Inpainting: Tips

Back to the Paintings Main Page

Back to Inpainting Chapter Index

The information presented on the Paintings Conservation Wiki is the opinion of the contributors and does not imply endorsement or approval, or recommendation of any treatments, methods, or techniques described.

Compiler: Catherine A. Metzger

Egg-Tempera for Retouching Cracks in Old Master Paintings

As Chief Restorer at the Metropolitan Museum of Art in New York, John Brealey introduced to his many interns and fellows the following formulation of dammar/egg tempera. It was used primarily in the retouching of cracks, not to inpaint the crack away but only to reduce the visual distraction of dark cracks on lighter areas, mainly in skies or flesh tones. A principle advantage is that the cracks can be covered quickly. The thin paint is pulled into the crack by capillary action and any “overpaint” can be buffed off and away from a varnished surface with cotton after it is dry.

Recipe:

Cleared egg yolk: 1 part

Thick dammar: 1 part

Water: 1 part

Put the egg yolk into a small-necked jar. Slowly add dammar in small amounts, shaking vigorously. After dammar, slowly add water in small amounts, again shaking vigorously to create an emulsion-like solution.

This medium is then mixed on the palette with pigments previously ground in water. In skies, raw sienna and titanium white mixtures were most commonly used.

On a painting with a sufficiently dry varnish, cracks can be painted over quickly. The paint will be drawn into the cracks, and any excess can be buffed off the surface on drying.

Laurent Sozzani

Submitted June 2004

Pastel Used to Match Oil/Wax Emulsion

A system was developed by trial and error to successfully retouch a badly scratched and scuffed painting by Arthur Dove. The medium was one of his oil and wax emulsions on a painting dating from 1937. The surface had a velvety, soft surface quality to it. Pastel was mixed on a piece of mat board to match the desired color, in the process creating a small mound of medium to use as the dry pigments for retouching. The pastel mixture was then applied to the affected areas with a brush using Jim Bernstein’s “dry water” (½ water:½ acetone). The combination of a fast-drying diluent with a chalk/pigment medium provided a near perfect match for Dove’s uniquely matte paint film.

Lilli Steele

Submitted June 2004

Pastel Used to Hide Stains in Unpainted Canvas

It isn’t always possible or advisable to remove discreet marks and stains from unpainted canvas, especially when dealing with color field paintings on cotton canvas from the 1950s and 60s. Disrupting the delicate surface structure of the fabric through the application of local treatments, both wet and dry, can quickly lead to unintended problems that may be difficult or impossible to correct.

In some instances, when treating these paintings the author has resorted to toning marks, stains, and abrasions using pastels with good results. The pastels can be applied using a paper stump and adjusted using a sable brush.

Like the canvas, pastels have a nonreflective, matte appearance and are easily reversed with a soft brush and/or vacuum suction should the need arise. The primary downside is that you are applying a material on top of the canvas to simulate a color that comes from within the canvas, and the inpainting will always be evident when the painting is viewed from an extreme angle.

The soft pastels that the author uses are manufactured by Rembrandt and Sennelier. As a class of materials, pastels can be less than lightfast when compared with other artists’ materials, but he has used five colors that presumably contain very little if any fugitive colorants (raw sienna 234.10, yellow ochre 227.9, and yellow ochre 227.10 from Rembrandt, and yellow ochre 116 and brown ochre 126 from Sennelier).

Jay Krueger

Submitted April 2010

Synopsis of Article on Selecting Pigments for Inpainting

Article title: “Multiple Pigment Selections for Inpainting Using Visible Reflectance Spectrophotometry.” In Studies in Conservation 47(2002): 46–61. By Roy S. Berns, Jay Krueger, and Michael Swicklik

When inpainting a painting, it is very difficult to attain a color match that does not change with the viewing conditions. Ambient lighting conditions with color temperature shifts in addition to changes in the viewing angle of the observer often cause the inpainted areas to shift in color, a phenomenon known as metamerism. This article shows how to utilize a small aperture spectrophotometer and spreadsheet software to select pigments that will minimize color shifts.

The method relies on a single constant simplification of the Kubelka-Munk turbid media theory and on restricting spectral analysis to 420nm and above in order to develop an easily usable database of pigments. By taking a spectral measurement of an area adjacent to the paint loss to be inpainted and subjecting the result to a multiple linear regression, the chosen spreadsheet software selects a set of pigments that potentially shows the least metamerism. The article outlines the theory of this methodology in depth and also provides a step-by-step description of how it should be employed for inpainting applications. The authors cite examples of two successful conservation treatments—for Barnett Newman’s Dionysius and Sanford Robinson Gifford’s The Desert at Siout Egypt—to illustrate the method.

Michael Swicklik

Submitted November 2008

Image Projection as an Aid to Reconstruction

A photographic or other reproduction that documents a former, better preserved state of a painting can be used to support reconstructive retouching of altered or lost features. The two techniques described below will maximize the accuracy of placement, proportions, and contours of reconstructed elements through the use of the reference image projected onto the work of art. These comments will not address philosophical or ethical issues surrounding degrees or styles of visual compensation, but will be restricted to basic practical matters.

Projection of the reference image allows the fullest application of the information it contains to the task of reconstruction, as well as the best possible reintegration of what might otherwise be completely decontextualized fragments of surviving original paint. Two points should be observed if the reconstruction is to take best advantage of the reference documentation:

The reliability of the reference image must be investigated as thoroughly as possible. Special care should be taken to consider the accuracy of the image’s tonal and/or chromatic rendering of the painting; specific distortions or deficiencies may require attention in interpreting the image.

The success of all projection-guided reconstruction depends on accurate registration between the reference image and the surface of the work of art. Reliable registration can be achieved by aligning at least three widely separated features. This usually entails considerable trial and error due to various possible sources of geometric distortion in the original photography and in the alignment of the projection device relative to the work of art. Perfect registration can be elusive, but the effort to achieve it pays off in superior results.

Technique One: Direct Projection

Equipment and Setup

For direct projection of the image onto the work of art using a slide-, overhead-, or digital projector, the projector should have the capacity to adjust image size as well as focus. An easel with smoothly operating mechanisms and lockable casters is important. The projector is set up at a suitable distance to produce an actual-size image of the painting or a part of the painting. The best possible registration may involve adjustments to the position of the projector, the painting on the easel, or the easel itself. It has proven helpful to start by having the surface of the painting plumb and perpendicular to the beam of the projector. The projector should be level, and the center of the projected image should correspond to the midpoint of the corresponding portion of the painting. It may be helpful to have shims or shallow wedges on hand. The registration and focus of the projected image are best checked with a piece of white paper held to the surface of the painting. Once the projected image is congruent with the painting, it is a good idea to unplug the projector rather than turn it off with the switch to avoid shifting it out of alignment. The light level in the workspace should be adjusted for an optimum balance between the visibility of the projected image and the painting itself. Because this method requires working with reduced visibility, it is primarily useful for ensuring accuracy of drawing, specifically the placement of contours and reference marks to be followed while retouching afterward under standard lighting conditions for retouching. It is preferable to apply any such guide drawing in a medium that can be removed later in the reconstruction process without disrupting the retouching or varnish surface.

Technique Two: Stereomicroscope Projection

Equipment and Setup

This technique was developed for reconstruction of fine details. It superimposes a photographic image onto the painting surface using a stereomicroscope. The microscope must have a continuously variable zoom and individually focusable eyepieces. Although it is possible that a number of models and makes of microscope could be used or adapted for the technique, it was developed using a Wild-Heerbrugg M8 with 10X oculars. The particular configuration of the M8 allows a disc of film cut from a detail slide of the source photograph to be placed, as described below, at a location in one ocular tube such that the photographic image and the surface of the painting can be viewed simultaneously in perfect focus. To compensate for the generally lower level of light reaching the eye through the photographic image, it is necessary to place a neutral density filter in the other ocular tube to obtain a convincingly merged stereo image. The best density for this filter can be chosen from a selection made from bracketed, out-of-focus transparencies of a neutral gray subject.

Scale and Preparation of the Photographic Image

It is desirable to make reference to and work on as large an area as possible, which means working in the range just above the microscope’s lowest magnification. The lowest magnification on the M8 stereomicroscope used, 6X, gives a field of view of about 1-⅝” diameter and at 9X, a field of view of about ⅞” diameter. At any given microscope magnification, the actual image size on the slides made from the source photograph must fall within a specific range. In our experience, the most useful ratio of the slide image has been 1/2 to 3/4 life size.

Any general reference on photography can provide guides for figuring magnification ratios, but here is a practical tip for achieving ratios in this range: the 35mm format seen in a camera viewfinder represents an area 25mm x 35mm (or roughly 1”x 1-½”.) A 50mm (about 2”)-long feature on the painting that fills the short (25mm/about 1”) dimension of the format will appear at 1/2 life size on the film; if that same 50mm-long feature is made to fill the long (35mm/about 2”) dimension of the format, it will be ¾ life size on the film. Note that the size of a feature on the source photo itself does not matter, just the relationship between the size of the feature on the slide image made from it and the actual size of the feature as measured on the surface of the painting.

Fig. 1: Page of Overlapping Slide Views

Fig. 2: Cutting of Film Discs

Fig. 3: Attachment of Film Disks to Microscope Eyepiece

Obviously, the reconstruction of an area bigger than the field of view of the microscope requires making a number of overlapping slide views (see fig.1). It is also useful to bracket exposures and select what provides the most information in any particular area.

To make the photographic film disc that goes in the microscope, the slide film is removed from its mount and placed on a light box. An eyepiece is removed from the microscope ocular tube and the short section of knurled, threaded tube at the end opposite the lens is unscrewed and removed. This piece is placed with its narrow end down on the slide film to select the area of interest. The circular shape of the tube’s exterior is scored on the film with a sharp needle and scissors are used to cut out the disc, which is then affixed to the narrow end of the tube with small tabs of clear adhesive tape (see fig. 2). The short threaded tube, with film disc attached, is then screwed back into the eyepiece, which is then inserted back into the microscope ocular tube (see fig. 3). A disc cut from the neutral density filter slide can be affixed to the other eyepiece in the same way. Either eyepiece can be used for the reference image, although a survey of a few colleagues has suggested there may be an advantage to pairing the photographic image with one’s dominant eye.

Focus and Registration

The painting is positioned under the microscope in the most convenient orientation for working; the image-containing eyepiece can be rotated within the ocular tube to match this orientation. To begin, the photographic image is focused using just the eyepiece diopter focus for that eyepiece; it helps to have an illuminated sheet of white paper under the microscope. The white paper is removed and the painting surface is brought into focus. If the painting surface does not have distinct enough features to allow focus to be judged clearly through the photographic image, a piece of white paper with some dark lines drawn on it can be placed on the paint surface as a focusing aid. Once the photographic image and the painting surface are in focus in one ocular, the other ocular can be focused. Next, the photographic image and the paint surface are brought into registration by using the microscope zoom to increase or decrease the magnification of the painting to match the scale of the photograph. Registration is confirmed by matching up at least three widely separated features visible in both the photographic image and on the painting (brushstrokes, cracks, losses, or the like.)

With everything in place, the image and paint surface in the closest possible registration, and with a little practice, it becomes possible to visualize the photographic features as resting on the paint surface, at which point they can be followed to reconstruct what is missing. Checking the progress and quality of the retouching while working is a simple matter of looking through the neutral density filter eyepiece only. Checking something on the photographic image only is easily done by slipping a piece of white paper under the microscope. In practice, the work may involve much looking through the “photograph” eyepiece only, in which case an eye patch can be worn on the other eye to eliminate the discomfort of keeping it closed for prolonged periods. The color and overall appearance of the reconstruction can be checked and adjusted under normal viewing circumstances, away from the microscope, as frequently as necessary.

When reconstructing a larger form that also contains fine detail, it is best to alternate between the direct and microscope projection methods, combining the particular advantages of each. Also, it is worth noting that the projection techniques described here could be equally useful for some types of objects other than paintings.

Images

Gerard David, Lamentation, John G. Johnson Collection, Philadelphia Museum of Art, Cat. 328

Figure 4 Detail from c. 1900-1910 photograph. Rocks at base of Cross and middleground tree

Figure 5 After 1930s treatment. Same view with rocks and tree removed in 1930s cleaning

Figure 6 Before inpainting, recent conservation treatment

Figure 7 Same view of painting with rocks and tree restored using directly projected photographic image

Stereomicroscope projection for reconstruction of fine details: Jacometto Veneziano, Portrait of a Lady, John G. Johnson Collection, Philadelphia Museum of Art, Cat. 243

Figure 8 Detail from c. 1900-10 photo

Figure 9 Same view of painting showing loss of painted details (strands of hair, eyebrows) in 1930s cleaning

Figure 10 Same view of lost details restored using stereomicroscope projection

Simon Marmion, St. Jerome and a Cardinal Praying, John G. Johnson Collection, Philadelphia Museum of Art, Inv.1329

Figure 11 Detail of archival photo (St. Jerome's beard)

Figure 12 Same view with large portion of beard removed in 1930s cleaning

Figure 13 Same view of beard reconstructed using a combination of stereomicroscope and direct projection

Mark Tucker

Submitted May 2004

Simulating Cracks in Old Master Paintings

In The Cleaning of Paintings, Helmut Ruhemann named three methods to simulate paint cracks in Old Master paintings: drawn, painted, or scratched (p. 184). The following tips, gathered as part of a limited and informal survey, suggest that conservators rely on these same techniques today.

Drawn cracks are made with a sharp pencil on top of the retouching.

Preference is divided between hard and soft pencils.

Painted cracks are made using a finely pointed brush and black paint.

The use of watercolor allows working back in with resin-based paints or varnishes without blurring the crack (Frank Zuccari, personal communication, 2010). Ox gall may be added to the watercolor to reduce surface tension.

When using watercolor, the cracks can be narrowed and sharpened further by brushing the edge with a clean, moist watercolor brush (Charlotte Hale, personal communication, 2010) or a microspatula (Dianne Modestini, personal communication, 2010).

Scratched cracks are made using a needle or needle-tool to incise lines into the fill or retouching.

Cracks may be scratched into a layer of Plaka. Used as a fill material, Plaka is easily carved and the colors are opaque. But, since Plaka must be removed mechanically, it should be applied only as a thin layer on top of a gesso fill that itself covers areas of complete loss in the painting

When a dark-colored fill is used beneath the retouching, the scratched cracks will be dark (Knut Nicolaus, The Restoration of Paintings, 1999, page 299; as noted by Joyce Hill Stoner in her book review that appeared in the January 2000 issue of the WAAC Newsletter).

Scratching may be useful if the cracks are wide (David Bull, personal communication, 2010).

Cracks are more often brownish, rather than black. Care should be taken to simulate fewer cracks than one might think necessary, to avoid the area taking on a life of its own. The underlying retouching should be kept slightly lighter, cooler, and brighter since the addition of cracks will make it darker, warmer, and dirtier (Dianne Modestini, personal communication, 2010).

Elizabeth Walmsley

Submitted December 2010