Reference-checking: how we verify historical period accuracy

Jonas sent a brief in March for "a 1340s English knight at Crécy, plate armour, two-handed sword, mounted." I had three problems with the brief before the email was closed. Plate armour wasn't standard in the 1340s — transitional kit (mail base, plate at the joints) is what most English knights wore at Crécy in 1346. The two-handed sword is wrong by about a century. And Crécy was a defensive infantry battle for the English; the famous knights were dismounted, fighting on foot with their horses behind the line. By the end of the day, after about forty minutes of cross-checking three museum archives and a battlefield-historian's account, the brief had become "a 1346 English man-at-arms at Crécy, mail hauberk with a coat of plates over the torso, plate gauntlets and knee cops, bascinet with a houndskull visor, an arming sword and a poleaxe, dismounted on foot." Different portrait. Better portrait. The reference check is the work most clients never see.

I'm Hector. I run Design Vortex, and the single most-asked question I get from clients after a few commissions with us is some version of: "how do you actually verify that any of this is right?" This piece is the studio's reference-checking workflow — what sources we use, in what order, how we cross-reference, the common errors we catch before the first sketch, and the homework a client can do on their own to make the back-and-forth shorter. It's the operational counterpart to the historical character art commission guide, which sits above this as the wider commission piece.

If you've already read the medieval armour reference piece or the Viking-era forbidden lands piece, you've seen the output of this workflow. This is the workflow itself.

Table of contents

• The two-pass reference system
• Source hierarchy: museum, academic, popular
• The first-pass brief review
• Cross-checking: when one source isn't enough
• Common brief errors we catch in pass one
• What clients can do at home
• When historical accuracy isn't the goal

The two-pass reference system

Every historical brief that comes through Design Vortex goes through two reference passes before painting starts:

1. Pass one — brief review. Within 24-48 hours of a brief arriving, I read it through once with no references open, then a second time with the studio's reference shelf open. I flag any anachronism, era-mismatch, or factual error and write back with a list of questions. This usually takes 30-60 minutes per brief.
2. Pass two — pre-sketch verification. Once the brief is locked, the painter assigned to the piece pulls three to five reference photographs from museum collections that match the brief's specifications. These get embedded into the project file as the painter's anchors. This usually takes 45-90 minutes.

After pass two we paint. The brief at this point reads as a specific century, with specific armour or clothing pieces, with specific weapons, and the painter has photographs of real surviving objects to work from rather than mental images.

Most studios don't do this. Most studios paint whatever the brief says, and if the brief says "12th-century knight in plate armour" the painting comes back as a 15th-century knight wearing armour that didn't exist for another 200 years — but with a 12th-century label. The label and the painting disagree. The brief looks like it was honoured, but the history isn't. We do the reference work because the alternative is making historical portraits that don't read as the history they claim.

Source hierarchy: museum, academic, popular

Not all references are equal. The hierarchy we work in, in declining trustworthiness:

• Museum object records. Photographs of real surviving objects in their original or near-original state, with dated provenance, in catalogues maintained by professional curators. The gold standard. Wallace Collection, Royal Armouries, Met Arms and Armor, V&A textiles, National Museum of Denmark, Tokyo National Museum, the Bargello in Florence. Searchable online, increasingly so each year.
• Academic reconstruction. Peer-reviewed reconstruction work by professional historians and material-culture researchers. Hilde Thunem's Viking textile reconstructions, the Royal Armouries' own combat-history publications, the Metropolitan Museum's catalogues, the Khalili Collection's Islamic arts research. Treated as authoritative when the museum object is unavailable or insufficient.
• Reenactor research at the high end. Some reenactment communities — Regia Anglorum for early-medieval, the Company of St. George for late-medieval, specific Sengoku samurai reenactor groups — do research at academic quality and publish detailed reconstruction notes. We use these critically, cross-checked against museum sources.
• Popular history books. Osprey Men-at-Arms series and similar. Decent broad-strokes reference, sometimes outdated in detail, often visually appealing but reconstruction-rather-than-photograph. Used as second-tier reference, not as primary.
• Films and games. Used for mood, atmosphere, costume drama scale, lighting. Never used as historical reference. Kingdom of Heaven is closer than most films and still mixes centuries. The History Channel's "Vikings" series is visually beautiful and chronologically scrambled. Assassin's Creed games are gorgeous and selectively accurate.
• Pinterest and Tumblr. Treated as untrustworthy by default. Often re-uses 19th-century history paintings (Victorian artists invented an enormous amount of what we now think looks medieval). Sometimes re-uses outright AI generations. Useful as a starting point for mood-board exploration, never as evidence for an actual claim about period accuracy.
• AI-generated reference images. Treated as adversarial. AI image generators are particularly bad at period accuracy — armour pieces hallucinate, textile patterns invent themselves, helmets generate impossible articulation. If a client sends AI images as references, we label them clearly as direction-of-travel rather than period-accuracy targets, and back them up with at least one museum photograph for any specific kit.

The single biggest move a studio can make to raise historical-accuracy quality is to inverse the typical reference funnel. Most artists work outward from Pinterest. We work inward from the Wallace Collection. The portraits read differently because the references read differently.

The first hour of a historical brief at our studio is spent reading museum catalogues. The first hour at most studios is spent on Pinterest. That's the whole gap.

The first-pass brief review

What happens in the 30-60 minutes after a historical brief arrives:

1. Read the brief through once with no references open. I'm listening for what the client wants and what they think they're describing. Sometimes the gap between those two is significant.
2. Identify the named period and the named objects. Note every specific date, every named weapon, every named piece of clothing or armour, every named location, every named person.
3. Cross-check each named element against the date. Is the weapon era-correct for that date? Is the armour era-correct? Is the location's named feature historically present at that date? (Edinburgh Castle in 1100 looked completely different from Edinburgh Castle in 1500.)
4. Identify mood references the client has provided. Films, games, novels, paintings. Note which century each mood-reference actually depicts. If they're misaligned, flag it.
5. Write back with specific questions. Not "this is wrong," but "I want to check: when you say plate armour at Crécy, do you mean the full plate of the 1450s, or the transitional mail-plus-plate that was actually worn in 1346? They look quite different and I want to paint what you actually want."

The single most useful question I ask in pass one: "is the period the look, or the look the period?" If the client wants a specific historical event painted accurately, the period leads. If they want a specific visual look that happens to lean historical, the look leads and we frame it as historical-flavoured fantasy rather than historical-strict.

Helene's 1898 spirit-photography brief came in clean — she had the year, the social position, the specific lighting, the prop. Pass one took 20 minutes. The brief I sent back asked one question: "S-curve corset or straight-front? S-curve is more 1900s; straight-front is 1890s." She replied with "straight-front, definitely — this is 1898 not 1903." That's a brief that lands cleanly because the client knew the difference.

Cross-checking: when one source isn't enough

For any non-trivial historical detail, the studio cross-checks against at least two sources. One museum object record alone is sometimes not enough, because:

• Objects in collections are often labelled with broad date ranges. A "13th-century helmet" might be 1220 or 1280, and the difference matters. Cross-checking against dated effigies and church brasses narrows the window.
• Some collection labels are outdated. Museum records get revised as scholarship advances. An object catalogued as "Viking" in 1960 might now be reclassified as "Carolingian" — the geography and period shift. We trust recent reclassifications.
• Reconstruction artwork can be wrong even when the underlying museum object is right. A 19th-century reconstruction drawing of an Anglo-Saxon helmet often makes choices the original object doesn't support. We use the photograph, not the engraving.

The cross-check process in practice:

1. Find the object in one source. Wallace Collection has a 1380s bascinet — record number, photograph, date range.
2. Find the same object class in a second source. Royal Armouries has three 1380s bascinets with comparable construction.
3. Find a contemporary depiction. Tomb effigies, church brasses, illuminated manuscripts. The Walter de Mauny effigy at the Charterhouse in London or the Black Prince's effigy at Canterbury are dated portraits in armour.
4. Compare and confirm. If all three sources agree on the construction, we paint from that consensus. If they disagree, we note the disagreement and paint to the museum-object consensus rather than the depiction.

A Sengoku samurai brief gets the same treatment: a kabuto referenced in the Tokyo National Museum collection, cross-checked against a comparable example in the Metropolitan Museum or the Bargello, then verified against a contemporary woodblock print or hand-scroll painting. Three sources, then paint.

Common brief errors we catch in pass one

The recurring mistakes that show up in historical briefs, in rough order of frequency:

• Wrong era for the kit. "12th-century plate armour" (covered in the medieval armour mistakes piece), "Sengoku samurai in Edo court silks," "1860s Victorian woman in an Edwardian S-bend corset." The single most common error, and the easiest to fix once flagged.
• Mixing trans-Atlantic centuries. "An English Victorian in 1880s clothing meeting an American Civil War officer in 1862" — works fine if the meeting is anachronistic by intent, but often the client hasn't realised the date gap.
• Weapons too late for the armour. Two-handed greatswords with 12th-century mail. Longswords with Sengoku armour. Match the weapon to the era.
• Hair and beard styles wrong for the era. Wild Viking beards, untrimmed Edwardian moustaches on a 1923 character, samurai with loose flowing hair. The grooming was almost always more controlled than the brief assumes.
• Lighting that breaks the era. Electric light in an 1840s rural cottage, flash photography in 1922 (it existed but was rare and looked specific), a candlelit Edwardian study in 1925 in a wealthy urban home (probably electric by then).
• Architectural anachronism in the background. A Norman knight in front of a Gothic cathedral that was built two centuries after his death. The English Victorians built a lot of "medieval" architecture that didn't exist in the actual medieval period — the Houses of Parliament are Victorian Gothic Revival, not medieval. Briefs sometimes use the building as period reference without realising.
• Trademarked or specific named characters mis-dated. "Eleanor of Aquitaine in plate armour" — Eleanor lived 1122-1204, plate armour wasn't worn in her lifetime, mail-and-surcoat was. The Eleanor of Aquitaine commission piece covers her specifically.
• Class-position mismatches. A peasant carrying a longsword (a knight's weapon by both convention and law). A common samurai with a daisho he wouldn't have been entitled to wear. A 1920s working-class character in a three-piece bespoke wool suit. Period kit was deeply class-marked, and brief writers often miss this.

Selene sent a brief in July for "a 14th-century French peasant resistance leader in plate armour." Two errors in one line. Plate armour was a noble class marker in the 14th century, not a peasant garment. And 14th-century peasant resistance leaders — the Jacquerie of 1358 are the closest fit — generally wore the same wool tunic and hood they wore farming. We rebuilt the brief as a Jacquerie leader in working clothes with a captured weapon, and the portrait read as the actual social moment instead of a fantasy that didn't fit any century.

What clients can do at home

The reference-checking work the studio does in pass one can be shortened significantly if the client has done some homework. The three things that help most:

• Name the year, not the era. "1346" is twenty times more useful than "medieval." "1583" is much more useful than "Sengoku." If you don't know the exact year, name the half-decade or the named event ("at Crécy," "during the Boshin War," "the year my great-grandmother got married in 1898"). The narrower the window, the cleaner the brief.
• Look up one museum object that matches your character before you write the brief. The Met, the V&A, the Wallace Collection, the Royal Armouries, the British Museum, the National Museum of Denmark, the Tokyo National Museum — all have free online catalogues. Find one helmet, one dress, one sword, one ceremonial object that fits your character's period and class. Mention it in the brief. The painter will look it up, and you'll have anchored the entire piece in one citation.
• Separate "I want this look" from "I want this period." If you want a historical look but don't care about strict accuracy, say so — the brief becomes much easier to execute and we won't waste a pass-one cycle asking you to confirm accuracy you weren't claiming.

A brief that arrives with one museum-object citation and a named year is essentially pre-checked. The painter can start research from the citation outward rather than from the brief outward. Briefs that arrive like this from clients are rare enough that I remember the names of clients who do it.

For the brief-writing craft generally, the how to write a commission brief piece is the wider companion. The reference-checking workflow this piece describes plugs into the brief-writing workflow that one describes.

When historical accuracy isn't the goal

A real consideration that comes up often: not every historical-leaning brief is actually trying to be historically accurate. A lot of briefs are using "historical" as a flavour rather than a fact-check target. That's fine — but we frame it differently.

The three registers:

1. Historical-strict. The brief is depicting a specific event, person, or moment, and the goal is period accuracy. The reference workflow described in this piece applies fully.
2. Historical-informed fantasy. The brief draws on historical sources but mixes elements freely (a "samurai with a Western-style sabre," a "Viking with steampunk goggles"). The reference workflow still applies for the historical elements, but the brief is framed openly as fantasy. The painting is honest about what it's doing.
3. Genre-coded as historical. The brief is fantasy that happens to lean historical (most D&D characters, most Conan-adjacent work). The reference workflow is light — we check for grossly wrong combinations, but we don't fact-check everything. The brief is fantasy and the painter has fantasy freedom.

Asking which register the brief sits in is the first question pass one tries to answer. The reference work scales accordingly. A historical-strict brief for an 1898 widow with a spirit-photography studio gets the full Wallace-Collection-grade reference treatment. A historical-informed fantasy of a "warlord queen on a battlefield" gets a lighter check, with the painter given more interpretive room.

For the cross-genre split between strict and fantasy registers, the fantasy character art commission guide covers the fantasy-strict end, and this piece covers the historical-strict end. The middle is where most commission work actually lives, and we handle that case-by-case in the brief conversation.

Closing the loop

The reference-checking process described in this piece is the part of historical commission work that most clients never see and most studios don't do. The work is sometimes 90 minutes of reading museum catalogues before the painter has lifted a brush, and the difference shows up in the finished painting in ways the client can feel even if they couldn't tell you which century the kabuto is from. The portrait reads as the era it claims to be. The history holds up.

If you've got a historical character — strict or informed-fantasy, any century, any country — drop a brief through the order form. The portfolio has the closest visual references for what our historical work looks like across periods, and the character work services page lays out tiers and pricing. For longer-form multi-figure work — a Crécy battle scene, a Sengoku command tent, an 1898 séance with three sitters — the custom projects page handles the scale.

For the period-specific deep dives, three sibling pieces cover the bulk of historical commission territory: the medieval armour mistakes piece, the Viking-era forbidden lands piece, and the Edwardian and Victorian portrait commissions piece. For the samurai end of historical, the samurai character art Japanese periods piece and the samurai portrait kabuto and kimono piece are the close-in companions. For the named-character end, the Eleanor of Aquitaine commission piece and the Viking portrait helmets and braids piece walk through specific portrait briefs. For the 1920s investigator crossover, the horror character art commission guide covers the gaslit-investigator register. For brief-writing craft and pricing, the how to write a commission brief piece, the choosing a commission style piece, and the character art commission pricing piece are the operational companions. For the painting process itself, the sketch-to-color-to-final walkthrough shows how a historical brief becomes a finished portrait once the reference work is done.

A historical portrait that's worth painting is one where the period reads instantly to anyone who knows the era and reads as quietly coherent to anyone who doesn't. The reference work is what gets you there. Name the year, cite one object, tell us which register you're in. The painter takes it from the brief outward.