How Realistic is Photorealistic?

S. Lyu and H. Farid

IEEE Transactions on Signal Processing, 53(2):845-850, 2005

Paper (ps.gz)    Paper (pdf)    Bibtex

Errata: the version of this paper appearing in print has several mistakes that are corrected in this reprint. Specifically, in the submitted version of this paper, the LDA and SVM classifiers were trained to yield a false-positive rate of 1% (a photographic image incorrectly classified as CG). After acceptance we changed the classifier to yield a false-negative rate of 1% (a CG image incorrectly classified as photographic). Several places in the text, and Figures 6, 7 and 9, however, were not adjusted accordingly. These errors are corrected in this reprint.

Computer graphics rendering software is capable of generating highly photorealistic images that are often impossible to differentiate from photographic images. As a result, the unique stature of photographs as a definitive recording of events is being diminished (the ease with which digital images can be manipulated is, of course, also contributing to this demise). To this end, we describe a method for differentiating between photorealistic and photographic images. Specifically, we show that a statistical model based on first- and higher-order wavelet statistics reveals subtle but significant differences between photorealistic and photographic images.