Untick what you never open: it leaves the rail and its card, and the
interface gets simpler. Hidden ones are still in Find a tool (Ctrl+K).
Your practice
Check your own image
Drop one image to view it full-screen with the tonal-value tools, or several to draw a timed session from them -
or a video, to draw from frames taken at random through it.
Your uploads
This browser has no WebGL 2, which the 3D forms need. Every other part of the board still works.
Click a form to select it, the selected one to light that spot · click the floor to throw the shadow there · drag to orbit · scroll to zoom · G / R / S move, rotate, scale the selected form, as in Blender
Prompt
Made here
What you generate shows here - click one to open it with every tool.
Prompt
Made here
What you make shows here, cut out - click one to open it with every tool.
Drop, paste or choose an image - a reference photo, a master painting, a painting of your own.
Paints
Gamut mask
Palette - colour, value and how to mix it
Point at a colour to see where it is in the picture; click to keep it.
No image yet.
Colour wheel - choose colours that go together
From the picture
Shadows are worked out the anime way: a value step darker and turned toward violet - the cool sky lights them - but skin's toward red, where light comes back out of it. Lights are paler and warmer. In watercolour a light is the same paint with more water, and an eye's catchlight is the paper.
Grey ladder - one mix, every value
Mixing chart - every paint with every other
Point at a square to see what it is; click it for the colour and what the mix does.
The sheet to paint
Mix this colour
Target Yours
Recipes you have hit
Glazing chart - one paint over another
Columns: the first wash, dry. Rows: the glaze painted across it. Click a crossing.
Light and shadow - what a light does to a colour
Skin - the four zones, in every light
The reflected light comes from the surface chosen above.
Pigment guide - what each paint does on the paper
Public-domain pictures from open collections. A painting opens in the workspace like a photo of your own, with every tool on it. Try:
Vanishing points
Size, mm×
Third point
The horizon is your eye level, and it fixes how tall anyone is anywhere in the picture. Drag the horizon and the white figure (its feet and its head) until the figure fits - on someone in a photo, or as tall as you drew them - then click the ground to stand people there.
m tall
Stand next
Time of day
Clouds
Under it
How fast a wash dries depends on the air. Set what a thermometer and a hygrometer say - or ask Open-Meteo (free, no key) about a place: only the place's coordinates are sent, and only when you press a button.
or press Space
Seasons and times of day
Harmony
Your paints
Saved
From a picture
A reference's own colours come from there.
Overlays hidden · o brings them back
Workspace
Layers
Your drawing
0:30
Every tool is pinned to the toolbar.
In the toolbar
Unpinned tools keep their keys, and Ctrl+K finds them.
space pause · ← → step · f flip · g gray · q squint · v value · +/- zoom · a angle · c guides · h contrast · e eyedrop · p pose · l head · x edges · b range · j toned paper · t temperature · u amounts · n tangents · y symmetry · z rhythms · k line weight · d compare · i info · r grid · o overlays off · w workspace · m similar · s skip · esc stop · ? help · ctrl+k find
↑ ↓ to choose · Enter to use · Esc to close
Theme editor
Changes show on the whole app as you make them. The picture, a
session's stage, value scales and paint swatches never take a theme's colours.
My materials
Tick what you have; the dot is what you are using now. The value
split's steps and the eyedropper then speak in its terms - a wash, a layer
of hatching, a pencil grade. The medium select in the info drawer switches
between the ones you have.
What's new
Earlier
Redraw a part
How different
Extend the picture
Which sides
How much
Cut out
Behind itLook
Before and after
BeforeAfter
How to draw it
Worked out from the finished picture - a guide to the order, not a record of how it was made.
Lightbox
Photograph your sketch - the whole sheet, from above - and trace it onto good paper laid over the screen.
Turn the screen's brightness all the way up, and tape the paper down.
Or drop a photo here. On a phone or tablet, Choose can take the photo too.
Corners Drag the four onto the corners of your sheet.
Measure the screen Hold a bank card - or any card that size, an ID, a driving licence - against the glass,
and make the outline exactly as wide.
Your data
Profile
Where it is kept
Backup
A backup is one .zip file: your settings, materials, themes, 3D scenes, practice log,
trainer scores and every upload. Keep it somewhere safe, or restore it on another computer.
Portfolio site
Your own work as a small gallery site - one page and its pictures, in a .zip - to publish on
GitHub Pages, as Refboard is. References are left out.
How to use Refboard
Browsing your library
The icons down the far left are the sections, grouped by the
stage of the work they are for - hover one for its name.
Prepare: the Library (shows or hides the folder panel),
All images and the Colour studio. Practise: Train and
3D forms. Paint: Start drawing - a timed session from the
packs you ticked - and My materials. Check: open any image,
a reference or a photo of your drawing, with the value tools. The
dashboard has the same four stages as cards, each with what is in
it. The
sliders icon top right shows or hides the session panel - its tabs are
Timing, Filter and Display, with the pose count
and Start always at the bottom. Both panels stay the way you leave
them. The 3D panel has tabs too: Object, Light, View and Use.
Find a tool - Ctrl+K, or the magnifier at the top of
the rail and in a session's toolbar: type a few letters of what you
want - "loomis", "flip", "green", "perspective" - and Enter. It finds
every section, every trainer and theme, and in a session every tool
and guide in the toolbar, also by the other words they go by. With
nothing typed, the ones you used last come first.
The brush near the bottom of the rail is My materials:
tick what you draw and paint with - watercolour, oil or gouache,
bottle ink, liners, ballpoint, watercolour pencils and markers,
graphite - and which one you are using now. Each gets darker in its
own way, so the tools speak in its terms: the value split names its
steps ("a light wash", "cross-hatching", "2B"), and the eyedropper
says which step a colour is, then, for paint, how to mix it. The
info drawer's Medium switches between your materials.
The moon near the bottom of the rail picks a theme:
Catppuccin Mocha (the default), Macchiato, Frappé and Latte, the
original Studio dark, and Daylight - black on white, for a tablet
outdoors. Edit colours... at the bottom of that menu opens the
theme editor: six colours, shown live on the whole app, with how
readable the text is on them; Save keeps it as your own theme, and
Export and Import pass it on as a small file. A theme colours the
panels and controls only: a session's
stage stays black, and the picture, value scales and swatches look
the same in every theme, so none of them changes how dark something
reads.
The library panel lists your packs. Clicking a pack or a folder
name opens it as a grid in the middle; clicking its
checkbox includes it in the pool a timed session draws from.
They are separate on purpose - browsing a pack to decide whether you
want it is how you end up ticking it.
In the grid, the search box matches file, folder and pack names, and
the box beside it sets the order: the folder's own (numbered poses are
usually numbered for a reason), by name across folders, lightest or
darkest first to pull a value study out of a pack, or shuffled -
picking Shuffle again deals a new order.
Clicking any thumbnail opens it full-screen in the same viewer a
session uses - zoom, the angle tool, construction guides, the
tonal-value split, the eyedropper, all of it - with ← →
walking that folder in order and no clock running. Esc returns
to the grid, on the image you stopped at. Draw these above the
grid starts a real timed session from exactly what is on screen, search
filter included.
The grid takes the keyboard too: arrow keys move between
thumbnails, Home / End jump to the first and last,
Enter opens one and S skips it.
Skip in a thumbnail's corner (or S in the viewer)
marks a pose never to draw again. Skipped images stay in the grid,
dimmed, so you can undo it.
Timed sessions
Tick one or more packs in the rail (the filter box above the list
narrows it by name for a large library, without changing which packs
are already ticked), then set up the right-hand panel: an interval -
a preset, a custom number of seconds, or a structured
warm-up/quick/long-study schedule (or build your own under
Custom: any number of steps, each its own seconds-per-pose and
pose count, saved for next time) - and any options: grayscale,
random mirroring, a grid overlay (rule of thirds, golden ratio,
diagonals, a center cross, or the classical armatures - dynamic
symmetry's diagonals and reciprocals, the harmonic armature, and
rabatment's squares - drawn to each image's own proportions), a
ghost of the previous pose, a
focal-point marker, a bell when the pose changes, fullscreen. Tone
filtering (high-contrast / low-key / high-key) and skipping
near-duplicates are both computed from your own library, not a fixed
guess. Click Start drawing.
During a session
Space pause/resume · ← → previous/next pose ·
F flip horizontally · G grayscale · Q squint ·
S skip this pose for good · Esc stop. The HUD's own
buttons do the same thing for a mouse.
Your toolbar - the session toolbar holds the tools you
pin; the rest are under More (the three dots), one click
away. At the bottom of More, every tool is listed: a lit one is in
the toolbar, and clicking it moves it to More or back. An unpinned
tool keeps its key, and Ctrl+K still finds it. Reset puts
back the first set.
Workspace - w, or the sliders in the toolbar: the
session's tools sorted by the question you are asking of the
picture. Value - how light or dark; Colour - what
colour and how to mix it; Construction - where things are and
at what angle; Figure - how the body is built; My work -
whether your drawing is right. Each row is the toolbar's own tool,
on or off the same as from its key.
Layers - the stack icon in the toolbar lists every overlay
that can lie over the picture: construction guides, angle lines, the
focal point, the ghost of the last pose, the head construction, the pose
skeleton, the grid and your drawing, top to bottom as they stack.
Each has a checkbox - the same as its own button or key - and its own
opacity, kept for next time, so several can be on at once without
drowning the picture: a faint grid under a strong skeleton. The
number on the icon is how many are on. O hides them all at
once without turning any off - a clean look at the picture - and
O again brings them back.
Tonal value / Notan study
The HUD's value select (or V to cycle) posterizes the
reference to 2-6 flat tones, shown side by side with the original -
block in light and dark masses before chasing detail. Two tones is a
pure Notan (light mass vs. dark mass); more tones reads closer to a
full value scale. For a quicker check without leaving the plain
original, Squint (Q) just blurs it directly - the
fastest version of holding a reference at arm's length and
unfocusing your eyes to see its big shapes and values.
Understand the pose
Open the Info drawer (I, next to the value select) for
everything about the current pose rather than the session itself:
Scheme - swap the tonal-value view's colour: grayscale,
three duotones (Van Dyke brown, Payne's grey, sepia/burnt umber),
a false-colour heatmap, or outlines only (for hard and soft
edges, the Edge map, x).
Blur - softens the reference before posterizing, so tone
blocks in as soft masses instead of hard edges around every small
bit of local contrast.
A legend showing exactly what each tone level maps to in
the current scheme, a brightness histogram, and the pose's
top-5 dominant colours.
Eyedropper (E, its own HUD button) - click anywhere on
the reference to read the exact hex colour and brightness percent
under the cursor, and several ways to mix it from real paint -
burnt sienna and ultramarine, white and phthalo blue - in parts by
volume, each with the colour that mix gives and how close it lands.
Medium says how you paint: in watercolour there is no
white - a recipe says how strong a wash to make (from barely diluted
to a pale tint), and white means leaving the paper; in oil or
acrylic lighter means white paint. Paints picks the tubes
they may use: a full palette, a split
primary, three primaries, Zorn's four, the earths. The mixing is
worked out the way paint mixes - by what each pigment absorbs across
the spectrum (Kubelka-Munk, via spectral.js) - so blue and yellow
make green and complements make greys, as on a real palette. Brands
differ and screens differ: take it as where to start, then judge by
eye. The last few samples stay listed
as small swatches next to the reading - click one to copy its hex,
to compare two earlier spots (skin vs. background, say) without
re-aiming at either one again.
Two more, both Options-screen checkboxes since they apply to every
pose in the session rather than being switched pose to pose:
Ghost previous pose overlays a faint edge-detected silhouette
of whatever was on screen just before this one, to help track how a
gesture or form carries from pose to pose; Focal-point marker
marks this pose's own visual-weight centroid - a rough guide to
where the eye tends to land first, from its actual brightness
distribution rather than a guess.
Zoom and pan
Scroll the wheel, or use the HUD's − / 100% /
+ (also -/+/0 on the keyboard), to zoom
into a detail - a hand, a face - up to 6x. Once zoomed in, drag
anywhere to pan; double-click, or the 100% button, resets.
Zoom and pan reset automatically on every new pose.
Angle tool
Toggle it with the HUD's Angle button (or A), then
drag anywhere on the reference the way you'd hold a pencil up to
check a shoulder or hip line before committing to it on paper. The
first drag after turning it on also becomes a reference
length - drawn faint and dashed once a second line exists - and every
drag after that reads its own angle and its length as a ratio
of that reference (a head height, a hand span). A few of these
comparison lines stay on screen at once (the oldest quietly makes way
for a new one past that), so you can measure a shoulder line and a
hip line and read both together instead of one at a time. Clear
lines, or turning the tool off and back on, wipes the slate for a
fresh reference. High-contrast overlays (Options, or H)
swaps the colour/opacity used to tell lines apart for a black-outlined
white line with its own dash pattern per line - the grid overlay gets
the same outline, more visible in bright light either way.
Construction guides
The HUD's guide select (C cycles) offers four more
traditional construction aids, one at a time:
Vanishing point - drag (following the cursor live) to drop
a point and fan guide lines out from it across the whole stage,
for perspective construction on a box, a room, or a foreshortened
limb.
Plumb line - drop a full vertical+horizontal crosshair
through a point, the digital equivalent of holding a pencil
straight up to check whether one landmark sits directly over
another.
Proportion divider - drag out one unit the same way the
angle tool's own reference length works, and the tool marks off
repeating multiples of it automatically along the same line - "how
many head-heights tall," measured instead of eyeballed.
Viewfinder - drag a frame over the reference: everything
outside it dims, thirds divide it, and its proportions are named
(4:5, 3:2...) - find a stronger picture inside the photo, cropped to
the shape of your paper. The Thumbnails schedule is the
exercise to use it with: eight small composition sketches.
Perspective check - finds the vanishing points from the
picture itself. Click two points along an edge, then two along an
edge parallel to it in the scene (the top and bottom of a window):
where the two lines meet is a vanishing point. Four more clicks on
another direction make a second, with the horizon through both and
how far it is tilted; it then says what field of view that implies,
and warns when the vanishing points are so close that square
things will look stretched - or on the same side of the centre,
which no square corner can do. A third direction finds the centre
of vision, which for an uncropped photo is the middle of the frame.
Works on your own drawing too - photograph it and open it with
Check your own image. Backspace undoes a point.
Only one guide is ever active at a time, and switching to a guide
turns the angle tool and eyedropper off (and vice versa) - all three
want the same drag gesture for different things. Selecting "Off," or
switching guides, clears whatever was drawn.
Pose skeleton
Pose (P) finds the figure's joints with a pose model
and draws over it what a teacher would: the tilt of the shoulders
and hips (and whether they tilt against each other -
contrapposto), the line of action in red - one curve from the
head down the spine to the leg that carries the weight, the gesture
to draw first - and the balance line in blue, a plumb line
from the pit of the neck, which on a figure at rest lands on the
foot bearing the weight. It stays on for every pose until you turn it
off. The switch at the top of its note redraws the pose in the 3D
figure's proportions - Anime (7 heads), Long-legged (9),
Chibi (2.5): every bone at the photo's angle, only its length
changed, standing on the same foot - in pink over the photo's own
skeleton, with a tick for each head down its side. The model runs in this browser; the first time, it is
downloaded once (about 6 MB). The image is never sent anywhere.
Head construction - Loomis and anime
Head (L) draws Andrew Loomis's construction over every
face in the picture, turned the way that head is turned: the
cranium as a ball, its sides sliced flat into the side
plane (blue - the ear sits just behind its middle), the
centre line (red) over the ball and down the face, and the
face in thirds (yellow) - hairline, brow, base of the nose,
chin. Lines on the far side of the head are dashed. The note says
how far the head is turned, tipped and tilted, and what that does to
the features. Draw the construction first, then check it against
this. It needs the face fairly large in the picture; like the pose
model, the face model runs in this browser and is downloaded once
(about 4 MB).
Anime, at the start of the note, keeps the ball, the side
plane and the centre line - how the head is turned - and draws the
face as anime does (pink): the eyes lower and much bigger, one eye
apart, with a heavy upper lash line and a tall iris; a small nose and
mouth close under them; the jaw nearly straight to a pointed chin.
Take a photo of anyone at a hard angle - from below, three-quarter,
tilted - and see where anime features sit on it. The choice is
kept.
Edge map - hard and soft
Edge map (X, or Workspace > Value) draws every edge
of the picture worth painting, coloured by how hard it is: red
where the change happens at once, blue where it fades over a
distance. Hardness is how quick the change is, not how big - a soft
edge between black and white is still soft. In watercolour, red is
laid on dry paper, blue run in wet-in-wet or softened with a damp
brush; the note says it in the terms of your medium (My materials),
and how the hard and the soft are shared out. Its opacity is in
Layers. For watercolour, How to draw it has the same map as a step.
Build up in stages
Tick Build up in stages under Timing for copying a master
painting or a long study. Each new pose is shown first as two masses
of light and dark, then as three values, then as soft masses of
colour, and only in the last stretch as it really is - the order a
painting is built in, so the eye cannot go to the detail first. The
interval is shared between the stages; → moves to the
next one early. It and memory drawing are one or the other.
Frames from a video
Drop a video on Check your own image (or pick one) and twenty
frames are taken from it at random - one from each stretch of the
clip, so they cover all of it - to draw a session from. Dancing,
sport and animals give poses in real movement that no posed photo
has. The frames are made in this browser; the video goes nowhere.
Draw from memory
Tick Draw from memory under Timing and every new pose is shown
only for the study time, then covered. The interval is your drawing
time; when it runs out - or you click the cover, or press → -
the reference comes back and the clock stops, so you can check.
D lays a photo of your drawing over it, the same as Compare
below. Previous shows an earlier pose plainly.
Compare your drawing
Compare (D) lays a photo or scan of what you drew over
the reference - pick the file, drop it on the stage, or paste it. The
clock pauses. Lines in red lifts just your marks off the paper
(an unevenly lit phone photo of grey paper is fine) and draws them in
red over the reference; See-through photo and Blink are
the other two ways to look. Drag to move it, Shift+scroll to
size it, Alt+scroll to turn it. Line up 2 points is the
quick way: click two landmarks on the reference (top of the head, a
heel), then the same two on your drawing, and it is sized, turned and
moved so they meet - what is left between them is your proportions
and angles as drawn. D hides it for a moment; it goes away with
the pose, and never leaves this browser.
For a painting, the last three choices (Where it parts)
are a map of tones, not lines: where the painting is darker or
lighter than the reference, warmer or cooler, stronger or
greyer in colour. A phone photo is never lit like the reference, so
its overall light and colour cast are matched first - a dim room does
not count - and the note says where the painting parts most, in
hundredths of a value step. Line the photo up first; the map follows it.
Clean scan... turns a phone photo of a sketchbook page into a
scan: click the four corners of the page (top left, top right, bottom
right, bottom left) and it is straightened, the shadow across it and
the lamp's tint are taken out and the paper comes out white. Choose
Painting for washes, so a wide dark patch is not taken for
shadow. The scan replaces the photo here and is kept in My work, tagged
digitised, with the reference it was drawn from.
Colour studio
Drop, paste or choose any image - a reference, a master painting, a
painting of your own. On the wheel, hue runs round and chroma
(how intense a colour is) out from grey in the middle; every pixel of
the image is a dot on it, so you see at once which hues it lives in
and how grey it really is. The numbered circles are its
palette, found from its pixels and sized by how much of the
picture each covers; the list beside gives each one with its
value as grey, and warns where two different colours have the
same value - they will merge in a black-and-white study.
Value shows the whole image that way.
A gamut mask (James Gurney's tool) is a shape on the wheel
that limits which colours a painting may use. Pick one, drag inside
it to turn it to other hues, drag a corner to reshape it, and
In gamut repaints the image with only the colours inside it,
its values untouched - try a palette on a picture before committing
paint to it.
Paints chooses a palette of real tubes - the same choice as
the eyedropper's. The dashed outline on the wheel is every colour
those paints can mix, and the note says how much of the picture lies
inside it: Zorn's four reach flesh and warm greys but no true blue or
green, so a sky has to become a cool grey - which is the point of a
limited palette. Under each palette colour is how to mix it from
them; click it for other ways. Point at a palette colour and the
picture shows only where it is used, the rest dimmed; click to keep
it that way. The image never leaves this browser.
Your practice
The dashboard counts what you have actually drawn: poses, time,
sessions, and the run of consecutive days you are currently on - a day
you have not started yet does not break it. The calendar under them is
the last thirteen weeks, one square per day, brighter for a heavier
day, measured against your own busiest one rather than a fixed number.
Hover a square for the date and count.
Only timed sessions count. Browsing the grid is looking, not drawing,
and is deliberately never logged. The most recent 50 sessions are kept
- enough for months of practice - and they sync between your devices
along with the rest of the setup where that is configured.
Check your own image
The drop zone on the setup screen works with no library at all -
it's the whole of what runs on the
live demo.
Drop one image to view it full-screen with every tool above, paused
with no timer running; drop several to run a real timed session
against them. Nothing is ever uploaded anywhere - it stays local to
this browser.
3D forms
Simple forms - cube, sphere, cylinder, cone, pyramid, prism, torus,
capsule, egg, or a model of your own - under light you control: the
forms every shading lesson starts with, where light, halftone, core
shadow, reflected light and cast shadow are easiest to tell apart.
Needs no library; the 3D engine (three.js) is fetched from a CDN the
first time you open the view.
Forms: up to six on the floor at once - Add form, pick
one by its number (in the panel, or the badge over it in the view), and
its shape, surface, placement, proportions and rotation are what the
panel then edits. Forms cast shadows on each other. Load
model... (or drop the file on the view) takes a .glb,
self-contained .gltf or .obj - a mannequin, a planes-of-
the-head bust, a skull - and treats it like any other form. The
Figure shape is a jointed wooden mannequin: click a dot on a
joint (or a part of it) to pick it, then drag its rings, press
R, or use the Bend / Twist / Lean sliders under Pose;
click the picked dot again for the whole figure. Pose has the
classic poses, Random pose, Mirror and Reset;
Alt+R straightens one joint. It stands on whatever part of it is
lowest, so a kneeling figure kneels. Above the pose, its
Body: Realistic, Anime, Long-legged or Chibi proportions - the pose
carries over - and Heads grid, a numbered line every head's
height, measured from the floor as it would stand straight; Bony
landmarks and Muscle groups label what an artist looks for
under the skin, on the figure, and turn with it - only what faces you is
drawn. A model file stays in
the browser, and only until the page is reloaded. Contour lines on a
model need texture coordinates in the file.
Head is a real head - a 3D scan - and Head planes is
the same head cut into a few hundred flat facets, the way the Asaro
planes head is: each turn of the form is one plane and one value, so
it shows where the light turns into shadow on a face and why. The
scan is fetched the first time you pick a head (about 400 KB, once);
until then, or offline, a simpler sculpted head stands in. Contour
lines on both run round the head and up it, like calliper lines on a
sculpture. Head scan: Lee Perry-Smith, Infinite-Realities
(CC BY 3.0),
via the three.js examples. Hand poses like the figure, joint by joint - the wrist,
three joints on each finger and the thumb - with its own poses:
relaxed, spread, fist, pointing, pinching, gripping. Foot is
the same for a right foot - the ankle, the middle of the foot and every
toe - with relaxed, pointed, flexed, on-the-toes, curled and spread
poses. In Draw random forms a head nods, tilts and turns and a
hand or a foot takes a new pose, instead of being swapped for a random
shape.
Anime head is built the anime way - the same proportions as
the Anime style of Head construction on a photo: a big round cranium,
the jaw almost straight to a pointed chin, a nose that is barely there,
and the eyes drawn on - the heavy upper lash line, the tall iris
darker under it, the gleam on the side the light comes from. Its face
is a flat mask, not the front of a ball, which is why anime eyes do
not turn the way real ones do: the note under the view gives the far
eye's width against the near one's at the angle you see it from, and
what it would be on a round head, with what to watch at that angle.
With a ComfyUI, Draw it at this angle asks Generate for anime
heads from the same side, to hold against the 3D one. The shapes are
listed in kinds: forms, figure and head, cloth, your models.
Its Hair (Object tab, while the anime head is selected) is in
locks, as anime draws it: a few pointed ribbons with some thickness
over a mass that hides the scalp, each lying on the skull and then
hanging from the widest point it passed - Short, Bob or Long, or none.
Its light is the ring anime paints: a hair is a line, and a line
shines where it lies square to the light, so the shine is a band round
the head, broken where each lock turns - turn the head or move the
light and it slides. On an Anime head it is cel shaded, and the fringe
throws a hard shadow of its own shape on the forehead. The colours are
named as Generate names them; your character sheets add their
characters' hair colours, and the form's own colour is its Skin.
Draw it at this angle sends the hair along.
Cloth over a ball, Hanging cloth and Curtain
are drapery, each one of the fold types drawing books teach: drop
folds falling from what holds the cloth up, diaper folds sagging
between two points, pipe folds running down from a gathered edge.
The cloth is simulated falling the first time you pick one - a
moment's wait, once - and lit and shaded like every other form.
Contour lines follow the weave, which is the easiest way to see how
a fold turns.
Move, turn and stretch a form in the view, the way Blender
does it. The selected form carries handles - Move,
Rotate or Scale, picked top-left of the view (pressing
the lit one hides them): drag an arrow, a ring or a square along its
axis, the middle square to slide anywhere on the floor, the white
circle to turn about your line of sight or scale evenly. Or, with the
pointer over the view, press G, R or S and move
the mouse; then X, Y or Z to lock an axis (twice
for the form's own axis while turning), type a number for an exact
amount, hold Ctrl to snap, and click or Enter to
keep it - right-click or Esc puts it back. Z is up and Y
into the picture, as in Blender; moving on Z lifts a form off the
floor (never through it), as does Lift under Placement.
Click a form to select it - a click on the one already selected
lights that spot, as before. Shift+D duplicates the form,
Delete removes it, Alt+G / R / S resets its place,
rotation or size, Ctrl+Z / Ctrl+Shift+Z undo and redo
any of these. The axis ball top right shows which way X, Y and
Z point: click a ball to look along that axis from its side - Z from
above, Y from behind - and click the one facing you to see the
opposite side. The numpad's 1, 3, 7 do the same
for front, side and top, and 4 6 8 2 orbit in 15° steps. The keys work on any keyboard layout.
Surface:Matte (plaster - no highlight), Satin
(a broad soft highlight), Glossy (a small sharp highlight on top
of the colour), Metal (mirrors its surroundings), Glass
(see-through; its cast shadow stays solid here) and Velvet
(brightest at the edges - the reverse of matte). Shine
fine-tunes any of them.
Drag to orbit and scroll or pinch to zoom. The lights
stay where they are while you walk around the forms, the way lamps in a
studio do - the Direction sliders update to say where they now are
relative to you. Reset view under Camera puts the camera back.
Aim the light in the view itself:click the selected form to make
that spot the brightest one; click the floor to throw the
shadow there (a far click is a low, long shadow). The yellow handle is
the light: drag it and a lamp follows the pointer - pulling it
away from the forms moves it further off, and its "lamp ×" number
changes as it goes; scroll over it - or Shift+scroll
anywhere - to move it nearer or further without moving it sideways.
The number is its distance in multiples of the scene's radius. All the
way out it becomes the sun: parallel rays, a shadow the width of
the form. Close in, a lamp: the side nearer it visibly brighter than
the far side, and a shadow that fans out. The lamp stands off the form
you last lit and goes with it when you move that form. A click on
another form only selects it - the light stays put until you click
the selected form or the floor. Brightness at the forms is
kept the same either way, so what changes is the character of the
light, not the exposure. A lamp out of the picture gets a hollow handle
on the frame edge that works the same way.
Second light adds a fill (soft, from the other side of the
camera - lifts the shadow side) or a rim (from behind - a bright edge
that separates the form from the background), shown as a small blue
dot, with a shadow of its own (untick Casts a shadow to leave it
out). Fill under Ambient lifts every shadow evenly; Bounce
is reflected light coming up off the floor in the floor's own colour -
set it to zero to see what it was doing inside the core shadow.
Occlusion darkens where a form shuts the sky out - a soft halo on
the floor round it, and a dark seam where two forms meet or one rests on
the floor; only the sky's light is dimmed, not the lamp's.
Light and shadow zones paints the families straight onto the
forms - highlight, light, halftone, terminator, core shadow, reflected
light, cast shadow - computed from the actual light and its shadows,
with a key in the corner. On the floor only the cast shadow is marked.
Perspective: the 1-point / 2-point / 3-point buttons
square the forms up and place the camera so exactly that many sets of
edges converge; Lens sets how hard they converge - 18mm is
steep, 200mm nearly parallel, and the forms stay the same size either
way. Vanishing points draws the box around the selected form
with every edge run out to where it meets, one colour per direction (a
point off the picture is pinned to the frame edge). Ellipses and
axis draws a round form's cross-sections and its axis - every
ellipse's minor axis lies along it, the rule a leaning cylinder most
often breaks. Floor grid, Eye-level line and
Cross-contour lines do what they say.
Anime shots, under the perspective buttons, move only the
camera - its height, lens, distance and roll - and leave which side of
the figure you see to you. Worm's eye goes down to the floor and
looks up (with the floor showing, the camera can go below the forms'
middle as far as the floor, and no further); Bird's eye looks
down from high up; Wide, close pushes a wide lens in so the
nearest hand or foot comes out huge; Dutch angle rolls the
camera - the Roll slider sets how far; Telephoto
flattens the depth; Fisheye bends straight lines round the
middle, 180° across the corners, with the Lens slider resting until it
is off. The eye-level line tilts and bends with them. Draw this
shot sends the angle and the lens to Generate's From and Lens
rows, with your ComfyUI.
Open in viewer sends the render to the full-screen viewer with
every tool above - try the tonal-value split on it. Draw random
forms renders a set of random poses and lighting and starts a timed
session with them, using the Interval and Schedule on the right; tick
Any shape to vary the forms and their proportions too.
Memory drill shows a fresh random scene for the time you set,
hides it while you draw it from memory, and shows it again when you
click. View drill sets a random scene and a task - draw it from
the other side, from above, under another light or through another
lens - and Show the answer swings the view to exactly that;
Check in viewer then opens the answer where Compare can lay a
photo of your drawing over it. Guides, zones and handles are never
part of these renders.
Clean (top left, or Shift+Alt+Z as in Blender) hides
every handle, number, skeleton and guide, leaving just the lit scene;
press it again or Esc to come back.
Saved scenes keeps named setups in this browser; Copy
link puts the whole scene in a link that opens it anywhere - on the
tablet, or for someone else (your own models are not included).
Train
Short drills, ten tasks a round, each scored out of 100 with the
answer shown and a line on why - your best and recent rounds are kept
on the cards. Enter goes on, Esc goes back to the list.
Value - how light is the patch? Pick its step 1-9 on a grey
scale, against surrounds built to fool you (simultaneous contrast,
gradients, strong colour), then with coloured patches.
Colour - mix the target on hue, chroma and value sliders;
the answer says which of the three was off and which way. The second
half shows the target on a coloured surround.
Anatomy - click the height of a landmark on an eight-heads
figure (chin, nipples, navel, crotch, fingertips, knees...) or on the
head (eyes, brow, nose, mouth, hairline); the canon is drawn in after.
Angles and proportions - turn a line to copy an angle; say
how many times one length goes into another.
Lines and ellipses - one stroke each: a straight line
between two dots, a circle, an ellipse of a given degree. Scored on
how far the stroke strays, with the true shape drawn over it.
Perspective - click where a box face's edges vanish, or
where eye level runs across a box in two-point perspective.
Gesture - a random pose on the 3D figure for 30 s, 1 or 2
minutes, then hidden while you finish; mark yourself against it, or
open it in the viewer and Compare a photo of your drawing.
Negative space - the empty shape by the dot: pick it out of
three, the other two the same area stretched or turned.
Tone map - on a picture from your library (or a render):
which of two spots is darker, where the darkest dark and lightest
light are, judged on the picture squinted.
Colour temperature - which of two is warmer or cooler; warmer
means leaning nearer orange, whatever the value or intensity.
Edges - the hardest or softest edge of four, with contrast set
to fool you.
Ellipses and centres - turn a cylinder's end ellipse to fit
its axis; click the true middle of a rectangle in perspective.
Your uploads and your data
Everything you give the app is kept: pictures and videos dropped on
Check your own image, photos of your drawing in Compare, the
Colour studio's pictures. They are listed under the drop zone as
Your uploads - click one to open it again in the workspace (a
video gives a new set of frames), or the cross on it to forget it.
The line above the list says plainly where they are kept.
Behind the container they are saved on its disk, with your
library, and join it as the Uploads pack - tick it and draw from
them like any pack. On the web page there is no server: they last
until the page is reloaded, unless you tick Keep in this
browser, which keeps them in this browser on this device.
Your data... beside the list (or Ctrl+K, "backup") makes a
backup: one .zip with your settings, materials, themes, 3D
scenes, practice log, trainer scores and every upload. Restore
from a backup reads one, says what is in it and asks before it
puts it back. Keep one somewhere safe, or use it to move to another
computer - from the web page to the container or back.
What's new - after an update, a short list of what came with
it, each with a button to try it. Again from the footer, Help or
Ctrl+K.
Rescan library
Shown on the setup screen once a library index has loaded. Click it
to ask the container to reindex right away instead of waiting for
its next scheduled pass - useful right after dropping a new pack
into the mounted folder.
All keyboard shortcuts
Space pause · ← → step · F flip
· G gray · Q squint · V value levels · +/- zoom · 0 reset zoom ·
A angle tool · C construction guides · E eyedropper · D compare your drawing · I info drawer ·
R grid · H high-contrast overlays · M similar (when available) ·
P pose skeleton · L head construction · S skip · Esc stop · ? this help
Everywhere: Tab and Shift+Tab move
between controls, with a ring round the one you are on; in a dialog they
stay in it, and Esc closes it. ↑ ↓ walk the rail of sections, Home and
End go to its ends. The first Tab offers Skip to the content,
past the rail. A screen reader is told which section you opened and
which pose is up.