Abdallah El BelkasyA. El Belkasy
Dubai --:-- 120 BPM Open to Rust · systems · backend

Off the bed

3D prints · No. 01

A Panamera in 44 slices

Eight versions drawn in code, a print that stopped after 22 minutes, and a first layer in a knot. The cause was under the base.

  • PLA
  • 0.12 mm layers
  • 7 h 37 min printing
  • 192 g
  • 2 prints, 1 stopped
The finished sculpture on dark granite, seen from the front and above: 44 black slices form the car on a red octagonal plaque with a thin black border, PANAMERA GTS in wide capitals below the car, the Nordschleife outline at the left and a pixelated patch at the right.
The finished piece. The track outline is the Nordschleife, traced from OpenStreetMap data (© OpenStreetMap contributors). A name on the plaque and the plate on the car's tail are pixelated in every picture.

At 22:53 on a Saturday night the printer stopped itself, 22 minutes into a print meant to run all night. Through its camera I could see a knot of red plastic at one corner of the plaque and loose strands lying across the lettering engraved under the base.

Phone photo of the red octagonal plaque's first layer on the black bed. A tangle of red filament sits at the far corner and loose strands lie across the mirrored letters engraved in the surface.
Fig. 01 · Attempt 1, two minutes after the stop, through the printer's door.

That was version 8 of a desk sculpture: a Porsche Panamera GTS cut across into thin black plates, each one the shape of the car at one point along its length. The plates stand in a row on a red plaque with gaps between them. Look from the side and the gaps close up and the car comes back. I'd seen synysthh's sliced Panamera GTS on Cults3D and wanted one of my own, of the first-generation car, the 970.

The whole thing is one script in CadQuery, a Python library for CAD. It makes 44 slices, each 1.6 mm thick, plus the octagonal plaque, the lettering, the border and a short message engraved under the base: From Abdallah, 2026. The printer has one nozzle, so the second colour comes from a pause. The red plaque prints up to 6.0 mm, the printer pauses before the first black layer at 6.12 mm, I swap to black, and it carries on with the car and the letters.

Render of the design from straight above: a red octagonal plaque with a thin dark border, the 44 black slices of the car in the upper half, PANAMERA GTS in wide capitals below, the Nordschleife outline at the left and a pixelated patch at the right.
Fig. 02 · What the script makes, seen from above. Everything red prints first, then the pause, then everything black.

Not even close

The first four versions were sliced from lnbjfu's print-in-place Panamera, a car model on MakerWorld. They sliced cleanly and looked like a generic fastback. My verdict on them was three words: not even close.

Version 1 · from the mesh

Pale blue render of the first version: a rounded, featureless car body cut into slices on a plain rectangular base.

Version 8 · from the car's numbers

Render of the final version from the side: the black slices close up into a long low car with a sloping roofline, on the red octagonal plaque.
Fig. 03 · Version 1 next to version 8, both drawn in CadQuery. The first is a loaf. The last one is a 970 from the side.

So version 5 threw the mesh away and started from the real car's numbers, scaled from its 4,970 mm length down to 140 mm. The roofline was traced from a side photo of a 970, the widths came from a published drawing, and every slice was rebuilt from those. Then a script laid the model's side silhouette over the drawing, and nothing moved on until the two lined up. That silhouette never changed again.

Making it print

The shape was right, but the first time I sliced it with supports allowed, the slicer wanted 44,275 support moves. Almost all of them were inside the wheel arches. Each arch was a groove with a flat roof 7.3 mm wide, in every slice. The support under those roofs would have sat in gaps about 1.6 mm wide, and no tool gets in there to take it out.

So every slice outline is now limited to how far a bare wall can lean out without support, every raised detail got a chamfer underneath, and the arch grooves got pointed roofs instead of flat ones. I tried making them shallower too, but at 1.7 mm deep the arches vanished from the renders, so they stayed 2.8 mm deep.

Before · version 6

Top view of the car before the fix: dense support under all four wheel arches, and small patches along the middle under the lettering on the base.

After · version 8

Top view of the car after the fix: the same 44 slices and no support anywhere.
Fig. 04 · Support the slicer asked for, seen from above and drawn from the G-code. The four dark blocks are the wheel arches. The small marks along the middle sit under the message engraved beneath the base, which got its own roof in the same fix.

But every slice has a seam, the small bump where each layer's outer wall starts and ends. I had rotated the model in the file so the slicer would put its seams on the hidden side. When the plaque became an octagon, the slicer's auto-arrange rotated it again, and the seams landed on the front. Now the model goes in unrotated, and before every print a script plots each seam from the G-code.

Back

Top view of the final file 22 mm up, title side at the bottom. Every slice's seam dot sits on its back end, at the top. A cluster of dots on the right end belongs to the raised panel on the tail.

Front · title side

Fig. 05 · The final file 22 mm up, from above. Each dot is a seam, drawn from the G-code, and they all sit along the back. The cluster on the right is the raised number plate on the tail.

Then the preview showed the plaque's raised letters, which have strokes 1 mm wide, printing as single beads 0.29 mm wide. They would have looked scratched in, not raised. The cause was a setting called Precise outer wall, which pulls the outer wall inward so outside sizes come out exact. On a stroke that thin it leaves room for one bead. Switching it off brought the letters back to full width.

The plaque, meanwhile, had been a plain rectangle all along, and a plain rectangle is what it looked like. I drew six shapes for it in one go: a stadium, a motorsport plaque (an octagon with a raised border), a road section with kerbs, a wedge, a two-tier step and a superellipse. The wedge and the step broke the single colour change, so they were out first. The octagon won.

Six renders of the same black sliced car on six red plaque shapes: a stadium, an octagon with a dark border line, a strip with dashed kerb blocks along its front edge, a plain wedge, a stepped two-tier base and a superellipse. A small patch on each plaque is pixelated.
Fig. 06 · The six plaque shapes, drawn in CadQuery, each with its name under it. The motorsport plaque won.

By then the shape matched the drawing, the supports were gone, the seams were on the back and the letters were full width. Attempt 1 started at 22:31. Every check I knew how to run said yes.

Back to 22:53

The usual suspects for a print that stops itself were a pause in the file or the filament sensors, which had paused a print on this machine before. Neither. The printer's log had the answer: Obico, an add-on that watches the printer's camera for failed prints, had seen the knot and paused it.

Layer 1 had come apart around the message engraved under the base. It prints as gaps in the very first layer, and I had never looked at what the insides of the letters became. The centres of the o, A, b, d, a, a, 0 and 6 in From Abdallah, 2026 were islands: eight separate pieces on layer 1, each 6 to 26 mm², joined to nothing. They lifted, and the nozzle dragged them around. A 0.12 mm first layer on a bed that is 0.27 mm out of flat had not helped.

Attempt 1

First layer under the base, the date 2026 seen from below, attempt 1. The centres of the 0 and the 6 are solid islands floating inside their letters, joined to nothing.

Attempt 2

The same patch on attempt 2. The 0 and the 6 have no floating centres any more: each centre is joined to the base by a thin bridge, so their gaps are open C shapes.
Fig. 07 · Layer 1 under the message, seen from below, drawn from each file's G-code. Filled: a letter centre held by nothing. The 2s have no centre, so they look the same both times. On attempt 2 the 0 and the 6 are open C shapes: the centre is part of the base.

The fix was a stencil trick, the little bridges that hold the middle of an O in place on a stencil. Each letter centre got a 0.9 mm bridge to the rest of the base, and layer 1 went from nine separate pieces, the base and its eight islands, to one. The letters still read fine.

The first layer itself also got thicker, 0.24 mm instead of 0.12, and slower, to ride over that high corner of the bed.

Fig. 08 · The file itself, drawn from the G-code one layer at a time: 382 layers in 8 seconds, every path the nozzle takes. Red for the first 49 layers, black from layer 50 at 6.12 mm.

Two and a half hours after the stop, with the bridges in the file, attempt 2 started at 01:23. At 03:42 it stopped where it was meant to, with the red plaque finished and flat.

Printer camera frame of the finished red plaque on the bed, flat and clean, at the colour-change pause.
Fig. 09 · The planned pause, 2 h 19 min in. The printer sits here with the nozzle parked until the black is loaded and I press resume.

Eight minutes later the black was loaded and the slices started going up.

Fig. 10 · Seven hours and thirty-seven minutes in eight seconds, from the first red layer to the last black slice at 09:24. The same camera, every twelfth frame.

At 09:24 it was done. The title, the track outline and the border came out crisp.

The front edge of the finished sculpture, close up: PANAMERA GTS in wide black capitals on the red plaque, the Nordschleife outline to the left, a pixelated patch to the right, the thin black border line, and the slices of the car above.
Fig. 11 · The front edge, the morning it came off the bed. The letters are full width.

It came off the bed in one piece, with nothing to cut away. The number plate on the tail, pixelated in these photos, came out a little soft. Everything else was sharp.

From the back

The sculpture from behind on a white shelf: the black slices seen from the car's tail, the red plaque with its border, V8 430 HP engraved in the back edge, the Nordschleife outline at the far corner, and a small pixelated patch between the slices.

The tail

Close-up of the car's tail end on the red plaque. The slices step down towards the rear, and a small raised panel on the last slice is pixelated.
Fig. 12 · Left: from the back. The spec line in the back edge, V8 and 430 HP, is the real car's. Right: the tail, close up. Each slice is 1.6 mm thick with about the same gap, which is why the car disappears from behind.
The sculpture exactly side-on at the car's height on a white shelf against a wood wall: the black slices close up into the profile of a long low car with a sloping roof, standing on the red plaque with PANAMERA GTS below it.
Fig. 13 · The view it is built for, side-on at the car's height. Stand up and the slices separate again. The right edge of the plaque is cropped off: a name is there.

What I check now

Before anything with text underneath goes to the printer, I count the islands on layer 1. It takes a minute, and it would have saved the first attempt.

The short version

What went wrong, and what fixed it.

  1. The first four versions didn't look like the car.Rebuilt every slice from the real 970's dimensions and checked the silhouette against a drawing.
  2. 44,275 support moves inside the wheel arches.No slice leans out more than 43° from vertical, 45° roofs on the arch grooves.
  3. The seams landed on the side you look at.Stopped pre-rotating the model, and plot the seams from the G-code before printing.
  4. Letter strokes printing as single 0.29 mm beads.Precise outer wall off.
  5. Loose letter centres under the base tore off on layer 1.0.9 mm stencil bridges, so layer 1 is one piece.
  6. A 0.12 mm first layer on a bed with 0.27 mm of spread.0.24 mm first layer at 30 mm/s.

Built on

Where the ideas came from. Their files stay on their pages.

How it was printed

Printer
Creality K1 Max, rooted, Klipper
Slicer
OrcaSlicer 2.4.2, profile K1M 0.12 Fine
Nozzle
0.4 mm, 210 °C (215 °C first layer)
Bed
60 °C
Material
PLA, red for the plaque, black for the car and lettering
Layers
0.12 mm, first layer 0.24 mm
Walls, infill
3 walls, 15 % crosshatch
Size
200 × 110 × 46 mm, 44 slices 1.6 mm thick
Supports
None
Colour change
Pause at 6.12 mm, red out, black in, resume
Time
7 h 37 min printing, 8 h 2 min with heat-up and the swap
Plastic
192 g

Settings that mattered

The profile’s value, then what this print used.

  • First layer height 0.12 changed to 0.24 mm
  • First layer speed 60 changed to 30 mm/s
  • First layer infill speed 105 changed to 40 mm/s
  • Precise outer wall on changed to off

PANAMERA and GTS are Porsche trademarks. This is a fan piece, not a Porsche product.

The files

The sculpture as one STL, a preview and a README with the settings. The name on the plaque and the number plate are left out of the file.

License: CC BY-NC-SA 4.0. Credit: Abdallah El Belkasy, abdaishere.com/prints. The track outline is © OpenStreetMap contributors (ODbL), as the README says.

Off the bed