Inspect and Select Light Frames with Blink and SubframeSelector

Blink lets you step through exposures and spot clouds, tracking failures, focus changes, or framing problems. SubframeSelector measures properties such as star size and shape so you can identify frames worth a closer look. Use the measurements alongside the images rather than letting one number make every decision.

Before the full WBPP run, use Blink for an initial check of the raw lights. For reliable SubframeSelector measurements, use correctly calibrated linear review copies; for a color camera, use the corresponding debayered copies. PixInsight staff explain that uncorrected hot pixels can be misidentified as stars in raw images.

That means the measured review stage comes after calibration, even though selection happens before the final registration and stacking run. This lesson uses a preliminary WBPP calibration pass, then carries the decisions back to an approved list of original raw lights. Keep those two file sets clearly separate.

Keep all original exposures unchanged. Make a small selection log with source path, filename, filter, exposure, decision, and reason. Use full paths if different sessions contain repeated filenames.

Start with one target, camera configuration, filter, exposure time, and binning. Repeat the review for other groups. A narrowband exposure and a broadband exposure can have very different star counts and backgrounds without either being defective.

Note changes of session, image scale, gain, or readout mode. These can affect measurements. Do not rank unlike groups together and automatically discard the faintest filter or the shorter exposures.

For practice, choose a manageable representative group. Include several normal exposures and any suspected failures so you can learn what their measurements look like.

Find Blink through All Processes, usually under ImageInspection. Load the raw light files and verify their names and counts. CFA color-camera data may show a mosaic pattern during this initial check; do not modify the raw files just to make the preview prettier.

Dithering shifts stars between exposures, and a meridian flip changes orientation. These do not automatically make a frame unusable; later registration handles compatible shifts and rotations when enough field overlap remains. Blink on unregistered data is not a registration test.

A single satellite trail may be handled by integration rejection when enough suitable frames are available. Severe aircraft flashes, large contaminated areas, or a tiny dataset need more care. Review the eventual stack and rejection maps rather than promising that every trail will disappear.

Keep mild seeing and transparency variations for measurement unless they clearly ruin the exposure. Selecting only the sharpest handful can discard useful total exposure time. The right balance depends on the desired detail, noise, and amount of data available.

Follow the WBPP lesson to load the candidate raw lights and correct calibration frames into a separate review output directory. Inspect the calibration assignments first. Configure the run to stop after calibration and, for raw color-camera data, debayering; disable later registration and integration stages using your version’s pipeline controls.

Open a representative output and verify that calibration and the intended color-camera conversion succeeded. Keep these copies linear and at their original geometry. Do not stretch, sharpen, denoise, resize, or register them before the measurements in this exercise.

Keep a traceable correspondence between each review copy and its original raw file. WBPP may add suffixes or use grouped output folders. Check the report and paths; do not rely on stripping a guessed suffix when filenames repeat.

These products are for inspection. The final workflow below returns to selected raw files for a complete WBPP run, trading some repeated processing for a clear beginner handoff. Do not put calibrated review copies into WBPP as raw lights with calibration enabled.

Find SubframeSelector through All Processes, usually under ImageInspection. Older references may describe a script with different controls. Use the installed process documentation for the current interface.

Compare the distribution within your group. A metric is a clue to inspect, not a complete diagnosis. Measurements depend on detected stars, sampling, and the selected model and settings.

Plot or sort by FWHM, then eccentricity, and inspect isolated outliers in Blink using the same calibrated copies. Look at star count and background changes too. Record the visible reason for each rejection, such as sustained trailing or cloud-obscured stars.

For example, if most comparable frames cluster near 3 pixels FWHM and a few measure near 7 pixels, inspect those few for focus or tracking problems. Those numbers illustrate an outlier pattern; they are not approval thresholds for your equipment.

For a first pass, use the Measurements approval controls to approve or reject individually. If adding an Approval expression later, base it on the measured group and inspect every borderline result. Check its syntax and evaluation in the installed help. Reapplying an expression can change earlier decisions unless they are protected by the version’s lock controls.

Approval decides whether to keep a frame; weighting decides how strongly an included frame contributes. This tutorial uses SubframeSelector for selection and lets WBPP calculate its own integration weights later. Leave custom weighting formulas and output weight keywords out of this first workflow.

Count the accepted frames and their total exposure time. Keep uncertainty visible in the log. If a large fraction is rejected, recheck the grouping, units, calibration, and detection settings before concluding the whole session is poor.

Save the measurements, configuration, and decision log using the available export options. Match each approved calibrated review file to its original raw light. You can also use Output Subframes to produce approved review copies in a new directory, but those remain calibrated copies, not raw WBPP inputs.

Create an Approved_raw folder containing copies of only the accepted original raw lights, preserving session and filter subfolders where needed. Alternatively, add the approved original paths directly to WBPP. Check the counts against the selection log and ensure that no frame appears twice.

Start the final WBPP configuration with that approved raw list and its matching calibration frames. Use a clearly named final output directory, verify calibration and CFA settings again, and enable the intended registration and integration stages. Do not accidentally reuse the unfiltered original light list.

If WBPP’s own Frame Selection stage is also enabled, review how its additional exclusions affect the accepted list. Keep interactive stages intentional. After completion, compare the report’s used-frame counts with your selection log and investigate unexpected losses.

FWHM is implausibly small or many tiny objects are detected: inspect Star Detection Preview and the calibration state. Hot pixels are not a sharp star field. Do not solve this by adopting a stricter FWHM cutoff.

Measurements differ dramatically between filters or sessions: separate comparable groups and check exposure, scale, and processing stage. Avoid a universal ranking across unrelated data.

The lowest-noise exposure is cloudy or lacks stars: inspect the image and other metrics. Smoothness alone is not a measure of useful target information.

Frames look equally bright despite changing sky conditions: review whether individual auto-stretches are masking those differences. Compare a consistent display stretch and the measurements.

Rejected frames return to the run: compare actual input paths and counts in WBPP. A decision in one process does not automatically remove a file from another process’s list.

The final stack still has poor stars: inspect registered lights, reference choice, rejection maps, and acquisition limitations. Frame selection cannot repair systematic optical or tracking problems present in every exposure.

Continue with the WBPP lesson, keeping the final integration linear until the appropriate calibration and processing steps are complete. This is an original practice sequence using the linked guidance. Older sources explain the concepts; current installed help governs controls and metric definitions.