Multi-Night Projects: Combining Data Across Sessions

There is a hard ceiling on what one night can give you. The target is only high in the sky for a few hours, the moon is only absent for part of the month, and at some point you need to sleep. Once you have hit that ceiling the only remaining lever is to add nights.

This is the technique that separates images that look competent from images that look serious, and it requires no additional equipment whatsoever.

Why it works better than a single long night

Three reasons, and the third is the one people miss.

Every frame is taken at high altitude. If you image a target only during the two hours around its meridian crossing, every frame passes through the least atmosphere available. Extend a single session to five hours and the last two are taken through thickening air near the horizon, which means dimmer, smeared, more absorbed light. Four nights of two good hours beats one night of eight mediocre ones.

Conditions vary, and averaging helps. One night has slightly better transparency, another slightly worse seeing. Combining them dilutes the effect of any single bad night rather than letting it define your result.

Satellites and aircraft average out. A satellite trail through one frame of a two hundred frame stack is rejected automatically. The same trail in a stack of twenty is much harder to remove cleanly. More total frames means better outlier rejection.^[1]

What has to stay the same

For frames from different nights to combine cleanly, some things must be consistent and some genuinely do not matter.

Must match: the instrument, obviously; the filter, absolutely; and the sub-exposure length, ideally. Mixing filtered and unfiltered data on the same target produces colour that fights itself and is far more trouble to reconcile than it is worth.

Does not need to match: the exact framing, within reason. Stacking software aligns on stars, so a modest rotation or offset between nights is handled automatically at the cost of a small crop at the edges. Nor does the sky need to be equally dark each night, though the darker nights will contribute more useful signal.

How to run one in practice

1. Pick the target from the calendar, not the mood. Choose something that crosses the meridian at a civilised hour for the next fortnight, and that suits the moon phase you are heading into.
2. Fix your settings on night one and write them down. Filter, sub-exposure length, gain. Changing them mid-project is the most common way a multi-night set becomes hard to combine.
3. Keep the raw frames. This is the part most smart-telescope owners skip. If your instrument only saves the finished stack, you can still combine those stacks, but keeping the individual frames gives you far more control later.
4. Label the folders by date. Trivially obvious and universally ignored until the first time you cannot remember which set was the good night.
5. Stop when the target sinks. Not when you get tired. The last hour below 30° elevation adds less than it costs.

A realistic first project

Pick the Rosette, the Heart, or M81 and M82. Give each ninety minutes a night across three or four clear nights within the same dark window. Total: four to six hours.

Compare the finished result against your best single-night attempt on the same object. The difference is usually larger than any equipment upgrade you could have bought for the same effort, which is a slightly deflating thing to learn and an extremely useful one.

Notes & sources

1. Outlier rejection during stacking and why larger frame counts improve removal of satellite trails and transient artefacts. Siril documentation, stacking and rejection algorithms ↩