Atrial fibrillation (AF), the most common arrhythmia, can be diagnosed from altered morphology of P-waves in patient's ECG. However, current analysis methods cannot identify specific type and location of electrical drivers for AF from ECG. This study aims to evaluate contributions of electrical activity in the right (RA) and left atria (LA) to the P-wave in both healthy individuals and AF patients. By comparing the impact of electrical excitation patterns in RA and LA on P-wave morphologies across each lead of ECG, this investigation aimed to provide mechanistic insights into the links between the P-wave morphology and the underlying type and location of the electrical drivers. Five 3D torso computational models generated from high-resolution CT patient scans were used for in-silico simulations of bi-atrial electrical activity and resulting 12-lead ECG. These models included the entire heart, blood pools and major organs including bones, lungs, liver, and kidneys. Electrical activity in sinus rhythm (SR) and AF was simulated using the Courtemanche ionic model and its AF variant. The simulations were conducted for control and AF cases involving RA-only, LA-only and bi-atrial scenarios, resulting in a total of six scenarios per patient. Body surface potentials were recorded in standard 12-lead ECG positions using bidomain equations and finite element methods. The distinct contributions of each atrium and the impact of conduction velocity (which was lower in AF vs. SR) on P-wave morphology were investigated by comparing signal metrics for each of the twelve leads. The study reveals varying ratios of RA versus LA contributions to the P-wave morphology across ECG leads, confirming previously published results that RA exerts a greater influence on the ECG signal than LA. Additionally, differences in conduction velocity between control and AF cases resulted in longer atrial activation times during AF, which was reflected in the prolonged P-wave duration.