Print settings
Cold Crabby splits settings three ways, the same way established slicers do. Knowing which of the three owns something is most of the battle.
| Row | Describes | Changes when |
|---|---|---|
| Printer | The machine | You buy a printer or change the nozzle |
| Filament | The spool | You swap material |
| Process | How to print this thing | Every print, potentially |
The three rows at the top of the panel name the profile in use for each. Click a row to show its settings; the ⋯ beside it swaps the profile, and every option below fills in. Managing those profiles is covered in Printers, filaments and profiles.
Can't find something?
Press Ctrl/⌘ + F and type. The search spans all three, so you don't have to guess which one owns it.
The three presets, all at once
The top of the slice sidebar lists what this plate is made of — its printer, its filament and its print profile. Every one names the preset it is using, and the three dots beside it open the menu of the others.
They sit side by side when the sidebar is wide enough to read them and stack when it is not, so widening the sidebar buys you a shorter header rather than just longer names. The whole lot stays put while the settings scroll, along with the search — you never lose sight of which preset you are editing.
There is no tab to press. Click a row and the settings below are that row's; the three dots beside it change which preset the row is using. Pointing the settings somewhere is what you do most, so it takes the whole row, and the dots sit beside it without ever getting in the way of it. Clicking a row that is already selected opens the same menu as its dots, and so does right-clicking any row — or a long-press on a touchscreen.
In that menu, every preset carries a pen on the right that opens that preset's own editor — not just the one in use. Editing takes you off the plate and picking does not, so the two are kept apart: the menu is where you choose, and the pen is the way out of it.
Links take you to the knob
Where a setting depends on something on another profile, the note under it links straight to that control — the right profile opened, scrolled to, and flashed — not to the top of a page of sixty settings with the rest left to you. The pen in the preset menu works the same way: it opens the preset named beside it rather than the list it came from.
A warning is never hidden
Settings are tiered: the everyday ones are shown, and Advanced and Expert reveal the rest when you go looking. A tier decides what is worth offering you — it never decides what you are told.
So a setting the slicer has something to warn you about is shown whatever tier it normally sits in, with the warning attached, for as long as the warning holds. Load ABS on a printer with no chamber heater and Chamber Temperature appears in the Filament tab with a note saying no chamber command will be emitted, even though it is an advanced setting you never asked to see. Change the filament and it folds away again.
Changes belong to the plate
Everything you see in the panel comes from the three profiles you picked. When you change one of those values, the plate remembers that one change — not a copy of every setting.
You can see which ones they are: a changed setting's name is italic, and its section gets a dot, so a collapsed Walls still tells you something inside it was touched. A strip pinned to the bottom of the panel counts them and offers Reset all, which hands every changed setting back to its profile. Setting a single value back to what the profile says does the same for that one — it stops being a change and starts following the profile again.
This matters for a reason that only shows up later. Edit a print profile — say you decide 4 walls, not 3 — and every plate using it prints with 4 walls, including the ones you already tuned. The only settings that stay put are the ones you deliberately changed on that plate.
Making a plate's changes permanent
Sometimes a change you made on one plate turns out to be one you want on every plate — the nozzle temperature you nudged for this spool, the wall count you settled on for this printer. Sync to profile, next to Reset all, opens a review of every changed setting: what the profile says next to what you changed it to. Switch off the ones you don't want carried over, then confirm. Accepted settings are written into the printer, filament or process profile they belong to and stop being a change on this plate — they are now what the profile says, same as if you had opened its editor and typed them in yourself.
Your changes are saved to the plate as you make them; the panel says so briefly underneath. Come back to a plate a week later, from the tab bar or your history, and it opens with the printer, filament and profile it was set up with, where you put each model, and the changes you made on top — even if you have been printing something else in PETG since.
Where that is saved depends on how you run Cold Crabby, and it is the same rule as your profiles: on the server if you self-host, on the machine if you use the desktop app, and in the browser only if you use the web version. Clearing your browser does not cost you your plates unless the browser is all you have.
Advanced — what actually gets sliced
Cold Crabby tells the slicer which printer, filament and print profile you picked — by name, not by sending copies of them — plus your list of changes. The slicer already has your profiles, and combines them itself: engine defaults, then printer, then filament, then process, then your changes, each winning over the one before. The combining happens in one place, so the command line, the desktop app and the browser cannot disagree about what a profile means.
Naming them rather than sending them is also what makes an edit stick. If every slice shipped a copy of your profiles, the last browser to slice would quietly overwrite a change you made in another tab, or that a colleague made on a shared slicer.
The one large thing still sent every time is the preview picture embedded in your G-code, because it is a picture of your 3D view — your camera angle, your theme, your filament colour — and only your browser can draw it.
The five settings that matter most
If you change nothing else, understand these.
Layer height (Process → Layer) — how thick each slice is. Default 0.2 mm. Smaller means smoother and slower; larger means faster and more visible layer lines. Stay at or below about 75 % of your nozzle diameter.
Walls (Process → Walls) — how many perimeter loops around the outside. Default 3. This, more than infill, is what makes a part feel solid. Going from 2 to 4 walls does more for strength than doubling infill.
Infill density (Process → Infill) — how much material inside. Default 20 %. Decorative parts are fine at 10 %; functional parts want 30–50 %. Above about 50 % you're usually better off adding walls.
Temperatures (Filament → Temperature) — nozzle 210 °C and bed 60 °C by default, which suits typical PLA. Your filament's label wins over any default.
Supports (Process → Support) — off by default. Turn them on when your model has overhangs steeper than about 45°. Check the preview afterwards: supports that touch nothing are wasted plastic and a worse surface.
Supports normally stand wherever they are needed, including on the model itself. Only from build plate restricts them to columns that can reach the bed through open space. Anything that would have to rest on the print is dropped, so those overhangs print unsupported — a deliberate trade you make when a support landing on the model would scar a surface you care about, or sit somewhere you could never get a tool into. Expect to lose coverage: on a shelf overhanging a wider base, the part beyond the base is still supported and the part above it is not.
Advanced and Expert
Each section shows the settings a print actually depends on, then a quiet Advanced row at the bottom with a count — Advanced 10. Pressing it expands that section in place, so you keep your scroll position and your context; a second press reveals Expert. Nothing moves, and there is no "expert mode" to switch the app into.
The split is about whether you can form an intention, not about how experienced you are. Seam position is Advanced because you can want the seam at the back. A bead-transition threshold is Expert because almost nobody can predict what changing it does — including people who have been printing for years.
Two things worth knowing:
- Search ignores all of this. Type into the settings search and you reach every setting the slicer has, whatever tier it sits in and whichever tab owns it. If you know the name — including the name another slicer uses for it — that is the fastest way there.
- Anything you have changed stays visible, even if it lives in Expert. A setting you can't find is a setting you can't put back.
Whole sections work the same way. A few hold nothing but advanced settings — Quality, Thumbnail, Time estimate — so they are not listed until you ask for them; a section header that opens onto nothing is worse than no header. The Advanced sections control at the bottom of the list brings them in, which is what keeps Process at seven sections rather than eleven.
Each section, and the list itself, remembers how far you opened it — so if you work in Advanced you only say so once.
If you always want everything in view, Settings → General → Settings detail sets where the panels open: Standard, Advanced, or Everything. It moves the starting point only — the per-section controls still open further, search still reaches everything, and nothing is hidden from you at any level.
Everything else, by group
Printer
| Group | What lives there |
|---|---|
| Hardware | Nozzle diameter, bed size and shape, kinematics, gantry clearances, whether the firmware can cancel a single object |
| Retraction | How far and how fast filament is pulled back on travel; Z-hop |
| Output | G-code flavour (Marlin or Klipper), start and end scripts, lifecycle markers |
Filament
| Group | What lives there |
|---|---|
| Material | What the spool is — type, name, colour, diameter, density, cost |
| Temperature | Nozzle and bed, with separate first-layer values |
| Cooling | Fan speeds, minimum layer time |
| Extrusion | Flow ratio, maximum volumetric speed, pressure advance — the numbers you calibrate per spool |
| Filament G-code | Custom G-code for this material |
Process
| Group | What lives there |
|---|---|
| Layer | Layer height, first-layer height |
| Walls | Wall count, wall generator, thin walls, extra perimeters, ordering, seam behaviour, fuzzy skin |
| Infill | Density, pattern, angle |
| Support | On/off, type, density, overhang threshold, interface layers, clearances, whether support may only start from the build plate |
| Speed | Per-role print speeds and accelerations, travel speed, and how gently travel arrives at an outer wall |
| Quality | Bridging, dimensional compensation, other accuracy options |
| Surfaces | Top and bottom solid layer counts, surface fill, ironing |
| Adhesion | Skirt, brim, raft |
| Objects | Print order, G-code run between objects |
| Thumbnail | The preview image embedded in the G-code file |
| Mesh | How the incoming model is interpreted |
Options that only apply in certain configurations hide themselves. Choosing the classic wall generator, for example, reveals options the Arachne generator doesn't use — so the panel never offers you a control that would do nothing.
Speeds are in mm/s
Every speed in the panel reads in mm/s, travel and retraction included. G-code measures those two in mm/min, and that is still what gets written to the file — you just don't have to do the arithmetic to compare travel against your print speed.
If you think in mm/min, press the unit beside any speed box. Every speed in the app switches together, the choice is remembered, and nothing about the saved value changes — only how it is written.
The thumbnail in your G-code
Most printers show a preview of the print on their screen, and that picture is embedded in the G-code file. Process → Thumbnail decides what goes in it: its size, the camera angle it's shot from, a light, dark or transparent background, and the model's colour.
How the model is rendered is not a print setting — the picture is taken in the 3D view, on your machine — so it sits with the other graphics options under Settings → 3D View → Thumbnail look. Plain, the default, shoots it flatly, which means the same plate previews identically wherever it's sliced. Match this view borrows your own scene's shading, gloss and contact shadow instead.
Infill patterns
| Pattern | Character |
|---|---|
| Rectilinear | Parallel lines, alternating direction each layer. Fastest. |
| Grid | Lines crossing at right angles. Stronger, slower. |
| Honeycomb | Hexagons. Good strength for the material spent. |
| Gyroid | A 3D curve. Equal strength in every direction; nice for flexibles. |
| TPMS-D (default) | Diamond minimal surface. Organic and isotropic — the same density holds up in every direction. |
Two special modes
Spiral (vase) mode (Process → Walls) prints a single continuous wall that climbs as it goes — no seam, no layer changes. For open, single-walled models only. Turning it on forces the settings it's incompatible with (extra walls, infill, top layers, retraction) off for you, and keeps your bottom layers as the base.
Ironing (Process → Surfaces) makes a second, hot, barely-extruding pass over top surfaces to smooth them. Slow, and only worth it on visible flat tops.
Fuzzy skin (Process → Walls) roughs up the outer wall with a random, hand-textured bump instead of a smooth surface — useful for hiding layer lines or giving a part a deliberately organic look. It's purely cosmetic: inner walls and infill print exactly as they would otherwise, and turning it off always reproduces the original smooth output.
Advanced — tuning the texture
Thickness is how far each bump can push in or out, in mm — larger values give a coarser, more pronounced texture. Point distance is how closely spaced the bumps are along the wall — smaller values pack in more, finer detail.
Advanced — tuning the ironing pass
Type chooses what gets swept: every top surface, only the single highest one (much faster on a tall model, and usually the only face anyone sees), or all solid surfaces.
Flow is how much material the pass adds, as a percentage of a normal bead — around 10 % is enough to re-melt the surface without raising it. Spacing is how far apart the passes run; well under a bead width is what flattens the ridges between them. Speed should stay low, because the nozzle needs dwell time to melt what it crosses. Angle defaults to following the layer's own fill direction; set an explicit angle to cross the fill instead, which flattens it more effectively.
Getting parts to the right size
A printer lays a bead slightly wider than asked, so parts come out a little large and holes a little tight. Both are consistent for a given machine, so both can be measured once and corrected (Process → Quality).
XY size compensation grows or shrinks every contour by a fixed amount. Print a test cube, measure it, and set the difference as a negative value if the cube came out oversized. Because the material spreads inward as well as outward, this also tightens holes.
Hole compensation adjusts holes on their own, so a peg that will not fit can be freed without changing the outside of the part.
Both default to off. Start from a measurement, not a guess — and keep the values small; a shrink larger than a thin feature will erase it, which the slice log warns you about.
The flare at the very bottom
Elephant foot compensation fixes a different problem from the two above. The first layer is deliberately squashed into the bed to make it stick, so it spreads sideways and only the base measures oversize — enough that a part won't sit flat, or won't drop into the hole it was designed for. XY size compensation would shrink the whole part to fix the bottom of it.
Measure the bulge with calipers, halve it, and put that in. 0.1–0.2 mm covers most machines. Only the first layer is corrected, because only the first layer is squashed.
It won't eat your first-layer detail
Shrinking the first layer sounds like it should wipe out embossed text and thin logo strokes — that's what a plain shrink does. This one measures how thin the geometry is at each point and simply stops there, so a fine feature keeps its width while the walls around it are corrected in full. It also leaves the base alone where the model already flares outward above it, and skips itself entirely when you print on a raft, where nothing touches the bed.
Advanced — tuning the correction
Elephant foot layers spreads the correction over more than one layer, ramping it to zero. Leave it at 1 unless a large correction leaves a visible step at the second layer.
Minimum contour width is the width the correction will never shrink a feature below. Left at 0 it works this out from your wall width; raise it to protect chunkier detail, lower it for a more literal correction.
First layer height (Process → Layer) prints the bottom layer thicker than the rest. The extra material absorbs what mesh bed levelling only approximates, which is why almost every profile sets it. It has no effect when you print on a raft, since the raft takes over contact with the bed.
Where your settings are saved
Changes in this panel apply to the current plate. To make them permanent, save them into a profile — see Printers, filaments and profiles.
Advanced — configuring outside the UI
The CLI and self-hosted server read a layered slicer.toml: built-in defaults, then your user config, then a project slicer.toml in the working directory, then command-line flags. Each layer deep-merges over the last, so a project file only needs the values it changes. See Configuration.