Industries / Consumer goods / Furniture
Finishing
Sanding wood smooth, then coloring and sealing it with stains, lacquers, oils or paint so that it looks right and resists wear, water and stains.
5
Milestones achieved, from 1920s to 1999.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1920s
Lacquer replaces shellac
Furniture makers in the early 1920s finished with shellac, which dries fast. DuPont developed the first commercial nitrocellulose lacquer, made from the cellulose nitrate used for explosives and film, and General Motors used it under the Duco name on almost all its cars by 1924. Lacquer dries only by its solvent evaporating, so fast that it has to be sprayed, and spray guns built by Joseph Binks and by Allen and Tom DeVilbiss were ready. Lacquer let finishers control drying in different weather by choosing solvents, and its price was steadier than that of shellac, a natural commodity. From the late 1920s to the early 1930s the furniture industry "changed over almost entirely from shellac to lacquer," the finishing writer Bob Flexner wrote in 2017.
Credit
DuPont; the spray gun makers Binks and DeVilbiss
1968
Light cures the finish
In 1968 Bayer AG in Germany made clear polishing finishes for wood from unsaturated polyester resins that harden under ultraviolet (UV) light instead of drying in air. Clear fillers for particleboard, applied by roller, followed in 1969. From the mid-1970s acrylate resins allowed faster conveyor belts and coatings with no solvent to evaporate, and by 1978 polyester acrylates were established for furniture, according to Manfred Müller, who joined Bayer in 1968 to work on these coatings.
LessonThe first particleboard fillers cracked because unsuitable fillers were used, Müller wrote.
Credit
Bayer AG's coatings chemists, as recalled by Manfred Müller (2004)
1990
Finishers move to Mexico
Between 1988 and 1990 between 11 and 28 wood furniture makers in the Los Angeles area, 1% to 3% of the total, moved to Mexico, affecting about 960 to 2,547 employees, the US General Accounting Office (GAO) estimated. About 78% of them cited strict air pollution limits on paint coatings and solvents as a reason. About 83% of the firms sampled cited the cost of workers' compensation insurance and wages.
LessonStrict limits on finishing solvents in one region can push the work to places without them.
Credit
GAO report NSIAD-91-191 (April 1991), requested by Representative John Dingell
1995
US limits finishing pollutants
On 7 December 1995 the US Environmental Protection Agency (EPA) set national limits on hazardous air pollutants, such as toluene, xylene and formaldehyde, from large wood furniture plants. EPA says a single plant can emit more than 23 tonnes a year of them. Existing plants must keep stains, sealers, topcoats and other finishing materials to at most 1.0 kg of these pollutants per kg of coating solids as applied, and new plants to 0.8 kg for most coatings, or capture the fumes with a control device.
Credit
US EPA, in 40 CFR part 63, subpart JJ
1999
Europe caps solvent emissions
The European Union's Solvent Emissions Directive (1999/13/EC) of 11 March 1999 limited emissions of volatile organic compounds (VOCs, solvent vapors that help form smog) from coating work, including the coating of wooden surfaces, above set amounts of solvent use. Existing plants had to comply by 31 October 2007, either by meeting the emission limits or by following a "reduction scheme" of their own that cut emissions by the same amount.
Credit
the Council of the European Union
Watch for
Spray robots taught by hand
Cefla Finishing says its T-Mate device records the path of a skilled finisher spraying a real part by hand and loads it into Cefla's iGiotto spray robot, which repeats and refines the movement. Cefla says this removes long programming sessions and lowers lacquer use, overspray, defects and rejects on doors, windows, stairs and furniture parts. The trade report gives no measured figures. LestaUSA showed spray robots that record a finisher's movements in 2021, also without published results.
Achieved whena plant using it publishes, or an outside party measures, lacquer used per part, the spread of film thickness and reject rates before and after, over at least a month of production.
Claimed by
Cefla Finishing, June 2026
Open
01
Graded photos of finish defects
An open set of photos of defects in clear and painted finishes on furniture parts, each labeled by two experienced finishers, and a baseline model that flags them. The defects include runs, orange peel (a bumpy surface like citrus skin), fish eyes (small craters from contamination), dust nibs, blushing (a milky haze from moisture) and sanding scratches showing through.
The briefCredit · unclaimed02
Finish spot tests, checked
A study of how often the usual solvent spot tests identify an old furniture finish correctly, checked against infrared lab analysis, and a phone guide that walks a refinisher through the tests and records the results with photos.
The briefCredit · unclaimed
Open 01 · the brief
Graded photos of finish defects
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open set of photos of defects in clear and painted finishes on furniture parts, each labeled by two experienced finishers, and a baseline model that flags them. The defects include runs, orange peel (a bumpy surface like citrus skin), fish eyes (small craters from contamination), dust nibs, blushing (a milky haze from moisture) and sanding scratches showing through.
Why it is still open
One public set covers coated wood. Jia, Chen, Liu and Wang released 13,400 images in June 2025 of four defects (scratches, cracks, bubbles and holes) in waterborne paint on wood products from a production workshop in Jiangshan, China; 336 of the images are originals and the rest are copies altered in brightness, orientation or noise. It has none of the common spray defects named above, comes from one plant and does not report how often labelers agree. Kruk and colleagues at the Warsaw University of Life Sciences detected "orange skin" on furniture parts from texture measurements in 2017, without a public image set. The largest wood image set, from VSB–Technical University of Ostrava (Kodytek and colleagues, 2021), labels more than 43,000 defects of ten natural kinds, such as knots, on sawn timber, not finishes. Public paint defect sets on GitHub and Hugging Face show car bodies. Searches of GitHub, Hugging Face, Zenodo, Crossref and the web in November 2026 found no public, labeled set of these spray defects on furniture finishes.
Built with what exists
A phone or camera with a raking light (a lamp at a low angle) and a polarizing filter to show surface texture; Label Studio for labeling; Ultralytics YOLO11 for detection, pretrained on the 2025 Jiangshan coated-wood set; Google's Gemini Decision 1, which returns a probability for each answer on a fixed list, to sort each part into accept, touch up or refinish (see the Note "A thousand questions, one pass").
First step
Ask a cabinet door maker to let you photograph every part sent to its touch-up station for one week, with the finisher's note of the defect.
Done looks like
3,000 photos from at least five finishing lines or shop finishing rooms, each defect labeled by two finishers, with their agreement reported. The model's recall for each defect type is published at no more than one false alarm per ten good parts, with the photos under an open license.
Open 02 · the brief
Finish spot tests, checked
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A study of how often the usual solvent spot tests identify an old furniture finish correctly, checked against infrared lab analysis, and a phone guide that walks a refinisher through the tests and records the results with photos.
Why it is still open
Before repairing a finish, refinishers test which solvent softens it on a hidden spot. Denatured alcohol softens shellac, lacquer thinner makes lacquer or a water-based finish sticky, and a finish untouched by both is a reactive one such as varnish (Fine Woodworking). Michele Derrick's 1988 infrared key identifies five natural resins in furniture finishes at concentrations as low as 1%, but it needs a lab instrument. A 2015 study of four pieces at the ASTRA Museum in Sibiu, Romania, by Timar and colleagues used solvent tests to choose which reference spectra to compare, and on one table aged shellac dissolved so poorly in alcohol that the tests pointed to a synthetic resin until infrared analysis showed shellac. Searches in November 2026 found such case studies but no published error rate for the spot tests.
Built with what exists
Denatured alcohol and lacquer thinner on cotton swabs, with nitrile gloves, good ventilation and no flames nearby; a phone camera for before-and-after photos; ATR-FTIR (an infrared instrument that reads a tiny scraping pressed on a crystal), available at many university chemistry departments; scikit-learn to sort the spectra.
First step
Buy ten used pieces of furniture of different ages, run the spot tests on hidden spots, and ask a university chemistry lab for a price to run ATR-FTIR on ten scrapings.
Done looks like
150 second-hand pieces from at least 30 owners or thrift stores, each spot-tested by three refinishers working separately and analyzed by FTIR. The share each refinisher got right is published for shellac, lacquer, varnish, oil and water-based finishes, with the spectra and photos released.