How to Reduce AMS Purge Waste (Filament Poop)

Slice a small multi-color model on a Bambu AMS and the numbers can be humbling: 40 g of model, 60 g of purge “poop”. That waste is the cost of clearing the melt zone at every color change — but the defaults are deliberately conservative, and there is a lot of room between “wasteful” and “color bleed”. Here are the levers, ordered by savings per unit of risk.

Why the waste exists at all

Single-nozzle multi-color systems — the Bambu AMS, Prusa MMU, and every AMS-style clone — swap filament behind one hotend. When red leaves and white arrives, the melt zone is still full of red, and the only way to clear it is to push new filament through until the old color stops contaminating the flow. That purged plastic has to go somewhere: out the back as poop, onto the prime tower, or into a part of the print you will never see.

The purge amount is set per pair of colors. White into red barely matters — a trace of white in red is invisible. Red into white is the expensive direction, because even 1% of red tints white pink. At default settings, a typical dark-to-light swap purges in the region of 1–3 g; light-to-dark is often a fraction of a gram. Multiply the expensive direction by hundreds of layer changes and the waste routinely outweighs the model on small, many-color prints. The fix is twofold: swap less often, and purge less per swap.

Lever 1: reduce color changes, not volumes

This is the highest-impact, zero-risk lever, and most people skip straight past it. Every color change you slice out of the file saves the entire purge for that swap — no bleed risk, nothing to tune.

Also think in blocks of height. A model that is red for the bottom 20 mm and white above it is nearly free; red text wrapping around every layer is the worst case. Stacking colors vertically instead of interleaving them is the cheapest design decision you can make.

Lever 2: tune the flushing-volume matrix

The flushing volumes live in a color-pair matrix, and the defaults are calculated to be safe for the worst realistic case — strong pigments, forgiving of partial mixing. Real pairs usually need less, sometimes much less. Cutting the global multiplier to 0.6–0.8 and then raising only the pairs that actually bleed will typically halve the flushed volume on a tuned printer. Dark-to-dark and same-hue pairs can often run far below default; anything into white, natural or a translucent filament should stay generous.

Bambu Studio: Prepare → Flushing volumes (button next to the filament list)

This lever deserves its own page: our flushing volumes guide covers how the matrix works and a test-strip workflow for dialing in each pair. The short version: change the multiplier first, test with a two-color print that alternates the expensive direction, and only edit individual cells when one pair misbehaves.

Lever 3: flush into infill and hidden objects

Purged plastic does not have to become poop. The slicer can route the flush into the sparse infill of objects on the plate, or into a designated sacrificial object, so the “waste” does structural duty instead of hitting the chute.

Two honest warnings. First, tinting: a single white wall over multicolor infill can read as blotchy in raking light — use 3 walls or skip this lever on display pieces. Second, strength: flushed material is whatever left the nozzle mid-transition, and PETG residue flushed into a PLA part’s structural infill barely bonds to the walls. Keep this lever for same-material, cosmetic-interior parts; turn it off for functional parts and any PLA/PETG or PLA/support-material mix.

Lever 4: right-size the prime tower — don’t delete it

The prime tower (purge tower) looks like pure waste, so it is the first thing people disable. Resist. After every swap the extruder’s pressure state is undefined — the purge tower is where flow re-stabilizes before the nozzle touches your model. Kill it and the first perimeter after each change gets the instability instead: blobs at seams, faint color ghosts, under-extruded starts.

Shrink it instead. A smaller footprint (width 25 mm instead of the default 35 mm) still gives every swap a landing strip, and if you flush into infill the tower only has to absorb the stabilization volume, not the whole purge.

Bambu Studio: Process → Others → Prime tower → Width

Lever 5: purge into a sacrificial object

A variant of lever 3: drop a simple extra model on the plate — a bin, a hook, anything you would print anyway — and mark it as the flush target. The purge becomes a free, randomly-striped object instead of poop. It costs print time and plate space and the surface looks tie-dyed, but on batch plates it converts nearly all transition waste into something usable.

Bambu Studio: add object → right-click → Flush options → Flush into this object

What about the poop you still make?

Some purge is irreducible, so people get creative: melting poops into sheets or coasters, shredding them as filament-recycler input, or using them as glue and paint test pieces. Be realistic — home recycling takes real equipment, and most purge still ends up as scrap. The kilogram you never purge beats the kilogram you recycle, which is why the levers above come first.

The fix ladder

1. Reslice to cut color changes: group colors by height or object, one-color first layer, drop rarely-used colors. Check the swap count in the preview.
2. Use “by object” print sequence when colors live on separate parts.
3. Lower the flushing multiplier to 0.6–0.8; tune stubborn pairs in the matrix (see the flushing volumes guide).
4. Flush into objects’ infill on same-material, non-translucent, non-structural parts.
5. Shrink the prime tower to ~25 mm; never disable it.
6. Add a sacrificial flush object on batch plates.
7. If light colors come out tinted after tuning, walk the volumes back up — our color bleeding guide covers recovering from an over-aggressive matrix.

FAQ

Can I set flushing volumes to zero if I flush into an object?

No. “Flush into objects” changes where the transition material goes, not how much is needed. Set volumes to zero and the transition happens inside your model’s visible walls instead — color bleeding with extra steps.

Why does the printer purge even when both slots are the same color?

The slicer treats each filament slot as a distinct material, and even “identical” spools can differ slightly in pigment and batch. Same-color, same-material pairs are exactly the matrix cells you can safely cut to near the minimum — a small purge remains to restore nozzle pressure after the swap.

How much can I realistically save?

Cutting swap count plus a tuned matrix commonly reduces total flushed volume by half or more versus defaults; adding flush-into-infill can push visible poop close to zero. Treat these as typical outcomes, not guarantees — strong reds into whites and translucents will always demand more purge than the average pair.