PrusaSlicer 2.9.4 with built-in filament calibration  

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Are the changes vast? ie, would porting the calibration tests over to PS3 be a hellish endeavour ?

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Posted by: @gb160

Are the changes vast? ie, would porting the calibration tests over to PS3 be a hellish endeavour ?

Until Prusa release the source code for it, no clue. As soon as the beta is released I'll have a look. 

RE: PrusaSlicer 2.9.4 with built-in filament calibration

First, a thanks to @hyiger for this as without it I probably never would have done any calibration at all! I can see why so many people can ignore it with PLA and PETG. They work well over a huge range of temperatures and other settings, and it can be difficult to narrow things down.

I did have trouble with the pressure advance calibration as no setting was any better than any other. It turns out you need to print fairly fast before PA matters. The trick is to go into filament-cooling and change the maximum speed for layers that complete quickly (typ. < 8 seconds). Crank it up to 120 mm/s or so and the double check the speed after slicing in the analysis window (or whatever the official name is). Once I did that the best setting became obvious. The default was too slow.

I don't have retraction sorted yet, but I suspect the Nextruder is so accurate that almost no retraction is needed. 

RE:

Hi @Conrad, good catch. I've added a parameter to set the speed for this test. Please try: https://github.com/hyiger/PrusaSlicer/releases/tag/v1.5.2-filament-edition

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Yes @hyiger, thanks, that's much better. I'm now moving on to the retraction test but with similar results to when I've tried previously. Fine stringing, maybe damp filament, but the retraction setting isn't having any influence at all. I'll keep investigating.

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Pretty slick - thanks.  

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Posted by: @conrad-2

Yes @hyiger, thanks, that's much better. I'm now moving on to the retraction test but with similar results to when I've tried previously. Fine stringing, maybe damp filament, but the retraction setting isn't having any influence at all. I'll keep investigating.

@Conrad. Thanks for testing. Any bugs or enhancement requests, please add them here: https://github.com/hyiger/PrusaSlicer/issues

RE:

Here's a feature list for the PrusaSlicer Filaments Edition fork (everything new vs. upstream PrusaSlicer 2.9.4):

Calibration menu — built-in pattern generators with auto-applied per-layer G-code:

• Temperature Tower (multi-tier with overhangs, holes, cones, labels)
• Flow Rate (YOLO 11-pad layout with Archimedean spiral)
• Pressure Advance (chevron tower; auto-picks M572 S / M900 K / SET_PRESSURE_ADVANCE for your firmware)
• Retraction tower
• Max Flow Rate (serpentine E-shape with per-layer speed override)
• Extrusion Multiplier (vase-mode cube)
• Fan Speed tower
• Dimensional Accuracy / Shrinkage gauge
• Bundled Calibration Guide linked from the menu

Bed Mesh Visualization — 3D heatmap rendered on the build plate:

• One-click Fetch Bed Mesh over USB serial from any Prusa printer (M420 V1 T1)
• Probe Bed Mesh… dialog with nozzle + bed heating, M73 progress, force-stop, and per-tool probing on the XL
• Save / Load / Compare with CSV — diff a fresh mesh against a saved baseline to see drift
• Diverging color ramp, iso-Z contour lines, per-cell value labels, Z exaggeration slider
• Warp report with least-squares plane fit, RMS, worst-point deviation, editable quality grade
• Per-printer auto-detection of expected probe counts (XL/iX/Core-One/MK4/MINI)

Native XY Skew Correction — per-printer shear transform applied directly in point_to_gcode() (Printer Settings → General, Expert mode)

FilamentDB integration — REST sync with a remote filament database:

• Auto-push on Save Preset (silent notification)
• Manual "Sync" toolbar button on filament tabs
• Auto-create + update upsert flow (works against an empty DB)
• Spool remaining warning after slicing if the selected filament doesn't have enough on the spool
• Live scan stream — subscribes to the DB's SSE feed and auto-selects the filament preset matching the spool you just loaded into the printer

Virtual Filaments — mixed-color filaments via per-layer tool alternation:

• Auto-generate virtual pairs from your loaded physical filaments, with perceptual (Kubelka-Munk-style) color blending
• Create from target color — type a hex code or CSS color name; the solver finds the closest integer-ratio blend across your loaded filaments (up to 9 components)
• Editable rows with user-supplied names, per-row delete
• Assign via the MMU painting gizmo or directly to whole objects/volumes from the sidebar dropdown
• Height-weighted cadence (thin layers repeat components more often than thick ones)
• Per-row run-length cap to avoid visible streaks on asymmetric ratios
• 3+ component gradient distribution
• Surface-bias offsets and top-surface dithering (data layer; G-code emission in follow-up)
• Full 3MF round-trip; tutorial included

Latest release: PrusaSlicer 2.9.4 — Filament-Edition-1.6.0

RE: PrusaSlicer 2.9.4 with built-in filament calibration

I’d like to take this opportunity to say a big thank you. Great work, Hyiger. I haven’t been able to test all the features yet, but there’s a lot of potential in your ideas.

RE: PrusaSlicer 2.9.4 with built-in filament calibration

After a bit of messing about with retraction towers, changing the code and consulting with Copilot, I'm reasonably confident that retraction is not being controlled by M207 (and deretraction by M208) when the G10/G11 commands are given. The gcode contains G1 E commands and Copilot suggests (in its annoyingly confident way) that FWRETRACT isn't enabled in the printer firmware (Core One), so M207 isn't supported. I'd fill out a bug report, but I'm not totally confident in any of this, only that there's no evidence of any change up and down my retraction tower.

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Posted by: @conrad-2

After a bit of messing about with retraction towers, changing the code and consulting with Copilot, I'm reasonably confident that retraction is not being controlled by M207 (and deretraction by M208) when the G10/G11 commands are given. The gcode contains G1 E commands and Copilot suggests (in its annoyingly confident way) that FWRETRACT isn't enabled in the printer firmware (Core One), so M207 isn't supported. I'd fill out a bug report, but I'm not totally confident in any of this, only that there's no evidence of any change up and down my retraction tower.

There are a number of Marlin g/m-code commands that are either not implemented or only partially implemented in Buddy firmware. The most annoying for me was skew correction. I ended up having to post-process g-code to do it. 

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Well, having more curiosity than common sense, I let Copilot generate a Core One retraction tower and ran it. Nervously. Amazingly, it generated a retraction tower that gave the expected results. I've attached it here so you can see their method. I didn't let it run all the way up as there wasn't anything more to see. It seems to confirm that the Marlin FWRETRACT code is MIA and that the G1 codes are necessary. It also explains why most other retraction towers I've seen published don't seem to work!

RE: PrusaSlicer 2.9.4 with built-in filament calibration

@Conrad thanks for finding this. It's a good catch!

I was able to fix the  Retraction Calibration tool. The old version used firmware retraction, it set the retraction length with M207 and triggered retracts with G10 / G11. As you found, Prusa's Buddy firmware does not implement those commands and ignores them. Every band printed with no retraction, so the towers came out identical regardless of the range entered.

The tool now uses standard slicer-side retraction G1 E, which the firmware does execute, and rewrites the retraction distance per band during post-processing. A few settings are locked during the test (wipe off, seam on the inward faces) so only retraction varies. The towers should now print a visible stringing gradient. I'll test this with various filaments later today. 

Check back in around 1hr, the builds should be finished by then: PrusaSlicer 2.9.5 - Filament Edition 1.7.0

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Ok, found a bug. The code is not ignoring the filament overrides. Will patch it in 1.7.1

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Gave it a quick test and just got home to see the print. Definitely working! Now I need to do a print using the new number and see if they agree. The bug isn't a big deal unless overrides have been set, so not concerned for initial testing. Thanks for doing that so fast- It had to be a significant effort to change the entire scheme.

I shouldn't be miffed at Prusa for removing the FWRETRACT code, as they don't list 207/208 in their list of available Buddy firmware codes, but I am. I can't imagine what the benefit was. The codes are so well documented everywhere it came as a bit of a shock to find they had been removed.

RE: PrusaSlicer 2.9.4 with built-in filament calibration

@Conrad, I put together a spreadsheet of unimplemented G/M codes in the Buddy firmware

RE: PrusaSlicer 2.9.4 with built-in filament calibration

So far so good. I've confirmed the retraction value shown by the tower is valid for unrelated prints. No stringing and I was able to reduce the retraction amount significantly from the default. I'm guessing that anybody who has heat creep issues would benefit from using only the amount of retraction needed, plus a small safety factor, also that wet filament needs larger values than dry filament. Thanks again! These are all the basic tools needed to quickly calibrate our filaments.

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Another thought on retraction and why one needs to test it. The necessary physical movement of the filament may be accurately known, but the retraction setting has to take into account any gear backlash in the print head. The setting probably gets larger as the gears get looser and the backlash increases. Retesting periodically seems like a good idea.

RE: PrusaSlicer 2.9.4 with built-in filament calibration

Posted by: @conrad-2

Another thought on retraction and why one needs to test it. The necessary physical movement of the filament may be accurately known, but the retraction setting has to take into account any gear backlash in the print head. The setting probably gets larger as the gears get looser and the backlash increases. Retesting periodically seems like a good idea.

Sound logic.

Don't have a background in mechanical engineering, but since the load is spread across 3 gears in the Nextruder planetary gears, I would expect wear which results in backlash would be a lot less than a spur gear train? Maybe a question to ask a LLM (with the obligatory "grain of salt").  Still worth considering.