Quick answer: purge grams × filament price
Purge material cost for one plate = additional purge grams ÷ 1,000 × filament price per kilogram. Full plate material cost = (part, support, purge and tower grams) ÷ 1,000 × that price. Divide the purge cost by the number of saleable parts only when you need the purge allocation per good part.
Use this separate formula only when the slicer reports purge or wipe-tower material apart from the model and supports. If its total filament value already includes all waste, multiply that total by the price per gram once. Adding the separate purge again would double-count it.
Start with one complete plate
Slice the same product in each configuration. Record the number of saleable parts, whole-plate print time and material use for that plate. Use the same quality requirements and include the supports you actually intend to print. A faster draft preset and a slower finished-product preset answer different questions.
Keep your two slicer outputs separate. Do not estimate multicolor time by adding an arbitrary percentage to a single-color print. Color changes, purge settings and layer layout can change both time and waste.
Choose one material input method
In the full cost-comparison calculator, if your slicer gives a complete filament total that already includes the part, supports and waste, choose “Total includes purge.” Enter that total in All material. The separate purge box becomes inactive.
If you have separate values, choose “Parts + separate purge.” Put part and support material in one box and all additional flushed material and tower material in the other. Do not add a tower twice because it appears in two different summary screens. If a slicer reports volume instead of grams, convert using the actual material density before entering a weight.
PrusaSlicer 2.9.6 can show flush and wipe-tower consumption as per-extruder weights in the right sidebar. When those are separate from the part-and-support figure, add the displayed flush and tower weights once in the purge field. If the complete filament total already includes them, use total mode instead.
OrcaSlicer documents `total_weight` / `extruded_weight_total` as the whole job weight, so use that value in total mode. Its `total_wipe_tower_cost` is money and `total_wipe_tower_filament` is volume in mm³—not grams. Do not paste either into a gram field. If you only have those tower values and cannot obtain a separate weight, use the whole-job weight instead of inventing a conversion or adding the tower twice.
If a slicer flushes a color change into an object, its infill or its support, do not assume that every flushed gram remains a separate waste category. Use only the split the slicer actually reports. When it reports only a complete filament total, use total mode: moving purge into printed geometry can change its category, but it does not remove those grams from material cost. Verify appearance and strength for your own saleable parts; this calculator does not predict either.
Compare cost per good part
Material is only one line of the result. Add electricity, machine wear, setup time, finishing, packaging and expected failures. Divide the total by the successful part count. A plate with ten parts should not be compared with a five-part plate using total cost alone.
Prusa documents that its smart wipe tower is shared by the objects on a plate and is influenced by color changes. That supports checking a multi-copy batch in the slicer; it does not establish a universal saving for every printer or model.
Decide whether a color premium is enough
Take the extra modeled cost per part into the pricing calculator. If the sales channel charges a percentage fee and you want to preserve a margin, the required increase in selling price is larger than the increase in production cost. Demand is a separate test: a customer may not pay that premium.
Use an average filament price for this version of the calculator. If the colors have materially different costs, calculate a weighted price from their actual weights first. No model files are uploaded here, and the example inputs are hypothetical rather than measured production data.