Sunday, October 28, 2012

How the Old Masters created the look of Gold in Painting

The detail (above) from Van Loo's painting, Marie Leszczinska Queen of France, is fascinating to me. I've been trying to figure out why I think his rendering of the gold table is unsuccessful. Though he's obviously a meticulous craftsman and clearly spent ages with a magnifying glass in one hand and a paintbrush in the other, there's something overwrought about the brushwork. His table looks plastic, but it took me a while to figure out why

Charles de Solier, Hans Holbein the Younger

Holbein was, of course, a master at painting the detailed effects of light and shade on any number of textures. I examined his Portrait of Charles de Solier (above) for clues as to his technique for painting gold.

Local color and value are notoriously hard to read, so I took color samples from the sword handle, (a) through (e) above, and then de-saturated them to get their values. Applying these five values against Munsell's value chart shows some interesting results.

The values are all gathered tightly around the bottom of the spectrum, between Value 1 and 3. There is nothing at all in the mid-tones from 3 all the way to 6, when all of a sudden we have our one and only highlight, (e). What surprised me was just how dark everything was. Even the brightest highlight on gold is only a 6, yet the highlight still jumped a full three values from its closest neighbor.

I knew that medieval painters, when planning to include gilding in a painting (say, on a halo), would map out their value composition as if the gold leaf was a dark element. This initially seems counter-intuitive - Gold isn't dark! - but when we see how dark the overall value scheme of gold is in the Holbein painting we can see why it works.

I tried the same experiment on another sample, this time a much brighter 'gold', taken from Le Brun's Hall of Mirrors painting in Versailles...

Firstly, it's interesting to note the palette swatches taken from the 'gold': They're kind of a dull brown, and not very gold at all. Again, regarding the values, we see that the majority of values are between 2 and 5, and the highlight jumps three values but is still no more than an 8.

Le Brun has used a mid-tone (d) on the circular frame, but the egg-and-dart molding has no (d) tone; it makes the 3-value-jump from dark tones straight to highlight just as in the Holbein painting, and is very successful for it.

Why is it that Van Loo's painting is unsuccessful?

Yikes! The first thing that stands out is the number of colors. Van Loo went a little overboard unnecessarily. The more colors you lay down on a space, the more you're making me stare at that area in order to figure out what's going on. Don't make me focus too much on a background object: the focus should be on the main subject.

Let's look at it more closely. Seen as black-and-white value reductions, the spread across the spectrum is much more evenly spaced than the (better painted) Holbein. Specular highlights on metallic objects are supposed to jump out at us. Van Loo's spectrum looks more like that of diffuse light, not specular.

If you want to represent specular highlights on metallic objects, you need to jump at least three values beyond your mid-tones. The way to do that is not to brighten the highlights, it's to darken everything else. Van Loo had nowhere to go from Value 7.5, as Value 10 is pure white, and as such is a purely theoretical limit.

It's like in Spinal Tap. When you turn the volume all the way, where do you go from there?

Spinal Tap, "This one goes to 11" scene [video link]

Justice Punishing (detail), Noel Coypel
In this detail from Justice Punishing, by Noel Coypel, we can see how effective hatching is at representing highlights on metallic surfaces. The hard jump in value from dark mid-tone to highlight works really well to suggest gold. Van Loo's smooth gradations look too soft by comparison. [Incidentally, Coypel's painting is also an excellent reference for the structure of the acanthus leaf].

David Briggs wrote a very concise description of diffuse and specular light, and explains exactly where we should place specular light reflections. Make sure to read all three pages.

Coving, the Nef (vessel) of Louis XIV (detail), René Antoine Houasse
Houasse's fresco from the Abundance Salon, Versailles, shows (along the bottom) gold embroidered cloth using a similar technique. But this time, instead of hatching, he's used dots to simulate the threads.

Look at these examples from other painters to get an idea of their method. Notice the jump from darks straight to highlight in the Rembrandt details. I love the way he painted light on metal. You might conclude that the bigger the gap in value between shadow and specular highlight, the more successful the illusion.

Rape of Prosperine (detail), Rembrandt van Rijn
Jeremiah Lamenting the Destruction of Jerusalem (Detail), Rembrandt van Rijn 
Portrait of Pope Leo X (detail) , Raffaello Sanzio

Madame de Haussonville, Jean Auguste Dominique Ingres

Ingres' detail shows the exact same phenomenon in the values on the painted gold frame. Even though the frame is in the background and the spread of values is not as great [he has rightly reserved his highest and lowest values for the main subject in the foreground], his specular reflections make the same jump.

Still Life with Silver Jug (detail), Willem Kalf


  1. This is a wonderful post. Thank you so much for sharing this information. Carolina Elizabeth

  2. What a fantastic study! I have a file with several "formulas" in it from studies I have made, but nothing on this level. Anyone who can tie in Spinal Tap is a very cool dude to me!!

  3. Oh my goodness... How did I not know about your blog?? This is fantastic!! Signing up now...

  4. You outdid yourself on this one, Alan!
    Wonderful and thoughtful.

  5. Alan, Has anyone remembered to tell you lately how brilliant you are? Extremely talented yet also witty, humorous and marvelously irreverent. Thank you for yet another amazing post. Educational yet with a 'dash' of Christopher Guest thrown in.

    Thank you, Robert D

  6. The Willem Kalf is my favorite. I kinda get excited when its time to paint the the whole thing comes to life. Great article.

  7. Che bella ricerca! Grazie Alan.Angela

  8. Excellent post, Alan. Thank you for such a thorough analysis that truly makes sense. And the Spinal Tap example was superb!

    Thank you. It's always an educational pleasure to read your blog.


  9. Alan, another brilliant and valuable study from someone who should consider writing his own book on the definitive trompe l'oeil.

  10. An Argyle wall! A canvas print such as this one of the Isle of Skye, painted by an American painter of the late 19th
    century, William Trost Richards, may take you right back to the wilds of the home country and go well with your Scottish
    wall. It can be ordered from

  11. I signed up and this gem came to my inbox. Thank you for imparting such amazing knowledge.

  12. That's very amazing post,I love this thank you so much.

  13. Alan a very detailed study of highlights and golds.


  14. Wow, this is a really beautiful and useful post, I feel like I understand something new, a sparkle light up in my mind, with darker tones and a specular highlight on it !
    Thank you very much!

  15. It's important to note that Rembrant used a tinted semi-transparent glaze over lead white paint mixed with an egg emulsion to achieve those warm glowing highlights. There are variations of this recipe but I've used the following: emulsion: 1/3 damar varnish, 1/3 of the following mixture (1/2 egg shell of water, 1/2 linseed oil), 1/3 wood turpentine. The varnish mixture is 1/3 varnish, 1/3 linseed oil, 1/3 turp.

  16. Forgot the egg! for the emulsion recipe include one egg yolk - very important to bind the water and oil.

  17. Your focus on value is of course the first step and well stated but I feel there is a missing component in the color aspect. What gives the gold the shimmering reflective quality is the variation of green and reds, more specific the red brown and green browns. Being complimentary these colors vibrate off each when neighboring each other and intensify the effect of the light refraction. Also the specular highlight has a similar effect in that core of the highlight and it halo has a warm and cool relationship also intensifying the light vibration. Hair also demonstrates these properties especially blonde. Flesh has these red and green vibrations but I like to think of them more in terms of pink and green.

  18. Faux paintings are definitely the best options of wallpapers.

  19. I love all the contents of your website

  20. WOW... so glad I stumbled into this somehow. Thank you. gonna share too.

  21. "still jumped a full three values from it's closest neighbor."

    Its. You meant "its".

  22. i dont think van loos painting is "unsuccessful"

  23. Very nice article. I'd say that Van Loo's work is wonderfully successful. If you look at gilt furniture of the period, you'll see that he's captured the variety of finishes very well indeed. The gilder's art in those days was practically like the painter's: matte and gloss and burnished, white gold, red gold, yellow gold--lots of colors and lots of finishes, all of which are accurately represented there.

  24. Interesting point, and well taken. The possibility of different types of gold is something I hadn't considered and might easily explain the range of colors. Your second point about Van Loo's gold possibly being dull in luster is also a good one, and could explain the lack of specular highlights.

    If Van Loo was going for strict mimetic reproduction of exactly what he saw in front of him, and assuming that your observations of actual gilded furniture of the period are correct, then yes; his painting is successful.

    However, I still feel that - speaking only of value - the specular highlights should be two to three Munsell value points away from their nearest neighbor. I realize that this is just a painterly contrivance, but I think it creates a more illusionistic and "believable" gold than Van Loo's.

  25. While the article is good and interesting to read and the core understanding is correct, I think that some of the understandings and painting analysis are oversimplified and made fit to your belief. Sure, if you take 17 color samples from one painting and 5 from another painting you will get a more smooth and even distribution on the munsell value chart.
    Painting successful gold is not just dependent on the value, but in my opinion also on the placement and size of the highlights, play with reflections and color.