Wan 2.5 Image to Video
Wan2.5-Preview creates stunning, high-resolution videos with flawless audio synchronization from multiple inputs.
API
If you're looking for an API, you can choose from your desired programming language.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
import requests
api_key = "YOUR_API_KEY"
url = "https://api.segmind.com/v1/wan-2.5-i2v"
# Prepare data and files
data = {}
files = {}
data['seed'] = 42
# For parameter "audio", you can send a raw file or a URI:
# files['audio'] = open('IMAGE_PATH', 'rb') # To send a file
data['audio'] = 'null' # To send a URI
# For parameter "image", you can send a raw file or a URI:
# files['image'] = open('IMAGE_PATH', 'rb') # To send a file
data['image'] = 'https://segmind-resources.s3.amazonaws.com/output/21aeb463-bb17-4536-864b-0bd1e11594a9-EMXN1y8qTgoGdXBsb2FkEg55bGFiLXN0dW50LXNncBo0YWlfcG9ydGFsLzE3NTM5NjM5NTIvbjdwNDlYOURCbS8yZTk0X2wwXzAwMS0wXzAuanBlZw_1000x1000.webp' # To send a URI
data['prompt'] = "Kitten in a McDonald's uniform stands on a stool, grilling burger patties. It flips the patties with a spatula, watches them sizzle, and occasionally looks around while steam rises from the grill."
data['duration'] = "5"
data['resolution'] = "720p"
data['negative_prompt'] = "unnecessary clutter, dark shadows"
data['enable_prompt_expansion'] = True
headers = {'x-api-key': api_key}
# If no files, send as JSON
if files:
response = requests.post(url, data=data, files=files, headers=headers)
else:
response = requests.post(url, json=data, headers=headers)
print(response.content) # The response is the generated image
Attributes
Sets a random seed for consistent outputs. Use values between 1 and 100 for variation.
Upload an audio file for syncing. Use a song clip or melody for dynamic results.
Input image for video generation. Choose a high-resolution image for best quality.
Text description for video creation. Include vivid visuals for creative animations.
Sets the video length. Choose 5 seconds for shorter clips and 10 for longer scenes.
Allowed values:
Set video quality. Use 1080p for high-quality renders and 480p for faster results.
Allowed values:
Avoid certain elements in generation. Include unwanted objects or colors for exclusion.
Activates prompt optimizer for enhanced results. Set to true for more detailed outputs.
To keep track of your credit usage, you can inspect the response headers of each API call. The x-remaining-credits property will indicate the number of remaining credits in your account. Ensure you monitor this value to avoid any disruptions in your API usage.
Resources to get you started
Everything you need to know to get the most out of Wan 2.5 Image to Video
# Wan2.5-Preview: Guide to Effective Use
Wan2.5-Preview is a state-of-the-art multimodal AI model that generates up to 10-second 1080p videos with synchronized audio, image input, and rich textual prompts. Follow this guide to optimize parameters and achieve cinematic-quality outputs across diverse applications.
## 1. Getting Started
1. Provide a high-resolution `image` URL (recommended 1080Ă1080 px or above).
2. Write a concise, vivid `prompt` describing scene composition, motion, and style.
3. (Optional) Upload an `audio` file (MP3/ WAV) to drive soundtrack or narration.
4. Adjust advanced parameters for duration, resolution, seed, and negative prompts.
## 2. Recommended Parameter Presets
Use Case | Resolution | Duration | Seed | Prompt Expansion | Negative Prompt
----------------------- | ---------- | -------- | ---- | ---------------- | ----------------
Professional Ads | 1080p | 10s | 12 | true | âgrain, low-contrastâ
Social Media Teasers | 720p | 5s | 42 | true | âblurry edges, dark shadowsâ
Educational Narration | 720p | 10s | 7 | true | âbackground noise, clutterâ
Music Video Snippets | 1080p | 10s | 29 | false | âstatic cameras, dull colorsâ
Quick Prototyping | 480p | 5s | 99 | false | âextra objects, lens flareâ
- **Resolution**:
⢠1080p for broadcast-grade quality
⢠720p for balanced speed and detail
⢠480p for fast previews
- **Duration**: 5 s for short loops; 10 s for extended scenes.
- **Seed (1â100)**: Ensures reproducibility. Reuse same seed + inputs for consistent results.
- **Enable Prompt Expansion**: Improves detail but may increase processing time. Set to `true` for complex scenes.
## 3. Prompt Engineering Tips
- Start with the subject: âA sleek drone sweeping over neon-lit cityscapesâŚâ
- Specify camera movements: âslow dolly, zoom out, pan left to right.â
- Include mood, lighting, color palette: âwarm sunset glow, soft shadows.â
- Use `negative_prompt` to exclude unwanted elements: e.g., âno rain, no subtitles.â
- Combine keywords and short phrasesâavoid run-on sentences.
## 4. Audio Synchronization
- Upload voice-over or music via the `audio` field.
- Ensure audio length matches video duration to prevent cuts.
- For multi-track sound design, overlay effects (e.g., âwind whoosh,â âcrowd murmurâ) in a single file.
## 5. Best Practices
- Test with lower resolution (480p) when crafting prompts.
- Lock your seed before final render.
- Review generated thumbnails before full download.
- Tweak negative prompts iteratively to refine outputs.
By following these presets and tips, youâll harness Wan2.5-Previewâs full potential for polished, synchronized audiovisual content every time.
Other Popular Models
Discover other models you might be interested in.
faceswap-v2
Take a picture/gif and replace the face in it with a face of your choice. You only need one image of the desired face. No dataset, no training

sdxl-inpaint
This model is capable of generating photo-realistic images given any text input, with the extra capability of inpainting the pictures by using a mask

sdxl1.0-txt2img
The SDXL model is the official upgrade to the v1.5 model. The model is released as open-source software

sd2.1-faceswapper
Take a picture/gif and replace the face in it with a face of your choice. You only need one image of the desired face. No dataset, no training
