The direct answer
A visual novel character is designed by locking four decisions before any painting starts: one recognizable silhouette, one fixed palette, one rendering finish, and one frozen list of expressions. The size of the sprite sheet then follows from simple multiplication, because every outfit must carry the full expression set. A lead with 6 expressions and 2 outfits needs 12 sprite files. A rival with 8 expressions and 3 outfits needs 24 files. The tool on this page performs that multiplication, pairs it with silhouette and palette direction drawn from the story role and the art finish, and returns an asset matrix plus a consistency checklist that can travel to an artist as one package.
That discipline matters because visual novel art fails in a particular way. A single portrait can look excellent while the full set falls apart, with heights drifting between outfits, eye styles shifting from face to face, and shadow colors wandering across files. Players notice these breaks quickly, since the same character stands on screen for hours beside dialogue text. The remedy is not greater talent but tighter specification: write the height, the palette, the line rules, and the export pattern down, then paint every file against those anchors.
Why a sheet beats a pile of portraits
A newcomer often commissions portraits one at a time, asking for a neutral face first, then a smile later, then a second outfit when the script demands it. Each request seems reasonable in isolation. Together they produce drift, because every new commission reinterprets decisions the earlier files already made. The smile arrives with a slightly different jaw. The new outfit changes the shoulder width. The painter adjusts the shadow tone to suit the new costume and silently breaks the match with the old one.
A character sheet prevents this by declaring the answers once. The turnaround drawing fixes proportions. The expression set fixes the face count. The outfit list fixes the costume count. The swatch sheet fixes every color. Later work then becomes reproduction rather than reinvention, which is faster for the artist and safer for the project. Even a solo developer painting their own cast benefits, because memory alone cannot hold exact color values across weeks of work.
The sheet also controls cost. Sprite pricing is commonly quoted per file or per hour, so the file total is the budget lever. Cutting two rarely used expressions from a two-outfit lead removes four files. Merging two situational outfits into one removes an entire expression set worth of painting. These choices are painless at the brief stage and painful after painting begins, which is why the scope verdict on this page appears before any artist is contacted.
How the tool math works
The core arithmetic is deliberately plain. The tool reads the expression count and the outfit count, clamps each into a sane range, and multiplies them into the total file count.
Total sprite files equals expressions multiplied by outfits.
The defaults show the standard lead: 6 expressions times 2 outfits equals 12 files. Change the inputs and the total follows the same rule. A love interest with 5 expressions and 1 outfit totals 5 files. A rival with 8 expressions and 3 outfits totals 24 files. A mentor with 10 expressions and 4 outfits totals 40 files, which the tool flags as heavy scope because forty paintings for one character will consume most of a small art budget.
The estimated paint load uses one further multiplication. The tool applies an illustrative planning assumption of two and one half hours per sprite file, then multiplies that assumption by the file total. Twelve files therefore show about thirty hours of painting load. Twenty-four files show about sixty hours. Forty files show about one hundred hours. These figures are explicitly labeled as an illustrative planning assumption inside the results. They exist so a planner can compare scope options against each other, not so anyone can predict a real invoice. Actual speed depends on the artist, the finish, the revision rounds, and the complexity of each costume, so the honest move is to use the estimate for comparison and confirm real pricing directly with the artist.
The scope verdict applies two bands to the file total. Twelve files or fewer reads as lean scope, meaning one character can be painted carefully without crowding out the rest of the cast. Thirteen through thirty files reads as moderate scope, meaning the list should be frozen and ranked by script frequency before painting starts. More than thirty files reads as heavy scope, meaning the brief should lose expressions or outfits before commissioning. These bands are planning judgments, not industry standards, and a well funded project may carry heavy sheets deliberately. Their value is forcing the conversation early, while cuts are still free.
Worked numbers make the tradeoffs concrete. Suppose a lead carries 7 expressions and 2 outfits. The total is 14 files, which lands in the moderate band, with an illustrative load around thirty-five hours. Removing one expression drops the total to 12 files and returns the sheet to lean scope, saving two files and roughly five illustrative hours. Alternatively, keeping all 7 expressions but merging the outfits into one drops the total to 7 files. Both cuts are legitimate; the first protects costume variety while the second protects emotional range. The right choice follows the script. A story with frequent costume changes keeps the outfits and trims faces, while a romance with long emotional scenes keeps the faces and trims outfits.
A second worked comparison helps with casting. Imagine four characters at 6 expressions and 2 outfits each. That is 12 files per character and 48 files for the cast, with an illustrative load near one hundred twenty hours. Reducing two supporting characters to 3 expressions and 1 outfit each changes their totals to 3 files apiece, saving 18 files across the pair and bringing the cast to 30 files. The leads keep their full range, the budget recovers nearly half, and players never miss faces the script never called for.
Designing by story role
The tool adjusts silhouette and palette direction from the selected role, because each role carries different visual duties on screen.
The protagonist needs instant recognition. This character appears in the most scenes, often at small dialogue sizes, so the design should survive shrinking. One signature color does more work than three competing accents. An upright, open stance reads as capable without stealing focus from whoever shares the screen. The palette direction for a lead therefore favors high contrast and one memorable anchor: a crimson scarf, a brass compass, a teal coat lining. Everything else stays quiet enough that the anchor can do its job.
The love interest needs warmth and gesture range. Intimate scenes live in small expression shifts — a softened gaze, a hesitant smile, a glance away — so the face set should include at least one tender and one vulnerable state alongside the standard happy and sad. Visible hands matter more here than for any other role, because reaching, fidgeting, and offering gestures carry romantic subtext. The palette direction favors warm, approachable bases with a soft secondary color reserved for close scenes.
The rival needs contrast against the lead. When both characters share the screen, which happens constantly in confrontation scenes, the player should read the opposition without studying details. An angled stance, a sharper shoulder line, and a clashing accent color all serve that purpose. If the protagonist wears warm red, the rival can carry cold steel blue. The contrast should be deliberate rather than accidental, which means choosing both palettes together instead of designing each character in isolation.
The mentor needs gravity and economy. Mentors deliver exposition, so their scenes are dialogue heavy and their sprites must remain interesting without becoming busy. A settled stance with a wide base, minimal head tilt, and a muted grounded palette with a single authoritative accent achieves this. Ornate costume detail works against the role, since it distracts during long speeches. One or two refined details — an engraved clasp, a patterned shawl edge — supply all the texture the role requires.
These directions are starting points, not cages. A timid protagonist or a flamboyant mentor can break every guideline above and still succeed, provided the break is intentional and applied consistently across all files. The tool supplies the conventional answer quickly so the designer spends creative energy on deliberate departures rather than on rediscovering the basics.
Choosing the art finish
The three finishes on this page describe how rendering is handled, and each one changes the labor behind the file count.
Anime finish uses clean cel shapes, bold line weight, and simplified shadow steps. It is the most forgiving choice for large sheets, because flat shadow shapes are faster to reproduce across many faces and outfits than blended gradients. Consistency is also easier to verify: a shadow shape is either the right color or it is not, with no ambiguous blending to debate. Most commercial visual novels use some variant of this finish for exactly these reasons.
Semi-realistic finish uses softer gradients, more detailed fabric folds, and restrained line work. It suits dramatic stories that want painterly mood without full painted labor. The cost is real, though. Every gradient must be rebuilt per file, and subtle blending drifts more easily than flat shapes. A semi-realistic sheet therefore demands stricter swatch discipline and more review passes, especially across outfits where fabric rendering could diverge.
Painterly finish uses textured brushwork, merged shadow edges, and painted fabric detail. It produces the richest single images and the heaviest sheets. Texture that looks effortless in one portrait becomes a consistency burden across forty files, because brush marks cannot be copied the way flat shapes can. Painterly suits small casts, premium positioning, and projects where each sprite doubles as promotional art. For large casts on tight budgets, it is usually the wrong choice, and the scope verdict will say so numerically.
Whichever finish is chosen, the rule is one finish for the whole cast. Mixing anime leads with painterly supports reads as two art directions colliding, and players interpret the mismatch as an error rather than a style. If a painterly look is affordable for only one character, the honest solution is to reserve it for CG illustrations and keep all sprites in the shared finish.
Building the expression list
Expressions are the emotional vocabulary of the character, and the list should be derived from the script rather than from habit. The reliable method is to walk through the outline scene by scene, note each distinct emotional state the character must display, and group near duplicates. A script that calls for joy, grief, fury, shock, resolve, and embarrassment justifies six faces. A script that calls for joy twice and grief once does not justify two joys.
The standard six cover most leads: neutral, happy, sad, angry, surprised, and one role-specific state. Neutral deserves emphasis because it appears more than all others combined. It should be genuinely neutral rather than secretly stern or faintly smiling, since any baked-in emotion will contradict half the lines it accompanies. Happy and sad should each be calibrated to the story tone: a comedy wants broader takes, a tragedy wants restrained ones. Angry and surprised handle plot turns. The sixth slot belongs to the role — determined for a captain, flustered for a love interest, contemptuous for a rival, solemn for a mentor.
Blush, sweat drops, and similar overlays deserve a decision of their own. Some engines composite these marks at runtime, which means they should not be painted into the faces. Others need them baked in. The brief should state which approach the project uses before painting, because converting between the two later means touching every file. The same logic applies to alternate mouth shapes for lip sync, if the engine supports them.
A frequent error is ordering exotic expressions before the basics are approved. The correct sequence paints neutral first, approves proportions and palette against it, then derives the remaining faces from the approved base. Any structural fix applied to neutral propagates cheaply; the same fix applied after six faces are painted costs six times as much. The checklist on this page encodes this order by placing pose approval before batch painting.
Deciding the outfit count
Each outfit multiplies the entire expression set, so outfits are the most expensive decision on the sheet. One default costume plus one situational variant covers the great majority of scripts. Common second looks include sleepwear, formal wear, battle gear, seasonal wear, and work uniforms. The test for a second outfit is whether costume changes occur visibly in the story. If the script moves characters from school to festival to beach, the outfits earn their files. If the story spans three days in one location, they do not.
Outfit design should preserve the character across the change. Keep the hairstyle, the signature color, and the overall value pattern recognizable so players never wonder who entered the scene. A captain who trades her naval coat for evening dress should keep the brass fittings or the crimson lining that mark her. Total redesigns between outfits read as different characters at dialogue size and should be avoided unless disguise is the plot point.
Technical continuity matters as much as visual continuity. All outfits share one canvas size, one character height, and one eye-line height. The head should sit at the same vertical position in every file so expression swaps and outfit swaps never cause jumping during scene transitions. Artists sometimes let poses drift for variety, and modest pose variation is fine, but the head anchor and the total height must hold. The review step that catches violations is cheap: place every outfit side by side at actual dialogue size and look for jumping heads, wandering widths, and palette drift.
Palette and silhouette craft
A visual novel palette must do two jobs at once: look appealing in a large gallery view and remain legible in a small dialogue box. These goals conflict, because detail that delights at full size dissolves into noise when shrunk. The resolution is hierarchy. Choose one signature color at full saturation, one or two supporting tones at medium presence, and quiet neutrals for everything else. Test the result by shrinking the sprite to the size it will actually display during play. If the character still reads, the hierarchy works.
Value structure matters more than hue choice. Squint at the design until colors blur into lights and darks. A strong design still shows a clear light-dark pattern: dark coat against pale skin, bright scarf against dark hair. A weak design collapses into uniform mid-tone gray. Players perceive value contrast before they perceive color, so a sheet with sound values survives palette experiments while a sheet with weak values fails in every recolor.
Silhouette works the same way. Fill the character shape with solid black and examine the outline. A strong silhouette shows recognizable features — a distinctive collar, an asymmetrical accessory, a characteristic hair shape — while a weak one becomes an anonymous figure. Since dialogue boxes often crop the lower body, concentrate silhouette interest in the head, shoulders, and upper torso. A striking hemline that never appears on screen is wasted design effort.
Skin tone and hair color deserve deliberate range across the cast. Visual novels present many faces in sequence, and a cast sharing one skin tone and one hair color forces players to distinguish characters by costume alone. Varying these features is both a readability measure and a respect measure, reflecting the audiences these stories serve. The brief table on this page treats palette as a per-character note precisely so cast-level variety can be checked across sheets.
Illustrative example one: a hypothetical starship captain
Consider an illustrative hypothetical captain for a science fiction romance. The designer names her Captain Ilsa Renn, assigns the protagonist role, and requests 6 expressions with 2 outfits in an anime finish. The tool multiplies 6 by 2 into 12 files, returns a lean-scope verdict, and attaches an illustrative planning load around thirty hours.
The brief table gives her an upright open stance with a forward-facing default pose, a high-contrast lead base with one signature color, and clean cel rendering with bold lines. The designer chooses a deep navy uniform with brass fittings and a crimson scarf as the signature anchor. The second outfit is shore-leave casual: the same scarf over a cream sweater and dark trousers, preserving recognition through the anchor color and the hairstyle.
The expression list follows the script. The outline calls for command scenes, a confession scene, a loss scene, a betrayal confrontation, a comic misunderstanding, and steady bridge dialogue. Those beats map to neutral, determined, tender, grief-stricken, angry, and flustered — six faces with no filler. Because the tool froze the count at six, the designer resists adding a second angry variant when a late scene suggests one, and instead directs the existing angry face with dialogue and music. The sheet ships at twelve files, the cast budget survives, and the captain stays consistent from first line to finale. This example is illustrative and hypothetical, built to show the workflow rather than to describe any real production.
Illustrative example two: a hypothetical rival with scope trouble
Now consider a second illustrative hypothetical character, a rival named Vex Marlow in a painterly school drama. The designer requests 10 expressions and 4 outfits: school uniform, stage costume, winter coat, and festival wear. The tool multiplies 10 by 4 into 40 files, returns a heavy-scope verdict, and shows an illustrative load near one hundred hours for this single character.
The diagnosis is straightforward. Four outfits quadruple every expression decision, and ten faces include near duplicates — two smirks, two furies, and a pensive look the script calls for once. The designer applies the tool recommendations in order. First the pensive face and one smirk merge, bringing expressions to eight and files to thirty-two. Then the festival outfit folds into the stage costume with accessory changes, bringing outfits to three and files to twenty-four. The verdict improves to moderate scope. A final pass cuts the winter coat after confirming that winter scenes reuse interiors, leaving two outfits and sixteen files.
The painterly finish survives because the budget now fits it. Had the designer kept all forty files, the honest advice would have been to switch the cast to anime finish or to demote this rival to a supporting set. Scope and finish are one decision, not two, and the file total is where they meet. Like the previous case, this example is illustrative and hypothetical, showing how the verdict guides cuts while they are still free.
Edge cases and failure modes
Small inputs create their own risks. A character with one expression and one outfit totals a single file, which the tool calls lean. That verdict is arithmetically correct and artistically dangerous if the character speaks across varied scenes with a frozen face. Single-face sprites suit silent cameos, background figures, and very short appearances. Any speaking role with emotional range needs at least neutral plus two or three states, and the lean verdict should not override that judgment.
Large inputs fail differently. Twenty expressions and six outfits total one hundred twenty files, the maximum the tool accepts. No independent project should commission that for one character. Beyond the raw hours, quality control collapses: nobody reviews one hundred twenty files with fresh eyes, so drift ships. The correct response is structural, splitting the concept into two characters or accepting that some states belong to CG illustrations rather than sprites.
Name handling carries a quiet edge case. An empty name field produces an unnamed-character brief rather than an error, because the arithmetic does not depend on the name. The designer should still fill it in, since unnamed briefs get misfiled when a cast grows. Keep working names short and stable across the whole production, and rename files only at defined milestones.
Style mixing is the most common failure mode in team projects. Two artists painting in nominally the same finish will still diverge on line weight, shadow temperature, and fold density unless the brief pins those details. The remedy is a shared style page: three approved close-ups at full resolution showing an eye, a hand, and a fabric fold, with written values for line width and shadow colors. Every artist matches the page, not each other. Without it, each painter matches their own habits and the cast fragments.
Palette drift across monitors and exports is a subtler failure. Colors approved on one display shift on another, and compressed exports shift them further. The defense is numeric: swatches defined by written values, checked with a picker rather than by eye, and re-verified on the final export format. Approving by impression on a single screen guarantees surprises on player hardware.
Pose drift deserves its own warning. Artists naturally vary poses to keep work interesting, and modest variation flatters a gallery. In the engine, however, expression changes swap files mid-line, so any head movement between faces produces visible jumping. The rule is strict for the head and lenient elsewhere: identical head position and size across all faces of an outfit, with freedom in shoulders and hands. Outfit changes may shift the body more, but the eye-line height still holds.
File naming and delivery round out the failure list. A sheet delivered as forty loosely named images becomes a misassembly hazard during engine integration. The naming pattern should encode character, outfit, and expression in every filename, and the delivery should include a contact sheet showing all files at uniform scale. Integration then becomes checking rather than guessing, and missing files surface before launch week.
Revision policy is the contractual edge case. Unlimited revisions on a forty-file sheet is an unbounded liability for the artist and a schedule hazard for the project. The workable structure approves in stages — turnaround, then neutral face, then one full outfit, then the remainder — with limited revision rounds per stage and additional rounds billed. The tool checklist mirrors this order so the brief and the contract reinforce each other.
Commissioning and review workflow
A well run commission moves through five stages. First, the designer completes one brief per character with this tool and reviews the cast totals together, promoting leads and demoting cameos until the budget fits. Second, the artist delivers a turnaround and a neutral face for approval, which locks proportions, palette, and finish. Third, the remaining expressions for the default outfit are painted and reviewed at dialogue size. Fourth, variant outfits are painted against the approved anchors. Fifth, the full sheet is exported, named, and assembled into a contact sheet for final sign-off.
Reviews should happen at the size players will see. Full-resolution inspection flatters detail that vanishes in play and hides proportion breaks that dominate small display. Every review round should therefore include screenshots of the sprite inside an actual dialogue box over representative backgrounds. A face that reads against both a bright bridge and a dark alley is approved; one that fails either returns for adjustment.
Feedback to artists should cite the brief, not taste. Instead of asking for a prettier expression, point at the locked anchor: the shadow tone deviates from the swatch by this amount, or the eye-line sits so many pixels above the approved height. Specific citations keep revisions fast and preserve the working relationship. The brief table exists to make this possible, translating vague dissatisfaction into checkable claims.
Payment milestones should follow the same stages as approvals. A deposit with the turnaround, a payment with the default outfit, and final payment with delivery keeps incentives aligned and protects both sides. The illustrative hour estimate from the tool can inform scheduling discussions, but money terms belong in a written agreement with the actual artist, never in a calculator output.
Depth beyond the basics
Several questions recur often enough to deserve fuller treatment than a short answer allows.
One concerns the tradeoff between expression count and CG illustrations. Both depict emotion, but they serve different densities. Sprites carry frequent, low-intensity states across thousands of lines, while CGs carry rare, high-intensity moments at full illustration quality. A confession lands harder as a CG precisely because routine feelings stayed in sprite form. Overloading the sprite set with extreme takes therefore weakens the CGs by spending peak emotion at routine quality. The guideline is to keep sprites in the performable middle range and reserve breakdowns, embraces, and spectacle for illustrated scenes.
Another concerns accessibility and readability. Dialogue-size sprites must serve players with low vision, color vision differences, and small screens. Strong value contrast, distinct silhouettes, and expression differences carried by shape rather than color alone all help. A blush that differs from neutral only in pink tint fails players who cannot distinguish that tint; a blush that also changes the mouth and eye shape succeeds for everyone. Designing faces through shape first and color second is the accessible order.
A third concerns cast planning as portfolio strategy. Many developers design characters in story order, briefing the protagonist first and whoever appears next second. Budget order is usually wiser: brief the full cast at file-total level first, allocate faces and outfits by screen time, then design in detail. This prevents the late-story favorite from starving because early minor roles consumed the art fund. The per-character totals from this tool add directly into a cast budget, making the portfolio view a matter of addition.
A fourth concerns reuse across projects. Well specified sheets outlive their games. A cleanly layered, consistently named, swatch-documented sprite set can cameo in the next title, appear in promotional crossovers, or anchor a remake. Disposable one-off portraits cannot. The extra rigor of full briefs therefore pays compound interest across a studio catalog, and archiving the brief alongside the exports is what makes that reuse possible.
A fifth concerns the boundary between design and writing. Sprite briefs sometimes attempt characterization the script never supports, ordering a brooding scowl for a character with no brooding scenes. The sheet should follow the script, not lead it. If the design demands an expression the outline never calls for, either the outline wants enrichment or the list wants trimming. Resolving that tension in pre-production improves both artifacts.
Where the work continues
A finished brief is a handoff, not a finale, and the first handoff is to the writing itself. The sheets designed here move next into Co-Writer at slash write, where each character gets drafted inside real scenes instead of living only as a turnaround. When the expressions are locked and the cast totals fit the budget, open Co-Writer and write the cast route by route: the matrices tell you what each character looks like, the swatch sheets hold the palette steady, and the frozen expression lists stop mid-draft redesigns. Bring the briefs along, and the adaptation to a visual novel later starts from finished characters rather than sketches.