What a notan is, and what it is for
Reduce a picture to two values, black and white with nothing between them, and what remains is the arrangement of light and dark. That is a notan, and the Japanese design tradition it comes from treats it as the skeleton of the composition rather than a study of it.
The test it performs is brutal and useful. If the two value version reads as a clear arrangement of shapes, the picture has a structure that will survive being painted. If it collapses into a grey speckle or into one large mass with a hole in it, there is nothing underneath the picture to hold a painting up, and adding colour will not supply one.
Painters have done this with a brush and one tube of paint for a very long time. The only thing that has changed is that you can now see it instantly and change your mind about the crop while you look.
How to read the result
Three failures show up immediately.
The first is a picture with no dominant mass. If light and dark are close to evenly divided and scattered, the eye has nowhere to rest and the composition will feel restless no matter how well it is painted. The fix is usually a crop, and occasionally a decision to make one area darker than the photograph did.
The second is a subject that disappears. A face at the same value as the background behind it vanishes entirely in two values, which tells you it will need a value change invented for it. That is a legitimate decision, and it is much better made deliberately at this stage than discovered halfway through an afternoon.
The third is speckle. If the whole frame breaks into small alternating patches, the picture is texture rather than structure. It may still be paintable, but it is not going to be held together by its design and you should know that going in.
A notan that looks wrong is information rather than a verdict. Plenty of good paintings have been made from photographs that failed this test, by changing the values on purpose.
From two values upwards
Once the two value version holds, go to three or four rather than to colour. Three values give you light, middle and shadow, which is enough to place every mass in a painting and is the level most block ins are actually working at.
Between three and seven is the working range. Below three there is not enough information to place a brush. Above seven the study stops being a plan and starts being a picture of the photograph, which returns you to the problem you opened the photograph to solve.
Pacyra runs value studies from two steps up to 45. The high end is for reading a subject rather than painting from it, and it is not the setting to print and tape to an easel.
Why value comes before colour, always
Value carries the form. Colour carries the mood. A painting with correct values and eccentric colour reads as a painting with a point of view. A painting with beautiful colour and muddled values reads as a mistake, and viewers who could not name the problem will still feel it.
This is also why the engine settles value structure before it assigns colour families. Two colours that read as the same value get the same number even where their hues differ, because on the canvas they are doing the same job.
In practice the discipline is simple to state and hard to keep. Squint at the picture until colour drops out and only masses remain. If it works squinted, it will work painted.