Refreshing Mona Lisa: conclusion

Colour adjusted Mona Lisa with lips tinted

"A portrait is a painting with something wrong with the mouth"

John Singer Sargent

It was with trepidation that I attempted to digitally restore the lip colour to Mona Lisa. It is one thing to make global changes to a photograph of a painting, but changing a part can never be completely accurate. I made many attempts and was not satisfied with any of them. This one was less unsatisfactory than the rest. With what must be an imperfect task, you can, at least, make sure that steps that you do have some control over are done properly. I am fairly sure that da Vinci would have used rose madder in a glaze for the lips, So I went to Winsor and Newton's website and found this example of the colour which also shows the white paint tube. I copied the photo and then adjusted the colour from the tube near where I wanted to sample a pixel of a deep tint to a reference white. Having got the colour, I made an overlay for the lips at 30% transparency and mimicked the glaze at the edges by anti- aliasing them to 5 pixels. As this overlay changed the value from the source image by 2% on a specific pixel I sampled, I then made that adjustment and the image was finished. I used ColorPilot and Fauve Matisse for this work.

Rubia tinctorum, Common Madder.  photo: H. Zell

Eastlake, in the quote I gave yesterday, mentioned da Vinci's use of lakes and black in painting flesh. The Old Masters would use rose madder lake for tinting lips and the blush on a cheek. A lake is actually from a dye rather than a pure pigment and is precipitated on (usually) a metallic salt to create a pigment. Genuine rose madder is a very beautiful colour, transparent, and of the colour of a Burmese ruby, it looks like liquid ruby as it comes out of the tube. It has only two drawbacks: it costs £29.20 for two and a half tablespoons and it is fugitive (fades in light). The latter is why you do not see it in the flesh tones and lips of the Mona Lisa. The lips are delineated, somewhat, so can be colour adjusted to an approximate level of accuracy, but the size and shape of any blush to the cheeks could only be guessed. Today, we have a permanent, synthetic, Alizarin red which is a lightfast substitute for rose madder.

"Mona", a fantasy of a preparatory Mona Lisa drawing.
by Carin Alizarin Perron (Garth Wright collection)

My late wife was fascinated by da Vinci's Mona Lisa and had planned to paint several copies of it. When she changed her surname to her maternal name of Perron (Daniel Perron was a direct ancestor and a seventeenth century colonist in Quebec), she also changed her middle name from Anne to Alizarin (after the pigment). Her nom de plume in her religious writing was Jesse Ancona which was derived from the Italian for a gessoed panel. She had made  several panels in preparation for her Mona Lisa copies and that included making the authentic final gesso ground for the panels from plaster which she slaked (rotted) herself. This (then inert) plaster would be mixed with rabbit's skin glue (she was allergic to cow-hide glue) to make the final gessoed panels  From the coarse gessso grosso to the silky gesso sottile , she followed Cennino Cennini's method. I used to help her in the production of the gesso.
Sadly, she did not live to do any of the copies.


A few years ago, there was an elaborate digital method devised to restore the Mona Lisa's colours. The image on that page and the larger one on their website might be a result of an over-eager web page graphic artist and not representative of the actual colours produced, but the colours are clearly wrong: you  can see that there is a strong purple tint to the flesh tones and the colours seem overly saturated. I am also skeptical of the method. The large number of layers in the glazing technique with each one affecting what lies below it, and the fact of various changes to the pigments over the centuries including the very fugitive rose madder, makes, I think, any automated method suspect. It is more important to take into consideration da Vinci's methods and skills along with the fact that he worked on this painting, obsessively, for years. I find it hard to believe that (if the graphic colours are even approximately correct) he would have used so much blue tinting over the entire painting to give such a purple tone to the flesh colours. To me, it makes it look like a poor photograph of a copy.

Compare the results with the copy of the Mona Lisa in the Prado (on the left) and then compare the Louvre Virgin of the rocks with the London version.. For many years, the London version was thought to have been a copy and that d Vinci only helped with the hands of the Virgin. Now it is considered (by the National Gallery where it resides) to be an earlier version entirely by da Vinci. The clumsiness of the infant's hands makes me skeptical of that too. There is something sublime about less controversial paintings like the Mona Lisa and the Louvre Virgin of the rocks that show no trace of the the more illustration art-like copies and overly saturated restorations. With everything, I think that taking flesh colours as paramount can give us the closest approximations of the original, pre-varnished paintings.

Tomorrow, another development in the teaching of evolution in U.S. schools


John's Coydog Community page