Research-Derived Protocol Rules

Target Optimization

Decades of target-pool literature converge on a short list of rules. A target should not have a distracting background. It should not contain multiple elements competing for attention. It should be visually coherent enough to sketch. Most pools follow these rules approximately, by curation and good intentions. We follow them absolutely, by construction.

A distracting background is not possible in our pool, because our targets have no background at all — every image is a single subject, fully isolated, on nothing. Multiple competing elements are not possible either: the pipeline enforces exactly one subject per image, verified by isolation analysis that rejects any image with stray fragments or secondary elements outside the subject's own mass, and every survivor still passes through human review before it earns a place in the pool. The entropy screening described in Shannon Entropy (ΔS) then removes the structurally boring remainder. The pool is not a folder of nice photographs. It is a manufactured instrument.

And with an account, you will never see the same target twice. Every target you have ever been shown is excluded from every future selection, permanently.*

* This guarantee requires an account for an honest reason: anonymous sessions are not tracked — we do not know who you are, by design — so there is no way to know what an anonymous visitor has already seen.