aliyun-wan-video

Name: aliyun-wan-video
Author: cinience

397stars34forks

How to install aliyun-wan-video?

cinience/alicloud-skills/aliyun-wan-video

$npx -y skills add cinience/alicloud-skills --skill aliyun-wan-video

Installs into the current project.

Summary

Use when generating videos with Model Studio DashScope SDK using Wan video generation models (wan2.6-t2v, wan2.6-i2v-flash, wan2.6-i2v and regional variants). Use when implementing or documenting video.generate requests/responses, mapping prompt/negative_prompt/duration/fps/size/

Files · 1

View on GitHub

SKILL.md

1	Category: provider
2
3	# Model Studio Wan Video
4
5	## Validation
6
7	```bash
8	mkdir -p output/aliyun-wan-video
9	python -m py_compile skills/ai/video/aliyun-wan-video/scripts/generate_video.py && echo "py_compile_ok" > output/aliyun-wan-video/validate.txt
10	```
11
12	Pass criteria: command exits 0 and `output/aliyun-wan-video/validate.txt` is generated.
13
14	## Output And Evidence
15
16	- Save task IDs, polling responses, and final video URLs to `output/aliyun-wan-video/`.
17	- Keep one end-to-end run log for troubleshooting.
18
19	Provide consistent video generation behavior for the video-agent pipeline by standardizing `video.generate` inputs/outputs and using DashScope SDK (Python) with the exact model name.
20
21	## Critical model names
22
23	Use one of these exact model strings:
24	- `wan2.6-t2v`
25	- `wan2.6-t2v-us`
26	- `wan2.2-t2v-plus`
27	- `wan2.2-t2v-flash`
28	- `wan2.6-i2v-flash`
29	- `wan2.6-i2v`
30	- `wan2.6-i2v-us`
31	- `wanx2.1-t2v-turbo`
32
33	## Prerequisites
34
35	- Install SDK (recommended in a venv to avoid PEP 668 limits):
36
37	```bash
38	python3 -m venv .venv
39	. .venv/bin/activate
40	python -m pip install dashscope
41	```
42	- Set `DASHSCOPE_API_KEY` in your environment, or add `dashscope_api_key` to `~/.alibabacloud/credentials` (env takes precedence).
43
44	## Normalized interface (video.generate)
45
46	### Request
47	- `prompt` (string, required)
48	- `negative_prompt` (string, optional)
49	- `duration` (number, required) seconds
50	- `fps` (number, required)
51	- `size` (string, required) e.g. `1280*720`
52	- `seed` (int, optional)
53	- `reference_image` (string \| bytes, optional for t2v, required for i2v family models)
54	- `motion_strength` (number, optional)
55
56	### Response
57	- `video_url` (string)
58	- `duration` (number)
59	- `fps` (number)
60	- `seed` (int)
61
62	## Quick start (Python + DashScope SDK)
63
64	Video generation is usually asynchronous. Expect a task ID and poll until completion.
65	Note: Wan i2v models require an input image; pure t2v models such as `wan2.6-t2v` can omit `reference_image`.
66
67	```python
68	import os
69	from dashscope import VideoSynthesis
70
71	# Prefer env var for auth: export DASHSCOPE_API_KEY=...
72	# Or use ~/.alibabacloud/credentials with dashscope_api_key under [default].
73
74	def generate_video(req: dict) -> dict:
75	payload = {
76	"model": req.get("model", "wan2.6-i2v-flash"),
77	"prompt": req["prompt"],
78	"negative_prompt": req.get("negative_prompt"),
79	"duration": req.get("duration", 4),
80	"fps": req.get("fps", 24),
81	"size": req.get("size", "1280*720"),
82	"seed": req.get("seed"),
83	"motion_strength": req.get("motion_strength"),
84	"api_key": os.getenv("DASHSCOPE_API_KEY"),
85	}
86
87	if req.get("reference_image"):
88	# DashScope expects img_url for i2v models; local files are auto-uploaded.
89	payload["img_url"] = req["reference_image"]
90
91	response = VideoSynthesis.call(**payload)
92
93	# Some SDK versions require polling for the final result.
94	# If a task_id is returned, poll until status is SUCCEEDED.
95	result = response.output.get("results", [None])[0]
96
97	return {
98	"video_url": None if not result else result.get("url"),
99	"duration": response.output.get("duration"),
100	"fps": response.output.get("fps"),
101	"seed": response.output.get("seed"),
102	}
103	```
104
105	## Async handling (polling)
106
107	```python
108	import os
109	from dashscope import VideoSynthesis
110
111	task = VideoSynthesis.async_call(
112	model=req.get("model", "wan2.6-i2v-flash"),
113	prompt=req["prompt"],
114	img_url=req["reference_image"],
115	duration=req.get("duration", 4),
116	fps=req.get("fps", 24),
117	size=req.get("size", "1280*720"),
118	api_key=os.getenv("DASHSCOPE_API_KEY"),
119	)
120
121	final = VideoSynthesis.wait(task)
122	video_url = final.output.get("video_url")
123	```
124
125	## Operational guidance
126
127	- Video generation can take minutes; expose progress and allow cancel/retry.
128	- Cache by `(prompt, negative_prompt, duration, fps, size, seed, reference_image hash, motion_strength)`.
129	- Store video assets in object storage and persist only URLs in metadata.
130	- `reference_image` can be a URL or local path; the SDK auto-uploads local files.
131	- If you get `Field required: input.img_url`, the reference image is missing or not mapped.
132	- `wan2.6-t2v` and `wan2.6-t2v-us` add multi-shot narrative support and optional audio input according to the official docs.
133
134	## Size notes
135
136	- Use `WxH` format (e.g. `1280*720`).
137	- Prefer common sizes; unsupported sizes can return 400.
138
139	## Output location
140
141	- Default output: `output/aliyun-wan-video/videos/`
142	- Override base dir with `OUTPUT_DIR`.
143
144	## Anti-patterns
145
146	- Do not invent model names or aliases; use official Wan i2v model I

People who liked this also viewed

6 picks

Type

aliyun-wan-video

397stars34forks

View on GitHub

How to install aliyun-wan-video?

cinience/alicloud-skills/aliyun-wan-video

$npx -y skills add cinience/alicloud-skills --skill aliyun-wan-video

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	Category: provider
2
3	# Model Studio Wan Video
4
5	## Validation
6
7	```bash
8	mkdir -p output/aliyun-wan-video
9	python -m py_compile skills/ai/video/aliyun-wan-video/scripts/generate_video.py && echo "py_compile_ok" > output/aliyun-wan-video/validate.txt
10	```
11
12	Pass criteria: command exits 0 and `output/aliyun-wan-video/validate.txt` is generated.
13
14	## Output And Evidence
15
16	- Save task IDs, polling responses, and final video URLs to `output/aliyun-wan-video/`.
17	- Keep one end-to-end run log for troubleshooting.
18
19	Provide consistent video generation behavior for the video-agent pipeline by standardizing `video.generate` inputs/outputs and using DashScope SDK (Python) with the exact model name.
20
21	## Critical model names
22
23	Use one of these exact model strings:
24	- `wan2.6-t2v`
25	- `wan2.6-t2v-us`
26	- `wan2.2-t2v-plus`
27	- `wan2.2-t2v-flash`
28	- `wan2.6-i2v-flash`
29	- `wan2.6-i2v`
30	- `wan2.6-i2v-us`
31	- `wanx2.1-t2v-turbo`
32
33	## Prerequisites
34
35	- Install SDK (recommended in a venv to avoid PEP 668 limits):
36
37	```bash
38	python3 -m venv .venv
39	. .venv/bin/activate
40	python -m pip install dashscope
41	```
42	- Set `DASHSCOPE_API_KEY` in your environment, or add `dashscope_api_key` to `~/.alibabacloud/credentials` (env takes precedence).
43
44	## Normalized interface (video.generate)
45
46	### Request
47	- `prompt` (string, required)
48	- `negative_prompt` (string, optional)
49	- `duration` (number, required) seconds
50	- `fps` (number, required)
51	- `size` (string, required) e.g. `1280*720`
52	- `seed` (int, optional)
53	- `reference_image` (string \| bytes, optional for t2v, required for i2v family models)
54	- `motion_strength` (number, optional)
55
56	### Response
57	- `video_url` (string)
58	- `duration` (number)
59	- `fps` (number)
60	- `seed` (int)
61
62	## Quick start (Python + DashScope SDK)
63
64	Video generation is usually asynchronous. Expect a task ID and poll until completion.
65	Note: Wan i2v models require an input image; pure t2v models such as `wan2.6-t2v` can omit `reference_image`.
66
67	```python
68	import os
69	from dashscope import VideoSynthesis
70
71	# Prefer env var for auth: export DASHSCOPE_API_KEY=...
72	# Or use ~/.alibabacloud/credentials with dashscope_api_key under [default].
73
74	def generate_video(req: dict) -> dict:
75	payload = {
76	"model": req.get("model", "wan2.6-i2v-flash"),
77	"prompt": req["prompt"],
78	"negative_prompt": req.get("negative_prompt"),
79	"duration": req.get("duration", 4),
80	"fps": req.get("fps", 24),
81	"size": req.get("size", "1280*720"),
82	"seed": req.get("seed"),
83	"motion_strength": req.get("motion_strength"),
84	"api_key": os.getenv("DASHSCOPE_API_KEY"),
85	}
86
87	if req.get("reference_image"):
88	# DashScope expects img_url for i2v models; local files are auto-uploaded.
89	payload["img_url"] = req["reference_image"]
90
91	response = VideoSynthesis.call(**payload)
92
93	# Some SDK versions require polling for the final result.
94	# If a task_id is returned, poll until status is SUCCEEDED.
95	result = response.output.get("results", [None])[0]
96
97	return {
98	"video_url": None if not result else result.get("url"),
99	"duration": response.output.get("duration"),
100	"fps": response.output.get("fps"),
101	"seed": response.output.get("seed"),
102	}
103	```
104
105	## Async handling (polling)
106
107	```python
108	import os
109	from dashscope import VideoSynthesis
110
111	task = VideoSynthesis.async_call(
112	model=req.get("model", "wan2.6-i2v-flash"),
113	prompt=req["prompt"],
114	img_url=req["reference_image"],
115	duration=req.get("duration", 4),
116	fps=req.get("fps", 24),
117	size=req.get("size", "1280*720"),
118	api_key=os.getenv("DASHSCOPE_API_KEY"),
119	)
120
121	final = VideoSynthesis.wait(task)
122	video_url = final.output.get("video_url")
123	```
124
125	## Operational guidance
126
127	- Video generation can take minutes; expose progress and allow cancel/retry.
128	- Cache by `(prompt, negative_prompt, duration, fps, size, seed, reference_image hash, motion_strength)`.
129	- Store video assets in object storage and persist only URLs in metadata.
130	- `reference_image` can be a URL or local path; the SDK auto-uploads local files.
131	- If you get `Field required: input.img_url`, the reference image is missing or not mapped.
132	- `wan2.6-t2v` and `wan2.6-t2v-us` add multi-shot narrative support and optional audio input according to the official docs.
133
134	## Size notes
135
136	- Use `WxH` format (e.g. `1280*720`).
137	- Prefer common sizes; unsupported sizes can return 400.
138
139	## Output location
140
141	- Default output: `output/aliyun-wan-video/videos/`
142	- Override base dir with `OUTPUT_DIR`.
143
144	## Anti-patterns
145
146	- Do not invent model names or aliases; use official Wan i2v model I

People who liked this also viewed

6 picks

Type