Model Studio Wan R2V

---

name: alicloud-ai-video-wan-r2v

description: Generate reference-based videos with Alibaba Cloud Model Studio Wan R2V models (wan2.6-r2v-flash, wan2.6-r2v). Use when creating multi-shot videos from reference video/image material, preserving character style, or documenting reference-to-video request/response flows.

version: 1.0.0

---

Category: provider

# Model Studio Wan R2V

Validation

mkdir -p output/alicloud-ai-video-wan-r2v
python -m py_compile skills/ai/video/alicloud-ai-video-wan-r2v/scripts/prepare_r2v_request.py && echo "py_compile_ok" > output/alicloud-ai-video-wan-r2v/validate.txt

Pass criteria: command exits 0 and `output/alicloud-ai-video-wan-r2v/validate.txt` is generated.

Output And Evidence

Use Wan R2V for reference-to-video generation. This is different from i2v (single image to video).

Critical model names

Use one of these exact model strings:

Prerequisites

python3 -m venv .venv
. .venv/bin/activate
python -m pip install dashscope

Normalized interface (video.generate_reference)

Request

Response

Async handling

Local helper script

Prepare a normalized request JSON and validate response schema:

.venv/bin/python skills/ai/video/alicloud-ai-video-wan-r2v/scripts/prepare_r2v_request.py \
  --prompt "Generate a short montage with consistent character style" \
  --reference-video "https://example.com/reference.mp4"

Output location

Workflow

1) Confirm user intent, region, identifiers, and whether the operation is read-only or mutating.

2) Run one minimal read-only query first to verify connectivity and permissions.

3) Execute the target operation with explicit parameters and bounded scope.

4) Verify results and save output/evidence files.

References