This study raises the topic of determining the water content of wall paint based on portrait distance on Nippon Paint paint brands with a water ratio of 1 : 0.75, 1 : 0.5, and 1 : 0. The problem is how to find out the comparison of accuracy levels in the recognition of levels water on a paint wall by applying the ANN recognition method using the GLCM feature and taking 4 different shooting distances. The training data and test data are extracted using the GLCM feature then ANN training is carried out using 17 training functions. The portrait distances used were 25, 50, 75, and 100 cm with the neurons used being 5, 10, and 20. Each architecture was run five times, with three number of neurons multiplied by five attempts to run the program so that a total of 15 trials, and three water comparisons done afterwards. Based on the test results it can be concluded that the air ratio of 1 : 0.75 using neuron 20 in the hidden layer obtained the best results in identifying the air ratio on the paint wall. In calculating the fusion matrix produces an average overall output that is equal to 80% accuracy, 65% precision, and 60% recall.