Skill: Painting Depth & Perception
What
Classical and modern techniques for creating convincing depth, volume, and spatial perception in 2D images — translated into prompt language for AI image generation.
Why
AI models generate flat-looking images by default. Without explicit depth cues, scenes look like theater backdrops — everything at the same visual distance. These techniques, drawn from 600 years of painting and photography, give the model precise instructions for creating images that feel three-dimensional. Combined with the Blender Camera Mindset, this turns flat renders into inhabited spaces.
1. Atmospheric Perspective (Aerial Perspective)
The #1 depth cue in landscape and room-scale scenes. Farther objects appear:
- Lighter (more white/grey mixed in)
- Less saturated (colors wash out)
- Less detailed (edges soften)
- Bluer (warm colors shift cool with distance)
Prompt Language
| Distance | Prompt |
|---|
| Foreground (0-2m) | "sharp detail, rich saturated colors, crisp edges" |
| Midground (2-5m) | "slightly softer, colors a touch muted" |
| Background (5m+) | "soft and hazy, desaturated, edges dissolving into atmosphere" |
| Far background | "barely visible through atmospheric haze, almost monochrome blue-grey" |
Example: "The doorframe in the foreground is sharp and warm — dark wood grain visible. The bed 4 meters away is slightly softer, colors cooler. The window behind the bed glows with diffused light, curtains barely defined."
When NOT to Use
- Interior close-ups (room too small for atmospheric effect)
- Stylized/flat art styles (Moebius, pop art) where flatness is intentional
2. Overlap and Occlusion
The simplest depth cue: objects in front block objects behind. ALWAYS specify what blocks what.
Prompt Patterns
- "The girl partially blocks the view of the bed behind her"
- "His shoulder overlaps the doorframe at the left edge"
- "The IV stand is visible behind and to the right of the bedside cabinet, partially hidden"
Describe the scene in layers, front to back:
LAYER 1 (closest): [object], sharp, large in frame
LAYER 2 (mid): [object], partially occluded by Layer 1
LAYER 3 (far): [object], partially occluded by Layers 1-2
LAYER 4 (background): [environment], visible in gaps between foreground elements
Key: Every layer must occlude the one behind it somewhere. If nothing overlaps, the brain reads everything as the same distance.
3. Size Gradient (Diminishing Scale)
Objects of known size appear smaller with distance. The brain uses this to calculate depth.
Prompt Language
- "The doorframe fills the left and right edges of frame (close). The bed is small at the far wall (4m away)."
- "Her hand, close to camera, appears larger than her head in the background"
- "Floor tiles receding — large and detailed near camera, shrinking toward the far wall"
The Size Ratio Rule
An object 4m away appears roughly 4x smaller than the same object at 1m. State this explicitly:
- "The cabinet at 1 meter is large and detailed. The identical cabinet at the far wall is a quarter the size."
4. Linear Perspective (Converging Lines)
Parallel lines converge toward vanishing points. The strongest architectural depth cue.
One-Point Perspective
Best for: hallways, roads, looking straight into a room
- "Hospital corridor stretching ahead, floor lines and ceiling edges converging toward a single point at the far end"
- "The room seen from the doorway — floor lines, bed rails, and ceiling all pointing toward the window at the far wall"
Two-Point Perspective
Best for: corner views, buildings, complex interiors
- "Standing at the corner of the room, walls receding to left and right, edges converging toward two separate vanishing points"
Prompt Pattern
[parallel elements] converging toward [vanishing point location], creating [strong/subtle] depth
5. Value Contrast and Chiaroscuro
Dark-light contrast creates volume. Adapted from Caravaggio, Rembrandt, Vermeer.
The Depth Rule
- High contrast = close/important — sharp light-dark transitions on foreground subjects
- Low contrast = far/ambient — background elements have compressed tonal range
Prompt Language for Lighting Depth
| Technique | Prompt |
|---|
| Rim light | "thin bright edge along her shoulder and hair, separating her from the dark background" |
| Under-lighting drama | "face lit from below, deep shadows in eye sockets, dramatic upward shadows" |
| Window light falloff | "bright near the window, light falling off rapidly, far corner of room in deep shadow" |
| Bounce light | "soft reflected light filling shadows on the shadow side of her face — not pitch black, slightly warm" |
The Vermeer Pattern
Vermeer's depth secret: one strong light source (window, left side), subjects arranged in layers of light and shadow.
Prompt: "Single light source from the left window. Subject closest to window is brightly lit. Second figure, further from window, receives less light — face half in shadow. Background wall barely touched by light, deep and muted."
6. Color Temperature Depth
Warm colors advance (feel closer), cool colors recede (feel farther).
Prompt Application
- Foreground: "warm tones — golden skin, warm wood, amber light"
- Midground: "neutral tones — true colors, balanced temperature"
- Background: "cool tones — bluish shadows, grey-blue walls, cool ambient light"
The Complementary Contrast
Place warm subjects against cool backgrounds for maximum depth pop:
- "Warm golden-lit figure against cool blue-grey hospital wall"
- "Red dress against muted green garden — the figure pops forward"
7. Texture Gradient
Near surfaces show texture detail. Far surfaces are smooth/uniform. The brain reads this transition as distance.
Prompt Language
- "Close floor: individual linoleum tiles visible, slight scuff marks, reflection of overhead light"
- "Mid floor: tiles still visible but smaller, less detail"
- "Far floor: smooth white surface, individual tiles no longer distinguishable"
Applied to Materials
| Material | Near prompt | Far prompt |
|---|
| Fabric | "weave pattern visible, individual threads, wrinkles and folds" | "smooth colored surface, general drape visible" |
| Skin | "pores visible, fine hair, subtle color variations" | "smooth tone, general form only" |
| Wood | "grain pattern, knots, surface texture" | "uniform brown surface" |
| Metal | "scratches, fingerprints, reflections of nearby objects" | "smooth reflective surface, general highlights" |
8. Depth of Field (Selective Focus)
Mimics camera lens behavior. Sharp focus on subject, blur before and behind.
Three Depth Zones
FOREGROUND (before subject): soft blur, out of focus
SUBJECT (focal plane): perfectly sharp, crisp detail
BACKGROUND (behind subject): soft bokeh, dreamy blur
Prompt Patterns
- Shallow DOF (portraits, emotional close-ups): "shallow depth of field — her face razor-sharp, the room behind dissolving into soft bokeh circles"
- Deep DOF (establishing shots, environments): "deep focus — everything sharp from foreground doorframe to background window"
- Rack focus effect: "foreground hand blurred, background face sharp" (or vice versa for the next panel)
When to Use Each
- Shallow DOF → emotional moments, character focus, intimacy
- Deep DOF → establishing shots, complex compositions, POV sequences (eyes don't have bokeh)
POV WARNING: First-person POV panels should generally use deep focus (human eyes don't produce bokeh). Only use shallow DOF in POV when the character is focused intensely on one object.
9. Cast Shadows as Spatial Anchors
Shadows pin objects to surfaces and show spatial relationships.
Prompt Language
- "Her shadow falls on the floor, stretching toward the bed — placing her 2 meters into the room"
- "The IV stand casts a long shadow across the bed sheet, confirming it's between the window and the bed"
- "His hand's shadow visible on the table surface before his hand reaches it — hand hovering 6 inches above"
Shadow Direction = Light Direction
Shadows always point AWAY from the light source. State both:
- "Window light from the right. All shadows fall to the left."
- "Overhead fluorescent. Shadows directly beneath objects, short and dark."
10. Master Painter Reference Patterns
Caravaggio — Tenebrism
Extreme dark-light. Figures emerge from blackness. Use for dramatic, emotional scenes.
Prompt: "Tenebristic lighting — figure dramatically lit from one side, emerging from near-black background. Deep shadows, powerful highlights, minimal midtones."
Vermeer — Domestic Light
Soft window light, intimate interiors, quiet depth. Use for calm, contemplative scenes.
Prompt: "Vermeer-like soft daylight from left window. Gentle falloff across the room. Warm on lit surfaces, cool in shadows. Quiet domestic atmosphere."
Monet — Atmospheric Color
Depth through color temperature shifts and dissolved edges. Use for outdoor/environmental scenes.
Prompt: "Impressionistic depth — foreground colors warm and saturated, middle distance cooler and lighter, far background dissolving into pale blue-violet haze."
Miyazaki/Ghibli — Layered Environments
Multiple parallax layers with distinct treatments. Use for environments with depth and wonder.
Prompt: "Layered depth like a Ghibli background — detailed foreground plants, mid-distance architecture with softer edges, distant mountains as pale watercolor shapes, sky with subtle cloud layers."
Spider-Verse — Stylized Depth Cues
Different artistic treatments at different depths. Bold and deliberate.
Prompt: "Foreground elements rendered with heavy ink lines and halftone dots. Mid-distance figures with clean cel-shading. Background as simplified color shapes with visible brush strokes."
11. Combining Techniques — The Depth Stack
For maximum depth, layer multiple techniques. Never rely on just one.
The 5-Layer Depth Stack (use for complex scenes)
1. ATMOSPHERIC: far objects lighter, bluer, softer
2. OVERLAP: every layer occludes the one behind
3. SIZE: near objects large, far objects small
4. SHADOW: cast shadows anchor objects to surfaces
5. FOCUS: subject sharp, fore/background softer
Minimum Viable Depth (use for simple panels)
At minimum, every prompt should include:
- ONE overlap statement ("X partially blocks Y")
- ONE size cue ("X fills the frame / Y is small at the far wall")
- ONE atmospheric cue ("background slightly softer and cooler")
Common Mistakes
- No overlap anywhere — everything side by side, same plane. Always have something partially block something else.
- Uniform sharpness — everything equally crisp from edge to edge. Add at least subtle softening to the background.
- Same color temperature everywhere — no warm/cool separation. Make foreground warmer, background cooler.
- Forgetting cast shadows — objects float without ground contact. Add shadows to anchor them.
- Describing depth abstractly ("there's a sense of depth") — describe the SPECIFIC techniques ("the linoleum tiles shrink toward the far wall, colors desaturating with distance").
- Using shallow DOF in POV panels — human eyes don't produce bokeh. Use deep focus for POV.
- Same level of detail everywhere — near and far objects described with equal specificity. Foreground gets granular description, background gets broad strokes.