P2KLPU .NET POC

This tool is used for Directly Connected Palette 2/2S to your Klipper Printer, like Voron using https://www.klipper3d.org/Config_Reference.html#palette2

It uses built.in Slicer wipe tower, there are plans to make calculation of wiping different, and this software recommending how much you need to add to not have short splices.

It requires 3 (4 optional) changes in PrusaSlicer against P2PP which requires complete rework:

A) Enable MMU (Single Extruder Multi Material in Printer settings)

B) If possible set everything in Printer Settings, Single Extruder MM Setup everything to 0, but this should be optional, if it does not work when you set some value, please create Issue, my goal is to strip all Prusa Commands so you have smallest setup possible

C) Print Settings => Output Settings => Post Processing Script => C:\Users\richa\source\repos\p2pp\dotnet-p2pp-poc\bin\Debug\net10.0\P2KLPU.exe;

Give it path to P2KLPU

D) In Printer => Custom GCode => Start GCode you add options below listed, for example my personal options are:

;P2KLPU PRINTERPROFILE=e666315ff39d9c78
; Printer Profile is random number you create, which represents your printer. If you omit this directive entirely, P2KLPU derives a stable profile hash from your PrusaSlicer printer profile name (printer_settings_id), so the same printer always gets the same ID automatically.
;P2KLPU SPLICEOFFSET=38
; Splice offset is offset of your mixing hotend part, if you have short hotend, you can have low offset (Almost 0), if you have long hotend, it can be up to 60. I have Dragon Water Cooled, with longer offset, 38mm works for me
; If your splices come too soon, make offset larger, if they come too late, make it smaller. Start at 25 and see. Depending on your tube length, one layer of large rectangle with 4 colors can be enough to do this.
;P2KLPU MINSTARTSPLICE=100
; This is just checking for your first splice, so it has at least 100mm, Palette 2S manual states so 85 is minimum, but I found 100mm as safe where filament does not break (It has tension from moving around Palete)
; Your value is honored as-is (no silent clamping); values below the manual minimum (85mm) produce a warning.
;P2KLPU MINSPLICE=70
; This is just checking for every remaining splice, so it is not shorter. If splice is too short, sometimes they can break inside Palette, Palette 2S Manual states 60mm minimum, I found 70 mm safe.
; Your value is honored as-is; values below the manual minimum (60mm) produce a warning.
; A splice below the configured minimum is an ERROR: by default (STRICT=1) the export fails so PrusaSlicer
; shows you the problem instead of silently producing a file with splices that can break inside the Palette.
; Add ;P2KLPU STRICT=0 to downgrade errors to warnings and write the output anyway.

;P2KLPU EXTRAENDFILAMENT=150
; This is how much extra filament there is at the end, if you have like me hotend/extruder path 100 mm, its nice to have 50mm above to grip the filament and pull it out
;P2KLPU LINEARPINGLENGTH=1000
; This is how often Palette PINGS with Klipper, Depending on your tube length, your result may wary. Too small number means overcompensating too often, too large number, means no compensation at all
;P2KLPU MATERIAL_PETG_PETG_3_-1_-5
; This says, if you meet PETG and PETG, and there are no other directives, 3 heat, -1 Compression and -5 cooling
;P2KLPU MATERIAL_IN1_IN3_2_0_-5
; This says, even if you know material mixing, if you are mixing INPUT 1 and INPUT 3, 2 heat, 0 Compression and -5 cooling
;P2KLPU MATERIAL_IN3_IN1_2_0_-5
; This says, even if you know material mixing, if you are mixing INPUT 3 and INPUT 1, 2 heat, 0 Compression and -5 cooling
;P2KLPU MATERIAL_IN4_IN3_2_0_-5
; This says, even if you know material mixing, if you are mixing INPUT 4 and INPUT 3, 2 heat, 0 Compression and -5 cooling
;P2KLPU MATERIAL_IN3_IN4_2_0_-5
; This says, even if you know material mixing, if you are mixing INPUT 3 and INPUT 4, 2 heat, 0 Compression and -5 cooling

; You can also redefine materials as PETG2, see below, but point is, I have Gray PETG in Input 1 and Blue Matte in Input 3, Yellow Matte in Input 4, even tho all are PETG, 3 and 4 mix great, 1 and 3 mix horribly, and 1 and 4 mix horribly, I am giving extra control to user to decide

This script requires .NET Runtime 10.0.1 https://dotnet.microsoft.com/en-us/download/dotnet/10.0

Output of the script looks like this:

=== P2KLPU Analysis ===
Display name: xxx_-_sga_top_pylons.mcf.gcode
Extrusion mode: Relative (M83)
Total positive extrusion: 116421.693 mm
RAW_MMU effective positive extrusion: 116421.693 mm
Tower effective extrusion: 32600.053 mm
Model effective extrusion: 83821.64 mm
Tower XY bounds (from ;TYPE markers): X[60.698,100.7] Y[55.336,171.396]
Splices detected: 236
Palette pings (O31) planned: 118
O31 encodes a ping location along the extruded filament.
- In Palette 2/2S connected mode, P2PP uses O31 Dxxxxxxxx where Dxxxxxxxx is the hex of the float32 bit-pattern (little-endian) representing millimeters.
- In Palette 3 mode, it can appear as O31 L<mm> mm.
Ping plan (1-based):
  #  O31              Location(mm)
  1  O31 D44757335          981.80
  2  O31 D44f5866b         1964.20
  3  O31 D453841ad         2948.10
  4  O31 D45759afe         3929.69
  5  O31 D45997819         4911.01
  6  O31 D45b821a7         5892.21
  7  O31 D45d6cdd5         6873.73
  8  O31 D45f57673         7854.81
  9  O31 D460a10f3         8836.24
 10  O31 D46196b78         9818.87
 11  O31 D4628c25a        10800.59
 12  O31 D463816bc        11781.68
 13  O31 D46476cde        12763.22
 14  O31 D4656c480        13745.13
 15  O31 D46661944        14726.32
 16  O31 D46756ed1        15707.70
 17  O31 D4682617c        16688.74
 18  O31 D468a0bc0        17669.88
 19  O31 D4691b5d8        18650.92
 20  O31 D4699609e        19632.31
 21  O31 D46a10b07        20613.51
 22  O31 D46a8b511        21594.53
 23  O31 D46b05f2b        22575.58
 24  O31 D46b8092c        23556.59
 25  O31 D46bfb343        24537.63
 26  O31 D46c75d44        25518.63
 27  O31 D46cf0948        26500.64
 28  O31 D46d6b373        27481.73
 29  O31 D46de5d7b        28462.74
 30  O31 D46e607e8        29443.95
 31  O31 D46edb1fb        30424.99
 32  O31 D46f561c1        31408.88
 33  O31 D46fd0ccd        32390.40
 34  O31 D47025b74        33371.45
 35  O31 D4706307e        34352.49
 36  O31 D470a058f        35333.56
 37  O31 D470ddaa9        36314.66
 38  O31 D4711afb5        37295.71
 39  O31 D47158505        38277.02
 40  O31 D47195ab3        39258.70
 41  O31 D471d2fc1        40239.76
 42  O31 D47210553        41221.32
 43  O31 D4724dac7        42202.78
 44  O31 D4728afc9        43183.79
 45  O31 D472c84ef        44164.93
 46  O31 D47305a02        45146.01
 47  O31 D47342f2b        46127.17
 48  O31 D47380438        47108.22
 49  O31 D473bd938        48089.22
 50  O31 D473fae58        49070.34
 51  O31 D47438382        50051.51
 52  O31 D47475893        51032.57
 53  O31 D474b2e68        52014.41
 54  O31 D474f041d        52996.11
 55  O31 D4752da42        53978.26
 56  O31 D4756af51        54959.32
 57  O31 D475a8466        55940.40
 58  O31 D475e5ad6        56922.84
 59  O31 D47622fe7        57903.90
 60  O31 D476604f6        58884.96
 61  O31 D4769da31        59866.19
 62  O31 D476dafa6        60847.65
 63  O31 D477184ac        61828.67
 64  O31 D477559ad        62809.68
 65  O31 D47792f04        63791.02
 66  O31 D477d0410        64772.06
 67  O31 D47806ca0        65753.25
 68  O31 D47825739        66734.45
 69  O31 D478441c2        67715.51
 70  O31 D47862c43        68696.53
 71  O31 D478817e3        69679.78
 72  O31 D478a0269        70660.82
 73  O31 D478becf3        71641.90
 74  O31 D478dd77e        72622.99
 75  O31 D478fc200        73604.00
 76  O31 D4791acd6        74585.67
 77  O31 D47939780        75567.00
 78  O31 D47958269        76548.82
 79  O31 D47976cea        77529.83
 80  O31 D479957b4        78511.40
 81  O31 D479b4237        79492.43
 82  O31 D479d2ce9        80473.82
 83  O31 D479f17a5        81455.29
 84  O31 D47a10292        82437.14
 85  O31 D47a2ed4f        83418.62
 86  O31 D47a4d801        84400.00
 87  O31 D47a6c282        85381.02
 88  O31 D47a8ad04        86362.03
 89  O31 D47aa97ae        87343.36
 90  O31 D47ac822e        88324.36
 91  O31 D47ae6de8        89307.81
 92  O31 D47b058b8        90289.44
 93  O31 D47b24343        91270.53
 94  O31 D47b42df2        92251.89
 95  O31 D47b61878        93232.94
 96  O31 D47b80334        94214.40
 97  O31 D47b9ee1a        95196.21
 98  O31 D47bbd8fc        96177.97
 99  O31 D47bdc3f3        97159.90
100  O31 D47bfae79        98140.94
101  O31 D47c19915        99122.16
102  O31 D47c383f6       100103.92
103  O31 D47c56ed6       101085.67
104  O31 D47c7598d       102067.10
105  O31 D47c9443c       103048.47
106  O31 D47cb2ed7       104029.68
107  O31 D47cd1983       105011.03
108  O31 D47cf0441       105992.50
109  O31 D47d0eeeb       106973.83
110  O31 D47d2d96b       107954.84
111  O31 D47d4c428       108936.31
112  O31 D47d6aeb0       109917.37
113  O31 D47d899bd       110899.48
114  O31 D47da84a3       111881.28
115  O31 D47dc6f7f       112863.00
116  O31 D47de5a1c       113844.22
117  O31 D47e044d1       114825.63
118  O31 D47e22f5b       115806.71

Splice plan (1-based inputs):
  #  From->To   Location(mm)   Length(mm)   Algo (Heat, Compression, Cooling) [h,c,k]
  1  4->1            5783.66      5783.66   3,-1,-5
  2  1->4            7084.91      1301.25   3,-1,-5
  3  4->1           11152.06      4067.14   3,-1,-5
  4  1->4           12455.33      1303.27   3,-1,-5
  5  4->1           16519.65      4064.32   3,-1,-5
  6  1->4           17828.22      1308.57   3,-1,-5
  7  4->1           18879.15      1050.93   3,-1,-5
  8  1->4           20322.88      1443.73   3,-1,-5
  9  4->1           21254.48       931.60   3,-1,-5
 10  1->3           21465.84       211.36   2,0,-5
 11  3->1           21721.28       255.44   2,0,-5
 12  1->4           21930.83       209.55   3,-1,-5
 13  4->1           22748.67       817.84   3,-1,-5
 14  1->3           22958.32       209.65   2,0,-5
 15  3->1           23210.52       252.20   2,0,-5
 16  1->4           23421.93       211.41   3,-1,-5
 17  4->1           24245.07       823.14   3,-1,-5
 18  1->3           24453.64       208.57   2,0,-5
 19  3->1           24707.96       254.32   2,0,-5
 20  1->4           24922.54       214.58   3,-1,-5
 21  4->1           25739.78       817.24   3,-1,-5
 22  1->3           25954.62       214.84   2,0,-5
 23  3->1           26211.21       256.60   2,0,-5
 24  1->4           26423.50       212.29   3,-1,-5
 25  4->1           27240.46       816.96   3,-1,-5
 26  1->3           27459.31       218.85   2,0,-5
 27  3->1           27711.50       252.20   2,0,-5
 28  1->4           27930.04       218.53   3,-1,-5
 29  4->1           28751.78       821.75   3,-1,-5
 30  1->3           28966.01       214.23   2,0,-5
 31  3->1           29220.33       254.32   2,0,-5
 32  1->4           29435.22       214.89   3,-1,-5
 33  4->1           30252.54       817.33   3,-1,-5
 34  1->3           30466.53       213.99   2,0,-5
 35  3->1           30723.13       256.60   2,0,-5
 36  1->4           30933.76       210.63   3,-1,-5
 37  4->1           31748.49       814.73   3,-1,-5
 38  1->3           31961.54       213.05   2,0,-5
 39  3->1           32213.74       252.20   2,0,-5
 40  1->4           32426.11       212.37   3,-1,-5
 41  4->1           33248.39       822.28   3,-1,-5
 42  1->3           33458.69       210.30   2,0,-5
 43  3->1           33713.01       254.32   2,0,-5
 44  1->4           33924.99       211.98   3,-1,-5
 45  4->1           34741.37       816.38   3,-1,-5
 46  1->3           34951.57       210.21   2,0,-5
 47  3->1           35208.17       256.60   2,0,-5
 48  1->4           35417.26       209.10   3,-1,-5
 49  4->1           36234.11       816.84   3,-1,-5
 50  1->3           36444.15       210.04   2,0,-5
 51  3->1           36696.35       252.20   2,0,-5
 52  1->4           36906.07       209.73   3,-1,-5
 53  4->1           37728.42       822.35   3,-1,-5
 54  1->3           37936.44       208.02   2,0,-5
 55  3->1           38190.76       254.32   2,0,-5
 56  1->4           38398.12       207.36   3,-1,-5
 57  4->1           39214.63       816.50   3,-1,-5
 58  1->3           39422.56       207.94   2,0,-5
 59  3->1           39681.95       259.39   2,0,-5
 60  1->4           39888.98       207.03   3,-1,-5
 61  4->1           40706.71       817.73   3,-1,-5
 62  1->3           40914.43       207.73   2,0,-5
 63  3->1           41180.67       266.23   2,0,-5
 64  1->4           41388.64       207.97   3,-1,-5
 65  4->1           42214.55       825.91   3,-1,-5
 66  1->3           42420.71       206.16   2,0,-5
 67  3->1           42692.80       272.09   2,0,-5
 68  1->4           42900.02       207.22   3,-1,-5
 69  4->1           43720.53       820.51   3,-1,-5
 70  1->3           43928.13       207.60   2,0,-5
 71  3->1           44209.93       281.80   2,0,-5
 72  1->4           44414.05       204.12   3,-1,-5
 73  4->1           45242.94       828.89   3,-1,-5
 74  1->3           45448.39       205.45   2,0,-5
 75  3->1           45732.90       284.51   2,0,-5
 76  1->4           45937.14       204.23   3,-1,-5
 77  4->1           46770.61       833.48   3,-1,-5
 78  1->3           46973.57       202.96   2,0,-5
 79  3->1           47259.56       285.99   2,0,-5
 80  1->4           47465.49       205.92   3,-1,-5
 81  4->1           48286.50       821.02   3,-1,-5
 82  1->3           48490.22       203.72   2,0,-5
 83  3->1           48821.95       331.73   2,0,-5
 84  1->4           49025.13       203.18   3,-1,-5
 85  4->1           49911.74       886.61   3,-1,-5
 86  1->3           50115.13       203.39   2,0,-5
 87  3->1           50434.85       319.72   2,0,-5
 88  1->4           50646.31       211.46   3,-1,-5
 89  4->1           51739.14      1092.83   3,-1,-5
 90  1->3           51949.79       210.65   2,0,-5
 91  3->1           52258.88       309.09   2,0,-5
 92  1->4           52466.79       207.91   3,-1,-5
 93  4->1           53484.31      1017.52   3,-1,-5
 94  1->3           53694.52       210.21   2,0,-5
 95  3->1           54183.85       489.33   2,0,-5
 96  1->4           54392.91       209.06   3,-1,-5
 97  4->1           55048.74       655.83   3,-1,-5
 98  1->3           55255.47       206.73   2,0,-5
 99  3->1           56052.41       796.94   2,0,-5
100  1->4           56260.43       208.02   3,-1,-5
101  4->1           56546.19       285.77   3,-1,-5
102  1->3           56749.17       202.98   2,0,-5
103  3->1           57555.38       806.21   2,0,-5
104  1->4           57756.66       201.28   3,-1,-5
105  4->1           58043.21       286.55   3,-1,-5
106  1->3           58245.95       202.74   2,0,-5
107  3->1           59024.88       778.94   2,0,-5
108  1->4           59232.97       208.09   3,-1,-5
109  4->1           59519.31       286.34   3,-1,-5
110  1->3           59727.89       208.58   2,0,-5
111  3->1           60460.91       733.03   2,0,-5
112  1->4           60669.73       208.82   3,-1,-5
113  4->1           60955.50       285.77   3,-1,-5
114  1->3           61161.95       206.46   2,0,-5
115  3->1           61896.54       734.58   2,0,-5
116  1->4           62104.97       208.44   3,-1,-5
117  4->1           62391.52       286.55   3,-1,-5
118  1->3           62615.15       223.64   2,0,-5
119  3->1           63357.75       742.60   2,0,-5
120  1->4           63559.88       202.12   3,-1,-5
121  4->1           63912.37       352.50   3,-1,-5
122  1->3           64112.75       200.37   2,0,-5
123  3->1           64858.64       745.89   2,0,-5
124  1->4           65060.63       201.99   3,-1,-5
125  4->1           65411.66       351.03   3,-1,-5
126  1->3           65613.67       202.01   2,0,-5
127  3->1           66360.41       746.74   2,0,-5
128  1->4           66560.66       200.25   3,-1,-5
129  4->1           66942.63       381.97   3,-1,-5
130  1->3           67146.92       204.29   2,0,-5
131  3->1           67958.91       811.99   2,0,-5
132  1->4           68164.58       205.67   3,-1,-5
133  4->1           68835.19       670.60   3,-1,-5
134  1->3           68989.58       154.39   2,0,-5
135  3->1           69587.84       598.26   2,0,-5
136  1->4           69739.55       151.70   3,-1,-5
137  4->1           70909.63      1170.09   3,-1,-5
138  1->3           71061.49       151.86   2,0,-5
139  3->1           71330.80       269.30   2,0,-5
140  1->4           71482.65       151.85   3,-1,-5
141  4->1           72753.59      1270.94   3,-1,-5
142  1->3           72905.32       151.73   2,0,-5
143  3->1           73152.09       246.77   2,0,-5
144  1->4           73303.87       151.78   3,-1,-5
145  4->1           74407.99      1104.12   3,-1,-5
146  1->3           74559.84       151.85   2,0,-5
147  3->1           74806.54       246.70   2,0,-5
148  1->4           74958.25       151.71   3,-1,-5
149  4->1           75930.82       972.57   3,-1,-5
150  1->3           76082.62       151.79   2,0,-5
151  3->1           76381.40       298.78   2,0,-5
152  1->4           76533.23       151.83   3,-1,-5
153  4->1           78037.54      1504.31   3,-1,-5
154  1->3           78189.24       151.70   2,0,-5
155  3->1           78518.61       329.37   2,0,-5
156  1->4           78670.41       151.80   3,-1,-5
157  4->1           80004.46      1334.05   3,-1,-5
158  1->3           80156.30       151.84   2,0,-5
159  3->1           80516.39       360.10   2,0,-5
160  1->4           80668.10       151.71   3,-1,-5
161  4->1           82187.67      1519.57   3,-1,-5
162  1->3           82339.47       151.80   2,0,-5
163  3->1           82709.13       369.66   2,0,-5
164  1->4           82860.96       151.83   3,-1,-5
165  4->1           84438.00      1577.05   3,-1,-5
166  1->3           84589.72       151.71   2,0,-5
167  3->1           85181.71       591.99   2,0,-5
168  1->4           85333.51       151.80   3,-1,-5
169  4->1           86462.88      1129.37   3,-1,-5
170  1->3           86614.70       151.83   2,0,-5
171  3->1           87178.35       563.65   2,0,-5
172  1->4           87330.06       151.71   3,-1,-5
173  4->1           88518.33      1188.27   3,-1,-5
174  1->3           88670.14       151.81   2,0,-5
175  3->1           89211.32       541.18   2,0,-5
176  1->4           89363.14       151.82   3,-1,-5
177  4->1           90233.94       870.80   3,-1,-5
178  1->3           90385.67       151.73   2,0,-5
179  3->1           90917.21       531.54   2,0,-5
180  1->4           91069.02       151.81   3,-1,-5
181  4->1           91946.40       877.38   3,-1,-5
182  1->3           92098.20       151.80   2,0,-5
183  3->1           92557.09       458.89   2,0,-5
184  1->4           92708.82       151.73   3,-1,-5
185  4->1           93529.74       820.92   3,-1,-5
186  1->3           93681.55       151.80   2,0,-5
187  3->1           93959.72       278.17   2,0,-5
188  1->4           94111.52       151.80   3,-1,-5
189  4->1           95985.58      1874.06   3,-1,-5
190  1->3           96137.31       151.73   2,0,-5
191  3->1           96528.76       391.46   2,0,-5
192  1->4           96680.56       151.80   3,-1,-5
193  4->1           98780.08      2099.52   3,-1,-5
194  1->3           98931.87       151.79   2,0,-5
195  3->1           99258.95       327.08   2,0,-5
196  1->4           99410.67       151.73   3,-1,-5
197  4->1          101520.26      2109.59   3,-1,-5
198  1->3          101672.05       151.79   2,0,-5
199  3->1          101979.69       307.64   2,0,-5
200  1->4          102131.47       151.78   3,-1,-5
201  4->1          104459.43      2327.95   3,-1,-5
202  1->3          104611.16       151.73   2,0,-5
203  3->1          104877.86       266.71   2,0,-5
204  1->4          105029.65       151.79   3,-1,-5
205  4->1          105860.49       830.84   3,-1,-5
206  1->3          106012.27       151.79   2,0,-5
207  3->1          106291.71       279.44   2,0,-5
208  1->4          106443.39       151.68   3,-1,-5
209  4->1          107394.32       950.93   3,-1,-5
210  1->3          107546.08       151.76   2,0,-5
211  3->4          107833.84       287.76   2,0,-5
212  4->3          108825.33       991.49   2,0,-5
213  3->4          109324.91       499.58   2,0,-5
214  4->3          110013.64       688.73   2,0,-5
215  3->4          110825.74       812.10   2,0,-5
216  4->3          111199.33       373.59   2,0,-5
217  3->4          111901.04       701.72   2,0,-5
218  4->3          112188.30       287.25   2,0,-5
219  3->4          112412.33       224.03   2,0,-5
220  4->3          112631.98       219.65   2,0,-5
221  3->4          112859.96       227.99   2,0,-5
222  4->3          113080.76       220.80   2,0,-5
223  3->4          113300.28       219.52   2,0,-5
224  4->3          113519.38       219.10   2,0,-5
225  3->4          113738.66       219.28   2,0,-5
226  4->3          113954.98       216.32   2,0,-5
227  3->4          114128.39       173.41   2,0,-5
228  4->3          114345.70       217.31   2,0,-5
229  3->4          114494.03       148.33   2,0,-5
230  4->3          114753.05       259.02   2,0,-5
231  3->4          114900.55       147.50   2,0,-5
232  4->3          115141.78       241.22   2,0,-5
233  3->4          115290.26       148.48   2,0,-5
234  4->3          115524.21       233.95   2,0,-5
235  3->4          115672.54       148.33   2,0,-5
236  4->3          115866.70       194.15   2,0,-5

Input usage summary:
  Input  Material   Used(mm)   Min splice(mm)   Max splice(mm)
██ DI1    PETG       24200.60           151.68          1443.73
██ DI3    PETG       24855.82           147.50           812.10
██ DI4    PETG       67403.28           194.15          5783.66

Overall splice lengths: min #231 (3->4) 147.50 mm, max #1 (4->1) 5783.66 mm


Wrote G-code: C:\Users\richa\AppData\Local\Temp\.48308_4.gcode.pp

Press any key to exit...

In modern Windows Terminal it uses colors to make reading easier:

Please report any bugs with Github Issues, this script was heavily inspired by P2PP https://github.com/tomvandeneede/p2pp and https://github.com/vhspace/p2pp to both @tomvandeneede and @vhspace I am giving my respects to keeping this alive

Last final warning, this code was half written by GPT5.2, I have reviewed and written large part of code myself, but the project was too huge to analyze over weekened alone.... I will work on improvements myself.

Foot note, for development, clone the repo, have .NET 10 SDK, and use Visual Studio Code or Visual Studio 2026 Community, no special anything required, its console app

This folder contains a .NET C# proof-of-concept for a future rewrite.

Current focus:

• Drop GUI entirely.
• Primary target: Klipper + Palette 2/2S connected mode.
• Keep configuration inside the G-code via ;P2KLPU ... comment directives (so PrusaSlicer can drive it).
• Normalize emitted pauses (G4) to be more Klipper-friendly and optionally hook macros around ping blocks.

What it does today

• Accepts PrusaSlicer post-processing style invocation: input.gcode [output.gcode].
• Reads G-code, detects tool changes (T0, T1, … / ACTIVATE_EXTRUDER), tracks extrusion, and prints a splice plan.
• With --dry-run, writes nothing (analysis only).
• No CLI configuration flags besides --dry-run and --verbose.
• Supports in-file directives anywhere in the G-code (usually in slicer start-gcode comments), for example:
  • ;P2KLPU SPLICE_OFFSET=0
  • ;P2KLPU DEFAULT_ALGO=10,5,3
  • ;P2KLPU ALGO 1-2=12,7,0
  • Material-based algorithm overrides (no = form; legacy style):
    • ;P2KLPU MATERIAL_DEFAULT_0_0_0 (sets DEFAULT_ALGO)
    • ;P2KLPU MATERIAL_PETG_PLA_3_-1_-6 (applies to any transition where filament_type matches)
    • ;P2KLPU MATERIAL_DI1_DI2_3_-1_-6 (directly sets the algorithm for input transition DI1 → DI2)
  • ;P2KLPU SYNC_BEFORE_G4=1 (default is on)
  • ;P2KLPU G4_ZERO_TO_M400=1 (default is on)
  • ;P2KLPU REWRITE_M0_M1=1 (default is on)
  • ;P2KLPU DROP_M0_M1_AFTER_O1=1 (default is on)
  • ;P2KLPU SYNC_PING_MACRO_OVERRIDE=MyOwnMacro (replaces the ping-block sync barrier line)
  • ;P2KLPU PING_MACRO_BEFORE=PING_BEGIN
  • ;P2KLPU PING_MACRO_AFTER=PING_END
  • ;P2KLPU OCTOPRINT_STRIP_O_COMMANDS=1 (advanced; see OctoPrint section)
• Normalizes G4:
  • Rewrites G4 S<seconds> to G4 P<milliseconds>.
  • In Klipper mode, replaces G4 S0 / G4 P0 with M400 (configurable via G4_ZERO_TO_M400).
  • In Klipper mode, inserts M400 before non-zero G4 (configurable via SYNC_BEFORE_G4).
• Normalizes slicer pauses:
  • Rewrites M0/M1 to PAUSE (configurable via REWRITE_M0_M1).
  • Drops an M0/M1 immediately after an O1 ... line (configurable via DROP_M0_M1_AFTER_O1), because Klipper’s [palette2] O1 handler already pauses.
• If a ping block is detected (the file contains ; --- ... INSERT PING CODE ...), optional macros are inserted before the ping G4 and after the O31 line.
• In console analysis output, O31 pings are decoded to millimeters:
  • O31 Dxxxxxxxx is a legacy “hex float32” encoding of the ping position in mm (matches the common hexify_float behavior).
  • O31 L<mm> mm is a more human-readable form seen in some Palette 3 workflows.

RAW_MMU hardening / diagnostics:

• In RAW_MMU mode, the analysis prints both:
  • Total positive extrusion (all positive E, including toolchange prime/unload/reload), and
  • Effective positive extrusion (positive E excluding E-only toolchange logistics).
• When PrusaSlicer ;TYPE:... markers are present, the analysis also breaks effective extrusion into:
  • Tower effective extrusion (;TYPE:Wipe tower / ;TYPE:Prime tower), and
  • Model effective extrusion (everything else).
• The scanner prefers explicit toolchange markers when present:
  • ; CP TOOLCHANGE START/END and ; TOOLCHANGE START/END.
  • This makes it much more resilient to PrusaSlicer wipe tower sparse layers and other layout changes, because we don’t rely solely on a fixed “N lines after Tn” heuristic.

Build / run

• dotnet build
• Dry-run analysis:
  • dotnet run --project .\\P2KLPU.csproj --framework net10.0 -- input.gcode --dry-run

PrusaSlicer setup (post-processing)

PrusaSlicer runs post-processing scripts like:

<script> <input.gcode> <output.gcode>

So in PrusaSlicer you typically only enter the executable path; PrusaSlicer supplies the input/output file paths.

Windows examples:

• If you built the project and want to run via the .NET host:
  • dotnet "C:\\path\\to\\tool\\bin\\Release\\net10.0\\P2KLPU.dll"
• If you published a self-contained executable:
  • "C:\\path\\to\\your-tool.exe"

Where to put it in PrusaSlicer:

• Print Settings → Output options → Post-processing scripts

Notes:

• Quote paths with spaces.
• You do not need to add placeholders for input/output; PrusaSlicer appends them.

Do I need to configure anything else in PrusaSlicer?

Usually: no — just add the post-processing script.

You may need additional PrusaSlicer configuration depending on the workflow:

• RAW_MMU auto-enable: this POC can auto-enable RAW_MMU when PrusaSlicer footer contains single_extruder_multi_material = 1 and the file contains tool changes and does not already look Omega-processed.
• OctoPrint workflow (see below): to allow auto-detection of “printing via host”, PrusaSlicer must be configured to print to a host so it sets SLIC3R_PP_HOST.

Notes

• This is not a full-featured tool yet: it does not implement a full purge/tower rewrite pipeline.
• Today it’s most useful as a Klipper compatibility pass over connected-mode output (e.g., fixing G4 S0 and letting you hook macros around ping blocks).

Slicer feature edge cases (why Python cared more than this POC):

• Features like variable layer height and combine infill mainly matter when a processor is doing geometry- or layer-structured rewrite (e.g., purge tower geometry / per-layer heuristics), because those slicer options change layer structure and how extrusion is distributed across layers.
• This .NET POC’s RAW_MMU pipeline is primarily stream/timeline based: it plans splices and pings from toolchange events plus “effective positive extrusion” (E-distance excluding E-only toolchange logistics), and otherwise tries to keep print moves intact.
• Practically: variable layer height / combined infill may change where a ping falls along the E timeline, but they are not expected to break processing correctness the way they can for tower-geometry rewrite.

Klipper safety:

• Palette 2/2S “Omega” uses O.. commands (non-standard G-code). On Klipper, the built-in [palette2] module registers the O0..O32 commands directly (see klippy/extras/palette2.py), so a file like samples/output/example_processed.gcode can work without defining gcode_macro O21, etc.

Firmware flavor:

• PrusaSlicer writes ; gcode_flavor = ... into the generated G-code (typically in the config footer).
  • If the file is marked as klipper, the POC enables Klipper-specific pause/sync fixes.
  • If the file is marked as a Marlin flavor, the POC is pass-through by default (no G4 or M0/M1 rewrites), but explicit ping-block overrides (e.g. ;P2KLPU SYNC_PING_MACRO_OVERRIDE=..., ;P2KLPU PING_MACRO_BEFORE=..., ;P2KLPU PING_MACRO_AFTER=...) are still honored.

Custom purge tower (TOWER mode)

P2KLPU can generate its OWN purge tower and ignore PrusaSlicer's entirely. In this mode PrusaSlicer only decides when toolchanges happen; P2KLPU owns all purge — which makes minimum splice lengths guaranteed by construction (each transition's purge is sized as max(pair purge, MINSPLICE deficit)).

Setup:

1. In PrusaSlicer, disable the wipe tower (Print Settings → Multiple extruders → Wipe tower OFF). TOWER mode errors if the export still contains the slicer's tower (you'd get two towers). Purging into infill or objects is done by P2KLPU itself (see below), no slicer tower needed. If you prefer PrusaSlicer's own wipe-into-infill planning, keep its tower on and add ;P2KLPU TOWER_REPLACE_SLICER_TOWER=1: P2KLPU then removes the slicer's tower from the file (its purge lines, perimeters, sparse grids, the travels to it and its pressure-advance disable) and honors what PrusaSlicer wiped into the model.
2. Enable Label objects (Output options) — recommended, so the tower can be auto-placed near your objects using their real outlines.
3. Add ;P2KLPU TOWER=1 to your start G-code directives.

How the tower behaves:

• Grows layer-synchronized with the print (PrusaSlicer-style) and stops at the last toolchange: every layer up to and including the last color change's layer gets a tower pass — "sustaining" layers (walls + a sparse support lattice) on layers without a toolchange, perimeter + zigzag fill consumed up to the purge target on layers with one (full if needed). Nothing is printed above the last color change. The tower top is therefore always at the nozzle's current Z, so tower visits never move the toolhead down (only the usual hop); no P2PP-style lagging Z. Fill orientation alternates 90° per layer for anchoring.
• Sustaining passes are the price of that: a print whose color changes all sit high up (a two-color Benchy with the changes in the roof) spends a pass on every layer below them. The support lattice is printed only where something above lands on it: every layer that has a purge somewhere above it keeps the lattice (so the whole tower below the last color change is latticed, and only the top purge layer's unused remainder is left bare). TOWER_SUSTAIN_LATTICE_LAYERS=<n> restricts the lattice to the n sustaining layers directly below each purge layer and prints walls only further down — cheaper (n=1 leaves just the walls on a long single-color stretch, TOWER_SUSTAIN_PERIMETERS=1 halves that again), but the lattice under a purge then bridges the hollow tower wall to wall, which PETG or a wide tower may sag on; n=2 gives it a lattice of its own to rest on.
• No mid-air purge: the lattice under a purge layer (TOWER_SUSTAIN_SPACING, default 6mm; 0 = walls only) uses the same orientation parity as the dense fill — so the purge fill always finds perpendicular support lines one layer below and bridges at most one lattice gap, never the whole footprint.
• Adaptive lattice (the tower's adaptive cubic, on by default): the surface that needs support is the next purge layer above, so the lattice keeps the dense spacing within TOWER_SUSTAIN_DENSE mm of height below it (default = the spacing, 6 mm), doubles to 12 mm for the zone below that (12 mm tall), and 24 mm beyond (TOWER_SUSTAIN_SPACING_MAX, default 4× the spacing). Deep inside the tower a lattice line bridges up to 24 mm between the perpendicular lines under it, which is harmless there. TOWER_SUSTAIN_ADAPTIVE=0 keeps one spacing everywhere.
• First layer is printed full (dense fill) with a brim around it (TOWER_BRIM_LOOPS, default 4) for adhesion.
• The color change happens ON the tower: the purge visit replaces the toolchange immediately, and every purge is floored at SPLICEOFFSET + 15mm, so the splice junction (where the blend arrives, per your SPLICEOFFSET calibration) always lands inside the tower purge with margin after it.
• Footprint is auto-sized and auto-shaped from the worst layer's purge demand: a square is preferred, but when it doesn't fit the free bed space the planner tries rectangles (1.5:1, 2:1, 3:1, both orientations — e.g. a long thin tower in a strip beside the model), and when no single footprint fits, it splits the purge across two or three smaller towers. Placement hugs object edges exactly, prefers spots closest to your objects, and validates against the bed and object outlines (TOWER_WIDTH/TOWER_DEPTH and TOWER_X/TOWER_Y still override; explicit dimensions force a single tower).
• With multiple towers, purge is routed per layer across them (one oversized purge can span towers within a single visit, with a retracted hop between), every tower gets coverage on every layer (purge fill, lattice, or a sustaining pass with walls), and each tower is its own Klipper object (P2KLPU_Tower, P2KLPU_Tower2, ...).
• On Klipper, the tower is a real Klipper object: EXCLUDE_OBJECT_DEFINE NAME=P2KLPU_Tower plus EXCLUDE_OBJECT_START/END around every tower pass — it appears in Mainsail/Fluidd's bed map and object list, and shows as the currently printing object during purges. Do NOT cancel it: cancelling skips the purge extrusion and desyncs the Palette splice schedule (only the printing moves are wrapped, so printer position/retract state survive a cancel — the print's colors will not).
• The analysis (console + G-code footer) reports how much of each filament the tower consumes (purge attributed to the incoming filament, sustaining passes to the loaded one), in mm and cm³ per input.
• Purge lengths (filament mm): PURGE_DEFAULT=105 (≈250mm³) and per-pair PURGE_PETG_PLA=120 / PURGE_DI1_DI2=90 overrides (input pair > material pair > default).
• Sustaining passes use the currently loaded filament and also lengthen the current splice segment (free min-splice help).

Speeds: by default the tower prints at the sliced profile's own speeds — normal layers at the solid-infill speed (falling back to infill, then perimeter speed), the first layer and the calibration squares at first_layer_speed (percentages resolved the way PrusaSlicer does), and the flow cap taken from the smallest max_volumetric_speed / filament_max_volumetric_speed in the profile (none set = uncapped, like the rest of the print). Give TOWER_SPEED, TOWER_FIRST_LAYER_SPEED or TOWER_MAX_FLOW a number to override any one of them; PRINT on any of them selects the profile value explicitly. The console reports what was taken from the profile.

TOWER directives: TOWER=0|1, TOWER_X, TOWER_Y, TOWER_WIDTH, TOWER_DEPTH, TOWER_BRIM_LOOPS=4, TOWER_SPEED=<mm/min> (default: profile), TOWER_FIRST_LAYER_SPEED=<mm/min> (default: profile), TOWER_SUSTAIN_PERIMETERS=2, TOWER_SUSTAIN_SPACING=6, TOWER_SUSTAIN_LATTICE_LAYERS (default: all layers below a purge), TOWER_SUSTAIN_ADAPTIVE=1, TOWER_SUSTAIN_DENSE=6, TOWER_SUSTAIN_SPACING_MAX=24, TOWER_MAX_FLOW=<mm³/s> (default: profile cap), TOWER_SPLICE_DWELL=0 (ms), TOWER_EXTRUSION_WIDTH (default: footer extrusion_width, else 0.45), PURGE_DEFAULT, PURGE_<FROM>_<TO>, PURGE_INTO_INFILL=0|1, PURGE_INTO_OBJECT=<name>, PURGE_JUNCTION=MODEL|TOWER (default MODEL), TOWER_REPLACE_SLICER_TOWER=0|1.

Purging into the model: wipe into infill / wipe into object

Internal infill is invisible, so it can be printed in whatever comes out of the nozzle right after a color change. ;P2KLPU PURGE_INTO_INFILL=1 makes P2KLPU do that itself, with no slicer tower involved: for every toolchange it takes the ;TYPE:Internal infill blocks that PrusaSlicer would print later on the same layer (after the change, before the next one), in stream order, until their filament covers the transition's pair purge, and prints them inside the purge visit instead — travel there, prime, the block's own lines verbatim (same coordinates and speeds, the object's Klipper label re-applied so cancelling the object still works), retract, back to where the toolchange left the head. Their sliced position and the travel run that led into them are dropped (the block's M204 is replayed in the visit; the retract/unretract lines around the hole are kept or dropped so the extruder state in the file stays balanced). Blocks under 0.5 mm of filament stay where they are: a visit's travel, prime and retract cost more than they purge. Perimeters, solid and top infill are never touched. ;P2KLPU PURGE_INTO_OBJECT=<Klipper object name> (repeatable, comma-separated) does the same with everything of a sacrificial object — its colors will mix.

The tower then only takes what the model cannot: tower purge = max(pair purge − into model, MINSPLICE floor, junction floor). The junction floor is SPLICEOFFSET + 15 mm of purge span in front of the declared color change; by default (PURGE_JUNCTION=MODEL) the relocated infill/object counts toward it, so a transition whose model purge covers the whole tail needs no tower visit, and when that holds for every transition no tower is generated at all — the model is the purge system. PURGE_JUNCTION=TOWER keeps those SPLICEOFFSET + 15 mm on the tower instead (the color change stays visible on the tower, the model carries the tail). The footprint is sized from the remaining per-layer demand, so the more the model takes, the smaller the tower. The console and footer report the transitions that purge into the model, the mm they take, and the tower purge saved.

Alternative, ;P2KLPU TOWER_REPLACE_SLICER_TOWER=1: let PrusaSlicer plan the wiping ("Wipe into this object's infill" / "Wipe into this object" in the object's Wipe options; both exist only while the slicer's wipe tower is enabled, so keep it on). P2KLPU removes the slicer's tower from the file, measures what PrusaSlicer printed into the model after each toolchange (the extrusion up to its ; PURGING FINISHED marker) and applies the same purge math. Set PrusaSlicer's Purging volumes at least as large as your P2KLPU purges in mm³ (mm × 2.405: PURGE_DEFAULT=120 → 290 mm³, 180 → 435 mm³) so it assigns as much to the model as it can, and filament_minimal_purge_on_wipe_tower to 0. P2KLPU's tower goes where the slicer's stood if it fits. Side effect worth knowing: PrusaSlicer disables Klipper pressure advance for its tower (SET_PRESSURE_ADVANCE ADVANCE=0 at every toolchange) and never restores it; those lines go with the tower, so the model keeps its pressure advance.

SPLICEOFFSET calibration print

Instead of tuning SPLICEOFFSET print by print, let P2KLPU generate a one-shot calibration print:

1. Slice any small object with your normal PrusaSlicer profile (two or more filaments assigned, wipe tower off or on — the model is discarded anyway).
2. Add ;P2KLPU CALIBRATE_OFFSET=20,20,8 to the directives (start, step, count; optional 4th value = test input, e.g. 20,20,8,2, or ALL to cycle through every filament in the profile — DI1→DI2→DI3→DI4→DI1… — so each color pair's purge tail gets measured, not just the offset). Keep your other directives (MINSPLICE etc.). Optional: ;P2KLPU CALIBRATE_SCALE=1 adds a direct-reading scale beside every square (see below); it is off by default because the thin strokes are tedious to remove from the bed.
3. Export. P2KLPU replaces the sliced model with a grid of pads but keeps your start/end G-code, temperatures, and profile.

What prints: a grid of single-layer squares directly on the bed, each with its declared offset printed as raised digits in front of it (e.g. 57.5). The priming square, marked P, is printed first in the start color (it keeps the first splice long enough). For every following square, the frame (brim + wall loops) is printed first in the outgoing color, then the toolchange declared with that square's own offset (square 1 = 20 mm, square 2 = 40 mm, …), then the purge = the fill inside the frame, alternating start→test color and back. A dense fill is a filament-mm ruler (the legend tells you the mm per fill line), so the fill line where the color flips, counted from the square's front edge, reads the junction arrival directly — the frame around it is just a frame.

Reading it — any square where the color change is visible works, and all of them should agree. The number printed in front of a square is the offset that square was printed with, not the answer; the answer comes from where the color flips inside the fill:

1. Count old-color fill lines from the square's front edge up to the first line showing any trace of the new color: measured = lines × mm-per-line. Do not count to where the color is clean — that later point is the purge tail, often 60–100 mm further, and using it makes the offset far too small (the color then changes too early in real prints).
2. SPLICEOFFSET = square's declared offset − measured mm (+10 mm safety so the change lands just inside the purge, never before it).
3. Count fill lines from the first trace until the color is fully clean: that tail in mm is your melt-zone purge requirement — set PURGE_DEFAULT ≥ tail + 30.

Cross-check without counting: squares whose declared offset is below the true value show the new color already in their frame (the change came before the purge started); the smallest declared offset whose frame is still entirely the old color is within one step of the answer.

The squares print at the profile's first_layer_speed and flow cap by default, like any first layer of yours. Keep it that way: the junction arrives earlier the more the Palette's buffer drains around a splice, and the buffer drains with consumption, so squares printed far slower than your prints would read a lower offset than the prints need. TOWER_FIRST_LAYER_SPEED and TOWER_MAX_FLOW override it if you want to test a specific speed.

A square that is entirely the new color changed before it started (that offset is too small); one that stays entirely old never received the change (too large) — read a neighbor. Because the formula uses the measured position, a coarse sweep (e.g. 40,10,12) already gives mm-level results; a fine step just adds redundancy. One print, both numbers.

With ;P2KLPU CALIBRATE_SCALE=1 a scale is printed on the left of every square, counting down from declared + 10 at the front edge (a tick every 10 mm of filament, a number every 30 mm), so the number beside the color change is your SPLICEOFFSET with no counting. The columns are spaced wider to make room for it.

Adaptive meshing (KAMP etc.): the sliced dummy model's EXCLUDE_OBJECT_DEFINE is replaced by one covering the whole calibration grid, so object-aware start macros mesh the squares — not the dummy.

Palette buffer error 121 and tower pacing

There are no O-codes that feed filament — Klipper's [palette2] module is passive (it relays the header and answers the device), and the Palette advances filament purely from its buffer switches. While the Palette is making a splice it cannot feed at all, so the buffer must cover everything the printer consumes in that window. A print with many splices near MINSPLICE keeps the Palette splicing almost continuously; purging fast through those windows drains the buffer a little more each segment until the collapsed switch trips → error 121 (and the Palette can't advance right then because the filament is clamped in the splice core).

By default the tower runs at the profile's speeds and flow cap, no slower than the rest of the print. TOWER_MAX_FLOW=<mm³/s> can cap tower extrusion volumetrically regardless of layer height if the buffer collapses during splices (error 120): a cap around 1.8 mm³/s keeps buffer draw during a splice well inside what the Palette recovers between splices, and TOWER_SPLICE_DWELL=3000 pauses 3 s at the start of every purge visit so the Palette gets a head start. Do not reach for the cap on a 121 that comes with an idle Palette — that is the device stalling in its own splice sequence, not the buffer being outrun. The analysis warns you when most transitions sit at the MINSPLICE floor (continuous-splicing territory). Raising MINSPLICE/purge lengths also lowers splice cadence at the cost of more purge.

Not in v1: ramming, multiple towers, non-rectangular bed geometry (bounding box only), configurable z-hop (fixed +0.6mm).

Supported ;P2KLPU directives

Directives are case-insensitive and can appear anywhere in the file.

General:

• ;P2KLPU RAW_MMU=0|1
• ;P2KLPU STRICT=0|1 (default 1: error-level findings — short splices, absolute E in RAW_MMU, MMU priming enabled — fail the export with a non-zero exit code so PrusaSlicer shows them; set 0 to write output anyway)
• ;P2KLPU PRINTERPROFILE=<hex> (Palette2 printer profile ID; omit to auto-derive a stable ID from the slicer printer profile name)
• ;P2KLPU AUTOLOADINGOFFSET=<mm> (see note below)
• ;P2KLPU FILAMENTOVERRIDE_DI<n>=<name> (overrides PrusaSlicer filament_type[n-1] for MATERIAL matching)
• ;P2KLPU FILAMENTOVERRIDE=<name> (alias for FILAMENTOVERRIDE_DI1)
• ;P2KLPU EXTRAENDFILAMENT=<mm>
• ;P2KLPU MINSTARTSPLICE=<mm>
• ;P2KLPU MINSPLICE=<mm>
• ;P2KLPU SPLICEOFFSET=<mm>
• ;P2KLPU SPLICE_OFFSET=<mm> (alias of SPLICEOFFSET)

About FILAMENTOVERRIDE:

• This changes the material name used for MATERIAL_<FROM>_<TO>_h_c_k matching and for Omega’s material table.
• Use it to introduce custom names like PETG-MATTE / PETG2 even if PrusaSlicer’s filament_type is more generic.

Material aliases (Spoolman-style, recommended):

• You can attach a stable material token to the filament profile itself, so it follows whichever tool/extruder it is assigned to.
• Add p2klpu_material per-filament in PrusaSlicer metadata; the tool reads (first found wins): custom_parameters_filament, filament_custom_variables, filament_notes.
• Supported formats per filament entry:
  • JSON object (common in custom_parameters_filament): {"p2klpu_material":"PETG-MATTE"}
  • Key/value text (notes/custom variables): p2klpu_material=PETG-MATTE

About AUTOLOADINGOFFSET:

• In connected mode, Palette schedules splices/pings in terms of “mm of filament fed”.
• Some setups effectively have a fixed offset between what the Palette counts and what the printer has already consumed when printing starts (autoload / preloaded length).
• The processor uses this offset to shift Omega distances (notably O30 splice positions, O31 ping positions, and the total in O1) by the specified millimeters.

About MINSTARTSPLICE / MINSPLICE:

• These set minimum splice-length thresholds.
• Your values are honored exactly (no silent clamping); values below the Palette 2 manual minimums (85mm first / 60mm rest) additionally produce a warning.
• A computed splice below the minimum is an ERROR: with STRICT=1 (default) the export fails so the slicer surfaces it; with STRICT=0 it is reported and the output is still written.
• The check covers every splice including the final end-of-print splice.

About EXTRAENDFILAMENT:

• This adds extra “tail” filament to the final end-of-print splice (and therefore the Omega O1 total length) so there is additional filament available after printing finishes.

About the splice schedule (important):

• The splice list always ends with a final end-of-print splice covering the last tool's segment through the end of the print plus EXTRAENDFILAMENT. Without it the Palette would never schedule production of the last color segment. The O1 total equals the end of that final splice, matching what the Palette expects.
• SPLICEOFFSET shifts that end like every other junction, so the filament that actually arrives after the last extrusion is EXTRAENDFILAMENT + SPLICEOFFSET, and it must cover the extruder-to-nozzle path (~100 mm on a direct-drive toolhead) because the extruder can only push while it still grips the filament. A SPLICEOFFSET set too low therefore does two things at once: colors change too early and the last part of the print runs dry (a Benchy printed at −25 with the true value near 60 lost its chimney top). The analysis warns when the tail comes out under 150 mm; 150 + a correct offset is enough, 250 gives margin for an offset that is still being tuned.
• Extrusion is accounted NET (retracts subtract, unretracts add back), matching what the Palette's encoder physically sees — so ping feedback percentages stay close to 100%.
• Arc moves (G2/G3) are fully supported in the accounting, so PrusaSlicer “Arc fitting” can stay enabled.
• A P2PP-style splice/ping summary is appended to the output G-code as comments, so you can inspect the plan after the fact even though PrusaSlicer hides console output.

About algorithm overrides and material IDs:

• The Palette 2 selects splice parameters from the O32 table keyed by MATERIAL-ID pairs (from O25), not per splice. When you use input-pair overrides (MATERIAL_DI1_DI2_... / MATERIAL_IN1_IN3_... / ALGO 1-2=...), P2KLPU assigns each used input its OWN material ID so those per-input algorithms genuinely reach the device (with only material-name overrides, inputs sharing a material share an ID, exactly like P2PP).

Ping planning:

• ;P2KLPU PING_INTERVAL=<mm>
• ;P2KLPU PING_MAX_INTERVAL=<mm>
• ;P2KLPU PING_LENGTH_MULTIPLIER=<float>

RAW_MMU toolchange stripping heuristics:

• ;P2KLPU MMU_TOOLCHANGE_WINDOW_LINES=<int>
• ;P2KLPU MMU_E_ONLY_STRIP_THRESHOLD=<mm> (default 15: E-only moves inside a toolchange region are stripped only when their magnitude is at least this many mm — big unload/load/ram moves go away, small retract/unretract pairs survive so the toolchange wipe still controls ooze; set 0 to strip every in-window E-only move)

Algorithm selection:

• ;P2KLPU DEFAULT_ALGO=h,c,k
• ;P2KLPU ALGO 1-2=h,c,k (accepts = or : between the key/value)
• ;P2KLPU MATERIAL_DEFAULT_h_c_k
• ;P2KLPU MATERIAL_<FROM>_<TO>_h_c_k where <FROM>/<TO> are either material names (from PrusaSlicer filament_type) or DI1..DI4.

Ping-block macro hooks (applies when a ping block is present or inserted):

• ;P2KLPU PING_MACRO_BEFORE=<gcode>
• ;P2KLPU PING_MACRO_AFTER=<gcode>
• ;P2KLPU PING_MACRO=<gcode> (sets both before and after)
• ;P2KLPU SYNC_PING_MACRO_OVERRIDE=<gcode> (replaces the ping-block sync line when it is a zero-length dwell)

Klipper-oriented normalization:

• ;P2KLPU SYNC_BEFORE_G4=0|1
• ;P2KLPU G4_ZERO_TO_M400=0|1
• ;P2KLPU REWRITE_M0_M1=0|1
• ;P2KLPU DROP_M0_M1_AFTER_O1=0|1

Spoolman integration:

• ;P2KLPU SPOOLMAN_SET_ACTIVE_SPOOL=0|1
  • When enabled, the tool looks for per-filament spool IDs in PrusaSlicer metadata (custom_parameters_filament, filament_custom_variables, or filament_notes).
  • Supported key names inside those fields: spoolman_id, spool_id, target_spool.
  • When a tool change happens, it emits SET_ACTIVE_SPOOL ID=<n> (Klipper macro) when an ID is available for that tool.

OctoPrint/Marlin compatibility:

• ;P2KLPU OCTOPRINT_STRIP_O_COMMANDS=0|1
  • When enabled, the processor rewrites Omega O* commands into comment markers (;P2KLPU_OCTO O31 ...) so Marlin never sees unknown O* commands.
  • This is intended for Marlin workflows without an OctoPrint Palette2 plugin.
  • If the processor detects host printing via PrusaSlicer (SLIC3R_PP_HOST) and the file is Marlin flavor, it assumes OctoPrint is in the loop and automatically disables this option to remain compatible with the OctoPrint Palette2 plugin (which requires real O* lines).

OctoPrint (Marlin) connected-mode notes

The community OctoPrint Palette2 plugin intercepts Omega commands (O21, O1, O31, etc.) directly from the outgoing G-code stream.

Practical implications:

• For OctoPrint + Palette2 plugin workflows, the output must contain real O* lines (not commented out).
• The plugin typically suppresses O* lines so they are not sent to the printer, and replaces pings (O31) with a short dwell.

Auto-detection in this POC:

• If the slicer config indicates Marlin and PrusaSlicer provides SLIC3R_PP_HOST, the POC assumes printing via host (OctoPrint) and keeps Omega O* commands intact.