Speckle photography is a very accurate method of experimental strain analysis. Because specklegramm is not a counterfeitable document of the state of the analyzed object, the method can be used to test the most important elements. The analysis of the specklegramm must, however, be free of systematic errors. In the paper the sources of errors in point by point analysis are quantitatively described and some ideas of elimination are presented. The algorithms are tested on specklegramms recorded in laser light or in white light.