Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Specify lighting, atmosphere, and light transitions for Seedance 2.0 prompts using named light sources, core parameters, and atmosphere contracts. Use when the scene needs a specific mood, time of day, or lighting style, or when lighting is flat, inconsistent across shots, or clipping.
.claude/skills/bilal140202-seedance-lighting/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -14% | 0% |
Lighting as narrative state. Measurable atmosphere contracts for Seedance 2.0.
❌ dramatic lighting
✅ single overhead practical as hard key, 5600K, deep shadow fillwindow backlight casting long shadows toward camera
neon sign as key light, pink and blue, no fill
firelight flicker, warm amber, unstable
overcast diffused daylight, soft wrap shadows
single bare bulb swinging overhead
red emergency lamp as sole light source
candle practical, warm gold, deep shadow beyond 1m
streetlamp sodium key, cool fill from ground reflectionKey direction: camera-left / camera-right / above / below / behind (rim)
Contrast: low-key (deep shadows) / high-key (bright, minimal shadows)
Color temp: warm amber / cool blue / neutral white (Kelvin optional)
Shadow edge: hard-edged / soft wrap / no shadowsEvery atmospheric effect must be describable in physical terms.
thin fog catching the backlight, visibility 5m
dust motes suspended in a sunbeam from camera-left
light rain visible against the key light, not heavy
breath misting in cold air at each exhale
heat shimmer rising from asphalt, subtle
smoke drifting from off-screen right, backlitIf fog becomes a blur: reduce density → increase backlight specificity → simplify background.
| Chinese | English | Usage | |---------|---------|-------| | 主光 | Key light | Primary dominant source | | 辅光 | Fill light | Softens key shadows | | 轮廓光 | Rim light | Edge separation from background | | 逆光 | Backlight / Contre-jour | Subject silhouetted or rimmed | | 低调布光 | Low-key | Dark, high-contrast; drama | | 高调布光 | High-key | Bright, low-contrast; commercial | | 实景灯光 | Practical light | Visible source in frame | | 丁达尔效应 | Tyndall rays / God rays | Light beams through particles |
Use timestamps for in-clip light changes:
0–3s: warm golden overhead light.
3–5s: lights flicker and die.
5–end: only red emergency lamp remains; cool rain spill from open door as rim.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 10,741 | 5,458 | -49% | 1 | 1 | 0% | 1,453 | 1,609 | +11% | 0 | 0 | — |
case-03 | fail→pass | 13,687 | 6,639 | -51% | 1 | 1 | 0% | 2,222 | 1,872 | -16% | 0 | 0 | — |
case-04 | fail→pass | 11,528 | 5,846 | -49% | 1 | 1 | 0% | 1,793 | 1,740 | -3% | 0 | 0 | — |
case-01 | fail→fail | 17,966 | 5,716 | -68% | 1 | 1 | 0% | 2,523 | 1,680 | -33% | 0 | 0 | — |
case-05 | pass→pass | 8,493 | 5,140 | -39% | 1 | 1 | 0% | 1,419 | 1,642 | +16% | 0 | 0 | — |
case-06 | pass→pass | 10,321 | 5,936 | -42% | 1 | 1 | 0% | 1,758 | 1,843 | +5% | 0 | 0 | — |
case-07 | pass→pass | 9,224 | 6,155 | -33% | 1 | 1 | 0% | 1,391 | 1,783 | +28% | 0 | 0 | — |
case-08 | pass→pass | 7,824 | 6,986 | -11% | 1 | 1 | 0% | 1,391 | 1,992 | +43% | 0 | 0 | — |
case-09 | pass→pass | 10,128 | 5,296 | -48% | 1 | 1 | 0% | 1,697 | 1,698 | +0% | 0 | 0 | — |
case-20 | fail→fail | 15,067 | 9,567 | -37% | 1 | 1 | 0% | 2,461 | 2,277 | -7% | 0 | 0 | — |
case-10 | fail→pass | 7,529 | 3,622 | -52% | 1 | 1 | 0% | 1,196 | 1,414 | +18% | 0 | 0 | — |
case-11 | pass→pass | 9,625 | 2,449 | -75% | 1 | 1 | 0% | 1,656 | 1,216 | -27% | 0 | 0 | — |
case-12 | pass→pass | 6,626 | 3,194 | -52% | 1 | 1 | 0% | 1,087 | 1,210 | +11% | 0 | 0 | — |
case-13 | fail→pass | 8,826 | 5,824 | -34% | 1 | 1 | 0% | 1,270 | 1,712 | +35% | 0 | 0 | — |
case-14 | fail→pass | 13,821 | 6,426 | -54% | 1 | 1 | 0% | 2,053 | 1,759 | -14% | 0 | 0 | — |
case-15 | pass→pass | 9,410 | 6,828 | -27% | 1 | 1 | 0% | 1,268 | 1,797 | +42% | 0 | 0 | — |
case-16 | pass→pass | 11,298 | 5,055 | -55% | 1 | 1 | 0% | 1,719 | 1,606 | -7% | 0 | 0 | — |
case-17 | fail→pass | 13,163 | 5,396 | -59% | 1 | 1 | 0% | 1,886 | 1,507 | -20% | 0 | 0 | — |
case-18 | pass→pass | 10,711 | 4,292 | -60% | 1 | 1 | 0% | 1,500 | 1,500 | 0% | 0 | 0 | — |
case-19 | pass→pass | 12,340 | 5,814 | -53% | 1 | 1 | 0% | 1,870 | 1,686 | -10% | 0 | 0 | — |
case-21 | fail→pass | 8,824 | 4,542 | -49% | 1 | 1 | 0% | 1,536 | 1,481 | -4% | 0 | 0 | — |
case-22 | fail→fail | 10,690 | 10,719 | +0% | 1 | 1 | 0% | 1,709 | 2,561 | +50% | 0 | 0 | — |
case-23 | pass→pass | 10,734 | 5,395 | -50% | 1 | 1 | 0% | 1,422 | 1,606 | +13% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +30 percentage points is the difference between those two pass rates over the 23 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.