The example you provided works by measuring capacitance between the foil plate and the "ground". "Ground" in this case is whatever the arduino GND net is tied to. It can be protective earth and therefore the size of electrical installation in the building, or it could be a battery pack and therefore quite small.
Hand or other conductive object in proximity changes the capacitance because it adds parallel capacitance to the foil-ground capacitance that is already there. The added capacitance is either
- human-ground capacitance if hand is touching the plate, or
- plate-human and human-ground capacitances in series if hand is in proximity, but not touching.
What are the limits to the size I can make this foil area?
How would you make a 70cm x 100cm foil sheet good enough enough to detect if you're touching an area of it?
Capacitance when there is no human in presence depends on the plate area. Capacitance with human depends on human's area. Measurements are going to be noisy so there will be a lower limit of value change that can be classified as presence that is proportional to plate area. For a very crude example, if you have a 2 square meter plate and a 1-year-old with ~0,5 square meter skin area touches it, you might have 25% increase in capacitance. Is that reliably detectable? It depends.
I did some experiments with 50x45 cm foil sheet. All measurements at 10 kHz. Me / child.
- Sheet capacitance to protective earth pin: 15 pF, 12k ESR / 15 pF, 12k ESR
- mouse mat and hand on it: 28 pF, 18 k ESR / n.a.
- 80 g paper sheet and hand: 36 pF, 20 k ESR / 21 pF, 15 k ESR
- same but with two hands: same / same
- hand directly on foil (no insulator): 38 pF, 20 k ESR. / 26 pF, 20 k ESR.
What about human-foil capacitance then? The one in series when there is an insulator?
- foil-paper-hand: 1150 pF, 3 k ESR / 490-700 pF, 7 k ESR.