showmeai（showmeai 视频）

Showmeai Image, Video & 3D Generation

Generate images via Showmeai's OpenAI-compatible Images API (/images/generations), videos via Seedance API (/task/volces/seedance), or convert 2D images to 3D models (/task/gi/image-to-3d).

Basic Usage

python3 {baseDir}/scripts/gen.py --prompt "your prompt here"

Options

# Specify model (default: nano-banana-pro)
python3 {baseDir}/scripts/gen.py --prompt "..." --model nano-banana-pro

# Higher resolution (append -2k or -4k to model name)
python3 {baseDir}/scripts/gen.py --prompt "..." --model nano-banana-pro-2k

# Save image locally (default: NO save, URL only)
python3 {baseDir}/scripts/gen.py --prompt "..." --save

# Save to OSS directory (~/.openclaw/oss/)
python3 {baseDir}/scripts/gen.py --prompt "..." --oss

# Save to custom directory
python3 {baseDir}/scripts/gen.py --prompt "..." --save --out-dir /path/to/dir

# Aspect ratio
python3 {baseDir}/scripts/gen.py --prompt "..." --aspect-ratio 16:9

# Image count
python3 {baseDir}/scripts/gen.py --prompt "..." --count 2

# Edit image (provide --input)
python3 {baseDir}/scripts/gen.py --prompt "make it look like a painting" --input /path/to/image.jpg

Video Generation

# Basic text-to-video
python3 {baseDir}/scripts/video_gen.py --prompt "A detective enters a dim room, examines clues on the desk"

# With reference image (image-to-video)
python3 {baseDir}/scripts/video_gen.py --prompt "Girl holding a fox, opens eyes and looks at camera gently" --image /path/to/image.jpg

# First-and-last-frame video (requires both frames)
python3 {baseDir}/scripts/video_gen.py --prompt "A blue-green bird transforms into human form" --first-frame /path/to/first.jpg --last-frame /path/to/last.jpg

# Without audio (cheaper)
python3 {baseDir}/scripts/video_gen.py --prompt "..." --no-audio

# Draft/preview mode (faster, lower quality)
python3 {baseDir}/scripts/video_gen.py --prompt "..." --draft

# Save to local directory
python3 {baseDir}/scripts/video_gen.py --prompt "..." --save --out-dir /path/to/dir

# With custom resolution, ratio, and duration
python3 {baseDir}/scripts/video_gen.py --prompt "..." --resolution 1080p --ratio 16:9 --duration 10

# With watermark and fixed camera
python3 {baseDir}/scripts/video_gen.py --prompt "..." --watermark --camera-fixed

# With seed for reproducible results
python3 {baseDir}/scripts/video_gen.py --prompt "..." --seed 12345

Image-to-3D Conversion

# Basic conversion
python3 {baseDir}/scripts/image_to_3d.py --image /path/to/image.png

# With custom model and format
python3 {baseDir}/scripts/image_to_3d.py --image character.png --model Hunyuan3D-2 --format glb

# With texture and higher quality
python3 {baseDir}/scripts/image_to_3d.py --image character.png --texture --steps 10 --resolution 256

# Query task status
python3 {baseDir}/scripts/image_to_3d.py --query <task_id>

# Download when complete
python3 {baseDir}/scripts/image_to_3d.py --query <task_id> --save

Supported Models

nano-banana series (returns URL, fast):

• nano-banana
• nano-banana-pro ← default
• nano-banana-2
• nano-banana-pro-2k / nano-banana-pro-4k (high res)

gpt-image series (returns base64, always saved):

• gpt-image-1
• gpt-image-1.5

Video models (Seedance API):

• doubao-seedance-1-5-pro-251215 ← default (supports audio, draft mode, text-to-video, image-to-video, first-and-last-frame)

3D models (Image-to-3D API):

• Hunyuan3D-2 ← default (supports glb/stl output via type parameter)
• Hi3DGen (supports glb/stl output via file_format parameter)
• Step1X-3D (supports glb/stl output via file_format parameter)

Config

Set in .env or ~/.openclaw/openclaw.json:

• Showmeai_API_KEY — your Showmeai API key (required)
• Showmeai_BASE_URL — base URL with /v1 suffix (required); defaults to https://api.showmeai.art/v1 if not set

Save Behavior

Images:

• Default: no local file, output MEDIA:<url> directly
• --save: save to ~/.openclaw/media/
• --oss: save to ~/.openclaw/oss/
• gpt-image models always save to media/ (API returns base64 only)

Videos:

• Videos are saved when the API returns direct URL or base64 data
• For async task submission, the task ID is output
• Use --save to ensure local saving when video is available

3D Models:

• Conversion is async, returns a task ID
• Use --query <task_id> to check status
• Use --query <task_id> --save to download when complete
• Default save location: ~/.openclaw/media/

Video Prompt Parameters

New way (recommended): Use direct command-line parameters

• --resolution 480p/720p/1080p — video resolution
• --ratio 16:9/4:3/1:1/3:4/9:16/21:9/adaptive — aspect ratio
• --duration 2-12 — duration in seconds (use 0 for auto on 1.5 pro)
• --frames <n> — number of frames (alternative to duration)
• --watermark — add watermark
• --camera-fixed — keep camera fixed
• --seed <n> — random seed for reproducibility

Old way (still supported): Append to prompt text

• --ratio 16:9 / --ratio adaptive
• --rs 720p / --rs 480p / --rs 1080p
• --dur 5 / --dur 10
• --cf false — disable fixed camera

Video specs: 24 FPS, durations 2-12s, resolutions 480P/720P/1080P.