Image Count for a Block

Two divisions and two additions of one. The frames along a line are $\lfloor L/B \rfloor + 1$, because a line of length $L$ needs an exposure at its start as well as one at every base along it. The lines across a block are $\lfloor W/S \rfloor + 1$ for exactly the same reason. Multiply and you have the frame count.

The plus one is small on a long line and dominant on a short one, and that asymmetry is why narrow sites are so inefficient per hectare. A 400 m line at a 12.5 m base takes 32 frames, where the plus one is 3% of the total; a 40 m line takes 5, where it is 20%. The same effect at the block level makes a long thin site cost far more per hectare than a square one of equal area, which is worth saying out loud before quoting a corridor job at a per-hectare rate.

What this count leaves out is not small. Add two to four frames at each end of every line, so the exposures covering your actual area of interest are not the block's own edge. Add a line outside each edge of the area, for the same reason. Add the frames flown during turns, which are usually discarded but are still shutter actuations and card space. Add a crossed set at right angles if the site has structures or if the camera is being self-calibrated. Realistically the flyable number is 20 to 30% above what this returns, and a plan built on the bare figure runs out of battery before it runs out of block.

Then look at what the count costs downstream, because that is where a mapping job actually gets expensive. At roughly 12 MB a frame, a thousand images is about 12 GB of raw imagery before any product is generated. Feature extraction is close to linear in the count, but the pairwise matching that follows is not — it grows faster, and dense reconstruction grows faster again, so a block that doubles takes rather more than twice as long to process. On a large job the processing hours can exceed the flying hours by a wide margin, and it is the part clients never see and never budget for.

The lever that drives all of this hardest is the ground sample distance, and the relationship is quadratic. Halving the GSD means halving the flying height, which halves both footprint edges, which halves both the air base and the line spacing — so the frame count goes up FOURFOLD. Twice the resolution is four times the images, four times the storage, more than four times the processing, and four times the battery swaps. That is the calculation to put in front of anyone who asks for "the best resolution you can do", because it usually changes the answer.

One last practical note: this relation assumes a rectangular block. Real sites are polygons, and the honest way to handle one is to get its true area and its bounding dimensions first, then plan lines along the polygon's LONGEST axis so that the lines are as long as possible and the turns as few as possible. Orienting the flight lines with the long axis of an awkward site is often worth more than any amount of tuning the overlaps.