Pricing

You buy render credits up front. What a render costs is computed from the finished image and from two things you choose about how it is made:

Why megapixels, and not "renders"

"One render" is not a unit. A 512 × 512 look at a single repeater and a 16384 × 16384 regional study are both one render, yet they differ in size by a factor of a thousand. Any flat per-render price would be far too expensive for the first or far too cheap for the second.

The unit therefore has to scale with the job. The finished image is the base it scales from, and that choice has three important properties:

• It is known before any processing begins. The output image size is part of your request, so the exact cost can be calculated and shown to you before anything has been spent. This is what allows a prepaid model to work fairly: there is no bill after the fact.
• It does not change when the service changes. How a job is divided internally is an engineering detail that has already been revised twice. Had pricing been tied to internal work units, each revision would have changed the price of identical output by a factor of four. Basing the price on the finished image allows us to keep improving the service without changing what anything costs.
• You can check it. Multiply the two sides of your output grid, divide by a million, multiply by the square of your supersample setting and by the number of transmitters, and round up. Every one of those is a field in your own request, so you can verify your own invoice without relying on a figure only we can compute.

Why supersampling and transmitter count are in the formula

They were not included at first. Billing on the finished image alone made the two most expensive kinds of request cost no more than the simplest.

• Supersampling is passes, and passes are squared. Supersample ×2 renders four times and ×3 renders nine times; the result is then resolved into an output image of exactly the same size. Billing on the output alone charged the same for nine passes as for one.
• Each transmitter is its own propagation calculation. A co-channel study against eighteen neighbours computes eighteen coverage fields and combines them into one image of unchanged dimensions. That is eighteen times the work, which was previously charged at the price of one.

Both multipliers are settings you choose and can see before you spend. The cost can still be quoted in full before a render runs, which was the purpose of the original unit and remains so.

What a request costs in credits

The table below is arithmetic on your own request, not a price list. Multiply by the rate per credit once it is set.

Typical work runs between 512 and 8192 pixels per side. These are three independent factors, and they multiply. Doubling the grid is four times the credits, every time, and four doublings is 256×. Supersample ×3 is nine times. Eighteen transmitters is eighteen times. The last column applies all three at once, which explains its size, and it is a request any customer could make. Whether that detail exists in the terrain is a separate question, and usually the more important one.

Things that do not change what you pay

How the job is divided internally. This is deliberately invisible to you, and it is the main argument for this unit, as set out above.

The range you ask for is not a factor in the formula as it stands, so at a fixed output size a 100 km map and a 10 km map cost the same. Treat that as NOT FINAL rather than as a promise: it is an open question whether range gains a factor, and it is listed below with the others. What range does change is your ground sample — the same pixels spread over more distance — which is a resolution decision, and the one most worth considering.

Rounding up to a whole credit

Credits are charged per render, and any part of a credit is rounded up to a whole credit. A 4096 × 4096 render at supersample ×1 is 16.8 megapixels, and is charged 17 credits. The rounding makes the most difference at the small end: a 512 × 512 render at the same setting is 0.26 megapixels and is charged one credit, so very small renders cost more per megapixel than large ones.

Why prepaid, and not a subscription

Credits are bought before they are spent. This is a genuine trade-off, and we want to be clear about it: we hold your payment before the processing is performed, and you carry that balance until you use it. That is a real cost to you. Credits do not expire, so the balance remains available until you use it. The reasons for choosing this structure are as follows.

• There is no bill to dispute. There is no credit limit, no overage, no invoice arriving after the month has ended, and no collections. The balance check happens before a job is accepted; if your balance does not cover a render, it is declined up front with the shortfall shown, rather than run and billed.
• You do not pay for months in which you do not render. Coverage planning is intermittent. Most customers need a set of maps around a decision and then nothing for a long time. A subscription charges most heavily when it is least useful.

Purchases and your balance

These policies are decided and apply when payment opens.

• Charged per render, rounded up. Each render is charged in credits from the formula above, and any part of a credit is rounded up to a whole credit.
• The minimum purchase is $30.
• Larger purchases receive a lower rate per credit. The volume tiers are still being finalised.
• New accounts receive $20 of free credit, so you can run your first renders before making a purchase.
• Credits do not expire. A balance bought for one project remains available for the next.
• Unspent credit is not refundable at this time.
• A render that fails is not charged. Credits are charged only for a render that completes successfully, and a failed render can be run again.
• Texas sales tax applies and is shown at purchase, before you pay.

Numbers that are not decided

The parameters below are listed openly so that it is clear which figures remain undecided, rather than leaving gaps or publishing numbers that have not been set.

See also: Resolution & cost, and the FAQ — particularly what counts as a "full" render versus a "small" one.