Methodology

GlossAtlas compares paint finishes by their published gloss/sheen measurements, not by their marketing names, not by color, and not by any converted or estimated number. This page explains the rules behind every comparison on the site.

Why there is no 60°→85° conversion

Gloss is measured with a glossmeter at a fixed angle from the paint surface. Manufacturers commonly publish figures at 60° (the usual angle for mid-gloss finishes) or at 85° (near-flat finishes, where a shallower angle is more sensitive to small differences in sheen). The standard test methods, ASTM D523 and ISO 2813, define gloss measurement at these angles. Some finishes are published at both angles; many are published at only one.

GlossAtlas never converts a 60° figure into an equivalent 85° figure, or vice versa, and it never will by default. No defensible, generally applicable constant or curve exists for that conversion across brands and finishes — the relationship between a paint’s 60° and 85° reading is not a fixed ratio, it depends on the specific formulation, and manufacturers that do publish both figures for the same finish do not report a consistent relationship between them. Publishing a conversion anyway would mean presenting an invented number as if it were a measurement. So the rule is simple and absolute: a comparison only ever happens between two figures published at the same angle. When two finishes were only ever measured at different angles, GlossAtlas says so rather than bridging the gap with a fabricated number.

The classification tiers, in plain language

Every comparison GlossAtlas produces falls into one of a small number of outcomes:

• A strong overlap — the two published ranges share a large portion of their span. It is a statement about two published numbers, not about how the paints will look on a wall.
• A partial overlap — the two published ranges share some span, but a smaller portion of it. Still a real overlap, just a narrower one.
• The nearest published range, without overlap — the two ranges don’t touch, but one finish is still the closest available option in that brand’s lineup, and the gap between them is disclosed in gloss units.
• No close equivalent — the nearest available finish in the target brand is far enough away in gloss units that calling it “closest” would be misleading rather than useful, so GlossAtlas says plainly that no close equivalent was found.

None of these is ever presented as “identical,” “exact,” “the same finish,” or “manufacturer-approved” — GlossAtlas reports a comparison of two published numeric ranges, not a claim that two products will look or behave identically on an actual wall.

Why “no close equivalent” is drawn at a specific threshold

The line between “nearest published range” and “no close equivalent” is a disclosed editorial policy choice, not a scientific constant discovered in the data. It exists so that if a brand’s nearest available finish is dramatically different in sheen from what you started with, GlossAtlas says so honestly instead of recommending a technically-nearest option that isn’t actually a reasonable substitute. The value is defined once, in the site’s matching engine, and quoted in “How overlap is scored” below.

What “closest published match” does — and doesn’t — mean

A GlossAtlas result tells you which published finish, in the brand you’re comparing against, has the closest published gloss/sheen figure to the one you started with, at the same measurement angle. It does not mean:

• the two paints are the same product, formulation, or color,
• the two paints are guaranteed to look identical once applied,
• the manufacturer of either paint endorses or has reviewed the comparison, or
• every factor that affects perceived sheen (substrate, application method, film thickness, lighting, and the color itself) has been accounted for — published gloss ranges can overlap and a real wall can still look different for reasons a glossmeter reading alone doesn’t capture.

Every figure GlossAtlas shows is transcribed directly from a manufacturer’s own published technical or product data sheet, linked from the result itself, with the date it was last checked. GlossAtlas never hosts a copy of a manufacturer’s document and never estimates a figure a manufacturer hasn’t published. If you want to see the full picture behind any single result, follow the source links — that’s the actual evidence, not this page’s summary of it.

Why this differs from a typical “satin vs. eggshell” article

Most paint-sheen content online — including brand-published buying guides — describes finishes in qualitative terms: reflectance ranges quoted from marketing copy, or comparative language like “soft shimmer” versus “velvety.” That’s genuinely useful for a general sense of a finish, but it isn’t a measurement, and it doesn’t tell you whether two specific products from two specific brands actually land in the same range. GlossAtlas answers a narrower question with a harder-edged answer: does this brand’s published gloss-unit figure, at this angle, actually overlap with that brand’s published figure, at the same angle. See the 60° vs 85° measurement guide for what the angle itself means, and why “Satin” isn’t the same sheen in every brand for why the marketing name alone can’t answer this question.

The evidence model: three different kinds of statement

Everything on this site is one of three things, and the pages say which.

1. Manufacturer data. The gloss ranges themselves, transcribed from the manufacturers' own product data sheets. The dataset holds 33 finish records across 4 manufacturers, and all 33 are verified primary source records.
2. GlossAtlas derived analysis. What the fixed rules on this page, and the formulas below, make of manufacturer data: overlap, gap, "closest", and every statistic on the analysis pages and worked examples. It is interpretation of other people's numbers; it is not new data.
3. GlossAtlas first-party measurements. Gloss readings GlossAtlas takes itself, with a gloss meter, on painted samples. None has been published yet. They are kept in a separate dataset, shown beside the manufacturer figure and never mixed into it, and the matcher never uses them. See below.

No third-party laboratory measurements are used anywhere. GlossAtlas makes no claim of paint-industry expertise or of laboratory accreditation.

First-party measurements: protocol

The first study (five samples from three brands, read at 60° and 85°) has a protocol but no readings yet, so nothing from it is published and no conclusion is drawn. Every study follows the same protocol (version 1.0):

1. Samples are bought new, from a retailer, in the colour and base named in the study; the can's lot or batch code and the purchase details are recorded and photographed.
2. Each can is stirred as its label directs and applied with a film applicator that delivers a 100 µm wet film, in 2 coats (recoat time recorded) onto a sealed, non-absorbent white test chart. 3 separate charts are made per sample.
3. Charts lie flat and cure for 7 days at 23 ± 2 °C and 50 ± 5 % relative humidity, logged throughout.
4. Gloss is read with a glossmeter whose maker states conformity to the gloss test standards, at 60° and 85°. The meter is calibrated against its own standard before each session and checked against a second standard before and after.
5. 5 positions are read on each chart at each angle, with the meter oriented the same way every time. Charts carry opaque codes and are read in a shuffled order fixed before the session, so reading order is not tied to brand. Unless a study says otherwise, the person who prepared the samples also reads them, so the reading is not blind to brand.
6. Every reading is kept. Mean, minimum, maximum and standard deviation are computed from the readings by script; they are never typed.
7. The mean at each angle is set beside the manufacturer's published range at the same angle. A 60° reading is never compared with an 85° figure.

A study is published only when the data passes a gate: instrument and calibration recorded, environment logged, at least 3 charts and 15 readings per sample and angle, photographs of the cans, charts and instrument, every sample reviewed, and written findings. Raw readings are downloadable; mean, minimum, maximum and standard deviation are computed from them.

How overlap is scored

For two finishes at the same angle, the overlap is the width of the span the two published ranges share. That width is divided by the width of the narrower of the two ranges. If the result is at least 50%, the result is a strong overlap; a smaller shared span is a partial overlap. Ranges that only touch at a single value share no span and are reported as nearest, not overlapping. If the nearest range is more than 15 gloss units away, the result is "no close equivalent".

Derived statistics and formulas

The analysis pages and worked examples report counts, rates, gaps and rankings that GlossAtlas computes itself from the published ranges. They are calculated when the site is built, from the same data and the same matching rule as the tool, so the tables and headline numbers cannot drift from the dataset. A few ranges are also quoted inline in sentences; tests check the facts those sentences depend on. Every calculation takes two figures published at the same angle. No figure is converted, subtracted or averaged across angles. One thing does compare across angles: when a finish publishes at both angles, the matching engine reads both and orders the two results by its ranking rule (overlap fraction, then gap) to decide which is shown first. Both are always shown, and neither is converted into the other.

• Range, width, midpoint. A range is the manufacturer's lowest and highest published value at one angle. Width = highest − lowest. Midpoint = (lowest + highest) ÷ 2. The midpoint is a descriptive summary, not a gloss value anyone measured.
• Shared span and overlap fraction. Shared span = min(highest values) − max(lowest values); a positive span means the ranges overlap. Overlap fraction = shared span ÷ the narrower of the two widths. A fraction of 50% or more is a strong overlap.
• Gap. When the ranges do not overlap, gap = max(lowest values) − min(highest values), and the direction says which finish is glossier. A gap of 0 means the ranges touch at a single value and share no span.
• Reading. One starting finish against one other brand. The engine ranks that brand's finishes at one angle by overlap fraction or gap, and every finish tied for best is reported. When the starting finish is published at two angles, both are read and the better one is shown first.
• Same name. Two finishes share a sheen name when their names are the same word (Flat, Matte, Eggshell, Pearl, Satin, Semi-Gloss). A reading is "same name among the closest" when a finish with that name is in the best set; "closer elsewhere" when a same-named finish can be compared but a differently named one is closer; "not comparable" when the same-named finish cannot be set beside the starting finish at the angle of the reading (most often because they were published at different angles); and "no same name" when the other brand has none.
• Comparable pair. Two finishes from different brands are comparable when both publish a figure at one or more of the same angles. Each finish pair is counted once; a pair that shares both angles supports two same-angle comparisons.
• Midpoint spread. For one sheen name at one angle: the highest range midpoint minus the lowest, across every published range (a brand with two lines contributes two). Reported only where at least two brands publish that name at that angle.
• Midpoint shift. The distance between the midpoints of two ranges at the same angle, used to compare two lines of one brand.
• Step between neighbouring finishes. Within a product line, finishes are ordered by sheen name; step = upper finish's lowest value − lower finish's highest value, at an angle both publish. If the two neighbours share no angle the step is reported as not measurable.
• Source age. (the dataset's latest verification date − a sheet's printed revision date) ÷ 365.25 days, rounded to a tenth of a year. The reference date is a fixed date in the data, not the date a page is viewed.
• Pooled figures. The table of range widths and the distribution of gaps pool readings made at either angle. A width is a size, not a gloss value, and each gap in the distribution is a same-angle gap; nothing is compared across angles. Rounding and scope. Percentages are rounded to whole numbers and other figures to one decimal at most. All statistics describe this dataset (33 records, 4 brands, interior lines only); they are not estimates for the whole industry, are not weighted by sales, and carry no confidence intervals.

How pages are produced and checked

1. Data. Each record is taken from a manufacturer's own data sheet: the ranges are extracted from the document with a script (scripts/data/verify_sources.py in the project) and compared with what is stored, then re-checked whenever a source is re-verified. Corrections and revisions are dated in the change log.
2. Selection. The questions the analysis pages ask are chosen by GlossAtlas because they are ones the cross-brand dataset can answer and a single manufacturer sheet cannot. The set of finishes and lines in the dataset is a curated selection, not a scrape.
3. Computation. Every number on an analysis page is calculated from the data at build time (see the formulas above). Automated tests fail if the headline figures and facts a page's wording depends on change, so that wording is re-read when the data moves.
4. Writing. The interpretation, the choice of examples and the stated limits are GlossAtlas's own, published under the operator's responsibility. The page text was drafted by an AI model (Claude) working from the computed tables, then checked by automated tests and by separate review passes that re-derived the figures from the raw data. The model is not a source of figures; where a sentence quotes a range, tests check it against the data.
5. Disclosure. Each analysis page says what kind of statement it makes, links the formulas and the sources, and states its limits. GlossAtlas has not tested any paint; no figure here is a laboratory or independent measurement.

Freshness and rechecks

The figures were last checked against the manufacturers' documents on 2026-10-10. The source documents themselves carry revision dates from 2020-05-01 to 2026-08-01. GlossAtlas verifies its sources periodically, by hand and without a fixed schedule or guaranteed interval. The last verification is the date above. If a source is found to be unreachable or changed, the record is marked and the page that uses it says so.

Data change log

Every change to the figures or their sources, newest first. If you spot an error, the About page says how to report it.

• 2026-10-10 · Correction: Behr Premium Plus Interior Flat (1050) updated to its current data sheet (revision 5/2025): 0–5 at 60° and 0–5 at 85°. The earlier entry (0–4 at both angles) came from the 5/2020 sheet.
• 2026-10-10 · Citation update: Behr Marquee Matte, Eggshell, Satin and Semi-Gloss (1450, 2450, 7450, 3450) and Premium Plus Satin (7050) now cite Behr's own data sheets instead of Home Depot-hosted copies. The Marquee sheets are newer revisions (5/2024, 6/2024, 10/2023, 6/2024); the published ranges did not change. The Premium Plus Satin sheet is the same revision (5/2020).
• 2026-09-21 · Correction: Sherwin-Williams Emerald Matte updated to its current data sheet (101.08MAB): 0–5 at 60° and 4–8 at 85°. The earlier entry (0–5 at 85° only) came from the 11/2023 sheet.
• 2026-09-21 · Citation update: Sherwin-Williams Emerald Flat and Satin now cite their current data sheets (12/2025 and 04/2026). The published ranges did not change.
• 2026-09-21 · Addition: 17 finishes added from four product lines: Duration Home (Sherwin-Williams), Regal Select (Benjamin Moore), Premium Plus (Behr) and Ultra (Valspar).
• 2026-09-21 · Check: All 16 original records re-checked against their manufacturer documents.
• 2026-09-06 · Initial: Initial dataset: 16 finishes in four product lines, each checked against its manufacturer data sheet.

Limits of this dataset

• Limited coverage. 33 finishes from 4 brands, across 8 interior product lines. Exterior and specialty products, and any other interior lines a brand sells, are not covered. It is a set that can be checked record by record, not a catalog of every paint.
• Published ranges, not readings from your can. No source specifies the paint base or color the range applies to, and gloss can vary with both.
• Angle coverage varies. 12 finishes are published at 85° only, 13 at 60° only and 8 at both. Where two finishes share no angle, no comparison is made.
• Gloss only. Color, undertone, durability, coverage and price are not compared.

Source register

Every record, the document it was transcribed from, and the angle(s) it publishes. Each row opens the manufacturer's own document.

)

Back to the comparison tool · Same name, different published range · Brand independence statement