 0.
CURA-9486
|
|
CURA-9486
|
|
We got better results without, and it is just adding unneeded complexity.
CURA-9486
|
|
When angles are calculated on very small edge lengths the calculated results become very inaccurate. By imposing a min_edge_length we circumvent this issue.
CURA-9486
|
|
CURA-9486
|
|
CURA-9486
|
|
|
|
CURA-9486
|
|
'origin/CURA-9486_hide_seam_increased_randomness' into CURA-9486_hide_seam_increased_randomness
# Conflicts:
# include/PathOrderOptimizer.h
|
|
But rather look at all points in a 1mm range. This provides the best of both worlds; sharp corners with lots of points are still detected and each point is a candidate seam location
CURA-9486
|
|
But rather look at all points in a 1mm range. This provides the best of both worlds; sharp corners with lots of points are still detected and each point is a candidate seam location
CURA-9486
|
|
This reverts commit 4179bd4a08b5ab69ddd438959c04eac79d49c20d.
|
|
CURA-9074
|
|
CURA-9074
|
|
relates to CURA-7271
|
|
Co-authored-by: Remco Burema <41987080+rburema@users.noreply.github.com>
|
|
CURA-9486
|
|
user specified.
The simplified polygon was causing a lot of noise in the seam position since it was basing the seam location on simplified vertices.
CURA-9486
|
|
instead of an arbitrary epsilon value. This will result in the same behaviour for both very large and very small models. Without this change, extremely large models would never run the tie breaker and extremely small models would always run the tie breaker.
Only update best_score if we are changing the best_point. The previous implementation could result in slowly increasing scores being assigned to best_score but best_point never being assigned.
CURA-9486
|