If you learned band combinations on Sentinel-2 and then opened a Landsat scene, you probably produced a strange-looking image. The two satellites number their bands differently, so a recipe that works on one fails on the other. Landsat 8 and 9 carry 11 bands, and knowing which ones to combine is the difference between a useful image and a confusing one.

For a natural-looking Landsat image, assign band 4 (red) to red, band 3 (green) to green, and band 2 (blue) to blue. The result looks like an aerial photograph at 30 metre resolution, with the option of sharpening it to 15 metres using band 8, the panchromatic band, which records a broad slice of visible light at double resolution. Pan-sharpening is a standard QGIS operation and is worth learning once you are comfortable with basic composites.

True colour on Landsat is slightly less crisp than Sentinel-2's 10 metre view, but Landsat's archive goes back to 1972, which no other free dataset matches. For change detection over decades, a true-colour Landsat view from the 1980s next to one from today is an irreplaceable pair of images.

Landsat and Sentinel-2 are stronger together than either is alone. Landsat gives you history back to the 1970s, Sentinel-2 gives you detail and frequent revisits from 2015 onward, and their bands are similar enough that indices like NDVI can be computed across both, with care. The standard approach is to build a harmonised time series, matching bands, adjusting for resolution differences, and documenting the cross-calibration so a reader knows the series is not a single instrument's record.

Watch out for the resolution gap: a 30-metre Landsat pixel and a 10-metre Sentinel-2 pixel do not describe the same patch of ground, so fine features will appear and disappear at the sensor switch, not because the landscape changed. For regional trends this rarely matters, but for field-scale analysis it can create false change signals. Aggregate to a common resolution, usually 30 metres, before comparing across sensors, and the combined archive becomes one of the most powerful free datasets in remote sensing.