Choose a pointing and field orientation that keep the target inside the instrument footprint while capturing the richest set of useful comparison stars — similar brightness (a brighter comp would cap the exposure to avoid saturation), similar color, and well isolated.
A target name is resolved to RA/Dec (or you can type coordinates directly). Before anything else, the target's declination is checked against the chosen instrument's site latitude: if the target can never climb above 20° altitude there, the run stops immediately with an error instead of spending time on a catalog query for a field that can never be observed.
All catalog stars within a circle around the target are pulled from
Gaia DR3 — wide enough to cover the instrument's footprint even
when the optimizer ends up pointing off-target. The official ESA
archive is tried first; if it's unavailable, a VizieR mirror
(I/355/gaiadr3) is used instead. Because Gaia photometry
at a given sky position doesn't change, successful lookups are cached,
so re-running the same target (after tweaking the margin or magnitude
filter, say) skips the network round trip.
Each catalog star (other than the target) gets a usefulness weight.
Your Gmag range (or delta-Gmag) and margin settings are then applied, and anything below the usefulness floor is dropped entirely.
If an "avoid brighter than" limit is set, any star brighter than it (other than the target) becomes a hard constraint in the next step: no pointing may place it inside the footprint.
If an "avoid brighter than" limit is set, any star brighter than it (other than the target) becomes a hard constraint in the next step: no pointing may place it inside the footprint.
The field center doesn't have to sit exactly on the target. The optimizer grid-searches candidate field-center offsets together with a position angle (skipped if "allow field rotation" is unchecked, which fixes the field at PA=0°) — a square footprint repeats every 90°, so only that range needs to be searched. For every candidate pointing that still keeps the target inside the footprint (respecting your target margin) and avoids any "too bright" star, the comparison stars that land inside the footprint (respecting the comp margin) are summed by weight. The pointing and position angle with the highest total score wins.
If no candidate pointing can satisfy every constraint — usually because a bright star can't be dodged anywhere in reach — the optimizer reports that explicitly rather than silently returning a compromised field.
The winning field center, position angle, and offset from the target are reported alongside the selected comparison stars (magnitude, color, separation, and weight) and the footprint polygon, which is overlaid on the Aladin sky view for a visual check before you commit to a pointing.
Footprint size is read from data/FOV_*.vot.xml; the polygon shown
reflects the optimized pointing and position angle. Pan/zoom freely —
“Recenter view” returns to the optimized field.
Proper motion arrows (when enabled) point in each star's direction of motion, colored by category (target vs. comparison). Arrow length is exaggerated and scaled relative to the fastest-moving star shown — only useful for comparing direction and relative speed between stars, not the true angular scale. A comparison star with an arrow parallel to, and about as long as, the target's suggests a common proper-motion (co-moving) companion.
The "Overlay observed pointing" dropdown (shown once a previous observation of this target is found) lists every night this target was actually observed, across all instruments. Picking one draws its real footprint (magenta) alongside the optimized one, from the pointing actually used that night — not a proposal — read straight from the frames obslog rather than the Gaia-driven optimizer above. The footprint size reflects the read mode that night actually used; the orientation reflects the recorded position angle where the instrument logs one (muscat/muscat2) and is drawn unrotated (PA=0°) otherwise, since muscat3/muscat4/sinistro/sbig/qhy600 frames don't currently record a PA. The view re-centers to fit both the target and the observed footprint, so an oddly offset night doesn't end up silently off-screen.
A note under the field lists the separation between the observed pointing and the current target position, flagged when it puts the target outside that night's footprint entirely — a real telescope pointing error, not a display bug. MuSCAT and MuSCAT2 frames carry no native WCS: their header RA/Dec is only the TCS's own belief about where it was pointed, never independently plate-solved, so it can be unreliable — the note flags this for every muscat/muscat2 overlay.