Getting portraits right is a skill outside the reach of the beginner. Or is it ?
One of the crucial factors in drawing a portrait that resembles the model is to get the exact proportions. It takes considerable talent to get these proportions right when sketching from real life (as one has to rescale and project from 3d onto 2d), and even from a photo this is a particularly frustrating endeavour.
But why do we always want to draw with the photo sitting next to our drawing and not draw directly on the photo? I guess that’s just historical technological constraints turned into ill-placed dogma. After all, if classical painters used grids and other contraptions to get their proportions right, why should we stop short of what the digital era allows us?
Placing the original photo in a separate layer below the sketch layer is exactly the same as drawing on a light box (if you do not have a graphics tablet, a light box is certainly cheaper – or you can tape the photo and your drawing to a window). Now this will not teach you where to put your pencil strokes, but at least the mouth will be in the right place, with the proper angle etc. The higher the resolution of the original picture, the best it is, even for a rough sketch, as there are always fine subtleties than you want to zoom into in order to get the pencil stroke just about right.
Once the sketch is complete, you can either carry on in the digital world, or print out your sketch as many times as you need to practice painting on it.
The portrait below was painted with this technique. I still have a long way to go to avoid the temptation of photorealism when painting from a photo, but I can tell you the model approved the final result.