# Shallow-water ripples: keep the rock and shoreline fixed

An original shallow pool input for small surface wavelets, with rock-outline, shoreline and water-level checks.

These original SVG/PNG teaching inputs were authored for a method exercise. No video model was run and no generated result is included. PNG files are raster counterparts of the editable SVG sources.

## Input responsibilities

Upload only the clean first frame; diagram is for review.

Use the current controls at /generator/image-to-video; check actual accepted inputs and settings before submitting.

## Goal

Animate gentle ripples on the shallow water while the large rock and the curved bank remain in place. Separate changing highlights from moving solid geometry; the shot is a fictional visual exercise, not a hydrology record.

## Preparation

- Use the supplied shallow-pool illustration with one large angular rock and a clear curved bank. A replacement image needs a visible waterline against a solid object.
- Choose small wavelets rather than a surge, tide or rising water. Keep the same water level so the exposed rock area can be compared.
- Keep the camera fixed and omit falling objects, animals and splashes. They would introduce causes and interactions outside this surface-motion task.
- Use the large rock’s corner points and the bank curve as fixed references. The diagram is for observation; upload only the clean scene PNG.

## Copyable brief

```text
Use the uploaded shallow-water scene as the first frame. Small gentle ripples travel across the water surface and around the fixed large rock. Let only the surface highlights and soft wavelet lines change. Keep the rock’s angular outline, position and exposed area unchanged; keep the curved bank and water level stationary. The camera stays fixed and the daylight remains steady. Do not add a splash, a falling object, animals or a strong current. End with the same shallow pool and quiet ripples continuing.
```

## Procedure

1. **Record the solid boundaries** — Trace the main rock corners and the bank curve in the observation sheet. Do this before assessing the ripple pattern so moving highlights are not mistaken for moving stones.

2. **Separate texture motion from level change** — Request wavelets on the existing surface while keeping the exposed rock area stable. A waterline climbing the rock would be a different action with different checks.

3. **Inspect rock edges during a passing ripple** — Watch a ripple meet the rock’s visible boundary. The wavelet may break into highlights, but the solid corner must not soften, flow downstream or produce a duplicate rock.

4. **Check the quiet continuing ending** — Compare the final bank and waterline to the source, then replay for sudden surges. Continued ripples do not automatically make the clip a seamless loop.

## Observation record

Use checks.csv to record pass/fail/not-observable and the earliest affected time. Do not pre-fill a pass: there is no generated clip in this package.

- The large rock remains at the same location and scale.
- Its angular corners do not melt into the moving water.
- The curved bank keeps the same outline and position.
- The exposed rock area and waterline remain stable.
- Visible movement is confined to small water-surface ripples and highlights.
- No splash source, animal, new stone or strong current appears.
- The ending keeps quiet surface motion without a surge or camera drift.

## Revise one variable

**The rock drifts with the water**: Change only the fixed-solid sentence and retain the same ripple request. Compare the rock corners rather than the changing reflection.

**The water rises up the bank**: Replace only surge-like wording with small surface wavelets at the existing level. Recheck the rock waterline across the full shot.

**Ripples become a foamy torrent**: Reduce only the motion amplitude and remove current-speed language. Keep the same pool and inspect whether solid boundaries remain visible.

## Ending

End with ongoing quiet ripples and unchanged solid boundaries. Save separate surface and waterline observations; this scene does not prove conditions at any real location.

## Evidence boundary

Historical parent queries: “minimax h3 image to video”, “minimax h3 prompting image to video”. Exact long-tail demand is unmeasured. Source: docs/research/pseo-keyword-gap-2026-10-05/evidence/bing-live-keywords.csv. Existing broad tutorial: /blog/landscape-video-prompts. The owner covers multiple landscape layers and continuous stream flow; this task supplies a fixed shallow pool with only small surface ripples and a rock-waterline comparison sheet.
