The framing around AMS-style multi-colour printing has matured considerably in the two years since I started this site. The initial coverage — from this site and most others — focused on a fairly simple axis: purge waste is bad, here is how to reduce it. That is still true, and the waste reduction guides remain relevant. But it has increasingly felt like an incomplete account of what running a multi-colour workflow actually costs in practice. Filament waste is the part that shows up in the bin after a print. The costs that do not show up there — the project planning constraints, the queue disruption, the spool state management, the cumulative failure risk on complex long jobs — are harder to see and more consequential to how you actually work. This post is an attempt to account for all of them.
My own position, disclosed upfront: I limit my AMS use deliberately and lean heavily toward multi-part printing for projects that would otherwise be multi-colour AMS jobs. That preference is documented in the multi-part printing post and has hardened with time rather than softened. This post is not a case against multi-colour printing — it is a case for understanding what it costs operationally before defaulting to it, which is a different and more useful question than whether the AMS wastes filament.
Project planning changes: how multi-colour restructures creative decisions
The first cost is the one that starts before a print begins, and it applies at the model selection and design stages rather than at the slicer. Multi-colour printing imposes specific constraints on model geometry that single-colour printing does not, and those constraints restructure which models you choose, how you modify them, and what is achievable at all within a session.
A model optimised for AMS multi-colour printing needs its colour boundaries to fall along geometric layer transitions or within paintable regions — not at arbitrary points in 3D space. This is not a fundamental limitation on what AMS can print; it is a fundamental characteristic of how single-nozzle colour change works. Every colour transition happens when the AMS swaps filament between one layer and the next, or when the slicer paints colour zones that the AMS cycles through. Models where colour is integrated into complex organic geometry — gradient transitions, spots or patterns across curved surfaces — require the texture-to-colour painting workflow now available in Bambu Studio 2.7.0, which adds its own preparation overhead and imposes its own colour-count and transition-quality constraints.
The practical effect: when browsing for models to print with the AMS, you are implicitly filtering for a specific subset of available geometry rather than the entire model space. Models that would be straightforward single-colour prints — download, slice, print — become candidates for evaluation: does this model have colour zones that the AMS handles cleanly? Does it exist in an AMS-optimised variant? Is there a multi-part non-AMS version that would produce better colour separation without the purge overhead? This evaluation is fast once it becomes habitual, but it is a real cognitive step that single-colour printing does not require, and it runs upstream of every decision in the rest of the workflow.
There is also the slot assignment question. Four slots. Current filament. Which colour goes in which slot, which models this session can be batched together because they share colour configurations, and which require a mid-session spool swap that effectively halts the AMS while you reconfigure it. Experienced AMS users develop a mental model of their current slot state and project future prints through it automatically. But that mental model is maintenance overhead that single-colour printing simply does not require. The slot planning question only exists because the AMS exists, and it runs in the background of every print decision made during a session.
Purge tower management: the design problem, not just the waste problem
The purge tower discussion usually starts and ends at material cost, which misses the more interesting operational dimension. The purge tower is not just a waste generator — it is an active participant in the print that occupies plate real estate, affects print time through its own layer cycle, interacts with the model’s geometry and placement, and creates a structural risk that grows throughout the job.
Plate real estate is the first structural constraint. A purge tower placed beside a model on the A1’s 256mm plate occupies space that could otherwise be a second copy of the model, a companion piece, or simply additional breathing room that reduces the risk of warping at the model’s edges. On a complex multi-colour model that already covers a significant portion of the plate, the purge tower may not fit beside it in a comfortable position, forcing either a plate size compromise or a tower position that creates travel distance inefficiencies. The A2L’s 330mm plate changes this equation positively, which is one of the understated benefits of the larger bed for multi-colour work specifically.
The worst-case waste numbers are striking enough to quote directly. One documented example found a multi-colour print using 1.56g of filament for the actual object but generating 4.93g of waste — more than three times the model weight in purge material. Make: magazine tested AMS-style multi-colour printing and described a job that took almost four days from start to finish, with swap overhead accumulating across hundreds of colour transitions. These are extreme examples on swap-heavy models rather than representative of typical simple multi-colour jobs, but they establish the actual ceiling of what the waste and time overhead can become on the wrong model.
The structural risk aspect is the one most often overlooked. The purge tower is a separate printed structure on the same plate, built from whatever colours are currently active in the print. It is, by design, less carefully printed than the model — its function is to absorb waste material, not to be structurally sound — and it grows taller alongside the model throughout the job. As it gets taller, it gets more top-heavy. A purge tower that successfully maintained integrity through the first hundred layers of a complex print can fail at layer two hundred, particularly on models with many colour changes that deposit inconsistent amounts of each colour into the tower on different layers. A fallen purge tower mid-print does not necessarily end the job, but the toolhead travelling into a knocked-over tower structure introduces a very real risk of print failure, nozzle collision, and potentially a ruined plate. The No Sparse Layers setting — covered in the dedicated guide — addresses the height problem by only building the tower on colour-change layers, which reduces the structural risk significantly. But it requires OrcaSlicer, is not available in Bambu Studio natively, and adds its own edge-case management.
Print queue disruption: how multi-colour jobs block single-colour capacity
This is the cost that becomes most visible on a two-printer setup and was one of the clearest observations from the period when the A1 and Kobra X were running simultaneously. Multi-colour AMS jobs occupy the A1 for substantially longer per unit of model output than single-colour jobs, because the colour change overhead — the swap, the purge, the travel to and from the tower — is proportional to the model’s colour complexity rather than its size. A small complex multi-colour figurine may take four hours including colour changes. The same printer could run two substantial single-colour prints in the same four hours on the same plate, with no purge waste and no swap overhead.
The queue disruption is the opportunity cost of that four-hour block: every multi-colour job that runs is a single-colour job that did not. For a printing workflow that mixes both types of work — as mine does — this creates a scheduling question that does not exist when all printing is single-colour: is this job worth the queue blockage it creates, or would splitting the model into multi-part single-colour jobs free the printer for better overall throughput? The answer varies by model and urgency, but the question itself is a cost that the AMS introduces. The Kobra X parallel printing experience illustrated it vividly: running the Kobra X on multi-part single-colour sections while the A1 ran the AMS components produced total throughput that a single machine could not match, and the benefit was almost entirely in removing the queue blockage that long multi-colour jobs created.
The disruption extends to the kind of interruptions that multi-colour jobs require but single-colour jobs do not. A filament approaching end-of-spool during a multi-colour job requires intervention — either loading a new spool in advance with the colour and diameter matched, or catching the run-out and managing the AMS’s recovery behaviour, which is imperfect and sometimes involves manual assistance that a single-colour print’s simple run-out pause does not. A spool that looked adequately full at the start of a three-hour multi-colour job may not be adequately full at hour two. Tracking remaining spool amounts across four loaded slots — and projecting whether each colour has enough remaining for the current job plus anticipated upcoming work — is another background task that single-colour printing does not impose.
Spool management: the operational complexity that scales with colour count
The spool management overhead of multi-colour printing is the most genuinely insidious of the hidden costs because it is most diffuse and hardest to quantify. It does not add to any individual print’s time or waste directly. It accumulates as a background tax on how the filament collection must be maintained to support multi-colour workflows reliably.
The most common single AMS failure mode — responsible for 43% of all AMS failures according to community troubleshooting documentation — is filament tangling in the AMS spool holder. The AMS does not actively unwind filament from the spool; it pulls it through a PTFE tube, and if the spool binds or the filament crosses over itself, the pull force exceeds what the extruder can handle. This is almost entirely preventable by properly winding your spools and using filament that comes on well-wound spools. The key phrase is “properly wound spools” — maintaining that across four active spool positions, some of which may be partially used spools that have been removed and re-loaded between sessions, is an active maintenance task rather than a passive storage question. A full spool from a good manufacturer behaves predictably. A partially used spool that has been sitting on a shelf for three weeks, wound under slightly different tension, may tangle at the AMS. Knowing which spools in your collection are safe to load without inspection and which need checking first is operational knowledge that only accumulates through experience.
The moisture management dimension compounds this. Multi-colour printing means having multiple spools loaded in the AMS simultaneously for extended periods — across multiple print sessions, potentially over several days. The AMS Lite is not a sealed desiccant environment. Filament left in the AMS Lite for extended periods absorbs ambient humidity at a rate that depends on the filament type and local conditions. PETG in a humid UK workshop environment loaded into the AMS on Monday may be producing surface defects by Thursday — not because it is unusually hygroscopic, but because the combination of open storage in a sometimes-humid room and the AMS’s lack of active humidity management produces gradual moisture uptake that is only apparent in print quality after it has progressed enough to matter. Monitoring filament condition across four loaded slots, rotating spools back to airtight storage between uses, and drying spools that have been sitting loaded for several days is genuine operational overhead that single-colour printing — where you run a spool, put it back in its container, run the next one — does not create.
The partial spool problem deserves specific attention. Multi-colour printing rapidly generates partial spools — 300g remaining on a white that has not been used since the Mario series finished, 180g on a yellow that has only appeared in one project, a nearly-full red that somehow never gets fully consumed because the models needing red always need it in small proportions relative to the other colours. Managing these partial spools, understanding what is left on each, avoiding loading a partially used spool for a long job that requires more of that colour than remains, and eventually either using up or disposing of the small remnants that accumulate from completed projects — all of this is overhead that grows in proportion to the number of multi-colour projects run. The filament inventory feature in Bambu Studio 2.7.x and Bambu Handy 3.22.0, covered in the Handy update post, addresses this directly, but its current value depends on whether your filament collection is primarily Bambu-branded RFID-tagged spools or a mixture of branded and third-party. For a mixed collection, the inventory requires manual maintenance to be accurate.
Failure risk: why long multi-colour prints fail more
The failure risk of any FDM print is not zero, and on a well-maintained machine with good filament and accurate profiles, it is low enough that most prints complete successfully. Multi-colour printing does not change this at the per-layer level. It changes it at the system level by introducing additional failure modes that do not exist in single-colour printing, and by increasing the consequence of any failure because the longer the print, the more time invested in whatever fails.
The second most common AMS failure mode — after spool tangling at 43% — is filament not reaching the printhead, accounting for 31% of failures. The PTFE path from AMS to printhead is long, and flexible filaments or filaments with inconsistent diameter can struggle with the distance and friction. This failure mode does not exist in direct-feed single-colour printing because there is no long PTFE path from a separate unit. It is structurally inherent to the AMS architecture and appears at random in a printing session — not on every job, not predictably, but often enough that extended multi-colour print farms include it in their risk planning.
A multi-colour job that fails at the 60% mark after four hours is categorically worse than a single-colour job that fails at the 60% mark after ninety minutes. The time loss is proportionally different, the material loss is higher because of the purge waste accumulated during the completed portion of the job, and the psychological cost — setting up again, re-evaluating whether the job parameters were correct, deciding whether to restart or redesign — is cumulative in a way that the occasional single-colour failure is not. This is why experienced multi-colour printers develop a pre-print checklist that single-colour printers rarely need: checking spool tension and remaining amounts across all four slots before a complex job, inspecting the AMS wiper condition, verifying that the purge chute is clear. The checklist exists because the failure surface of a multi-colour print is larger, and the checklist is the operational response to that larger failure surface.
The cumulative failure risk matters most for the longest and most complex jobs — which are also the jobs where the benefit of multi-colour printing is greatest and the case for using it is strongest. A 20-minute two-colour print has minimal multi-colour failure risk and little reason to choose multi-part instead. A six-hour complex figurine with hundreds of colour changes has meaningful accumulated failure risk and a legitimate case for multi-part as the alternative. The risk and the benefit are inversely correlated in a way that the “use AMS for complex multi-colour work” heuristic does not capture without this nuance.
The operational framework: when to absorb the costs and when not to
None of what is above is an argument against multi-colour printing. It is an argument for being intentional about when you absorb the workflow costs rather than defaulting to the AMS because it is there and because the models are available in AMS-optimised form.
The cases where absorbing the costs is genuinely justified: models where colour is integrated into the geometry in ways that multi-part cannot replicate cleanly — fine painted detail, gradient zones, patterns across continuous surfaces. Repeat production runs where the upfront planning investment pays back over multiple identical prints. Short, low-swap-count jobs where the purge overhead is small relative to the model and the failure risk window is short. Models that only exist in AMS-optimised form and do not have credible multi-part alternatives.
The cases where multi-part is the more considered choice: complex figurines with clean geometric colour boundaries that multi-part handles naturally and AMS handles wastefully. Large-scale projects where the print time and failure risk of a single multi-colour AMS job are prohibitive. Projects where repairability matters — the multi-part post covers this extensively. Any project where the spool configuration for the AMS would require a setup that disrupts other ongoing work or empties partial spools that are needed elsewhere. And, honestly, most decorative display pieces at moderate scale — the category that makes up the majority of what most hobbyists actually print for enjoyment rather than function.
The decision framework I have settled into after two years of running both approaches: if I can achieve the same result with multi-part printing, I default to multi-part. The AMS earns its use specifically in cases where multi-part cannot replicate the colour integration I need. Not because the AMS is bad — it is genuinely impressive hardware — but because the operational overhead of multi-colour printing is real, and the cases where it is the definitively better choice are narrower than the default reach for AMS-optimised models implies.
