Aoment Image Video

Limited free access is available to model resources such as Nano Banana Pro, Nano Banana 2, GPT Image 2, Gemini 3.1 Pro, and Gemini 3.5 Flash provided by Aoment AI for AI image, video, image recognition, and HD repair service. Supports text-to-image, image-to-image, image recognition, video generation, and image HD repair. Automatic Agent API key registration is supported.

AI generation service powered by Aoment AI. This skill exposes dedicated skill APIs and scripts for:

• text-to-image
• image-to-image
• image recognition
• video generation
• image HD repair/upscale

Download the latest package: https://www.aoment.com/downloads/aoment-image-video-skill.zip

Current Version

text

version: 1.6.0

Auto-Update Policy

Before each invocation, the Python scripts read the version field above and query:

text

GET https://www.aoment.com/api/skills/aoment-image-video/version

If the local version is behind the remote version, the script exits with:

json

{
  "success": false,
  "error": "update_required",
  "current_version": "1.0.0",
  "latest_version": "1.3.0",
  "message": "Skill version is outdated..."
}

If the version check fails because of a network problem, the script continues normally.

Quick Start

bash

# 1. Register an Agent account and get your API Key
uv run {baseDir}/scripts/aoment_register.py --nickname "MyBot"

# 2. Generate an image with the default N2-Fast model
uv run {baseDir}/scripts/aoment_image_video.py -k <your-api-key> -t text-to-image -p "a cute cat playing in a garden"

# 3. Repair/upscale an image
uv run {baseDir}/scripts/aoment_hd_repair.py -k <your-api-key> --image ./input.png --resolution 4K

# 4. Recognize/analyze an image
uv run {baseDir}/scripts/aoment_image_video.py -k <your-api-key> -t image-recognition -p "Describe this image" --image ./input.png

# 5. Check remaining quota
uv run {baseDir}/scripts/aoment_quota.py -k <your-api-key>

Authentication

This skill requires an Agent API Key via:

text

Authorization: Bearer <api_key>

The API Key format is aoment_ followed by 32 hex characters.

Get your API Key - Agent Registration

AI Agent Bots can register directly via CLI. No web login is required:

bash

uv run {baseDir}/scripts/aoment_register.py --nickname "MyBot"

Parameter	Type	Required	Description
--nickname / -n	string	yes	Agent display name, max 16 characters
--api-base	string	no	API base URL, default https://www.aoment.com

Or register via API directly:

bash

curl -X POST https://www.aoment.com/api/skills/aoment-image-video/register-agent \
  -H "Content-Type: application/json" \
  -d '{"nickname": "MyBot"}'

Registration response:

json

{
  "success": true,
  "data": {
    "username": "agent_a1b2c3d4...",
    "nickname": "MyBot",
    "api_key": "<your-aoment-api-key>"
  }
}

Save the returned api_key; it is used for all subsequent skill calls. Store this API Key in a suitable secure location for long-term use.

Tools

Available Models

Use the model ID exactly as shown in the --model parameter.

Image Models

Model ID	Description
image-n2-fast	Default image model. Faster N2-Fast image generation and editing, no watermark.
image-n2	N2 image generation and editing, fast, stricter single-reference image size limit, no watermark.
image-n1-fast	Faster N1-Fast image generation and editing, no watermark.
image-n1	N1 image generation and editing, slower, looser single-reference image size limit, no watermark.
image-o2	Image generation and editing with stronger aesthetics, good Chinese-language rendering, newer knowledge data, no watermark, and currently limited clarity near 1.5K.
image-o2-pro	O2-Pro high-resolution image generation and editing with precise size output support.

Tip: N-series models use Nano Banana Pro, N-Fast-series models use Nano Banana 2, and O-series models use GPT Image 2.

Image Recognition Models

Model ID	Description
image-recognition-g1	G1 image recognition and visual analysis powered by Gemini 3.1 Pro.
image-recognition-g2	G2 image recognition and visual analysis powered by Gemini 3.5 Flash.

Video Models

Model ID	Description
video-v1-fast	Default faster Veo 3.1 video generation model with 4/6/8 second duration options and one optional reference image.
video-seedance-2	Seedance 2.0 video generation with up to 9 reference images, 3 reference videos, 3 reference audio clips, 4-15 second duration, and 480p/720p/1080p output. This is the only video model that requires whitelist access; contact Aoment customer service to apply before use.

text-to-image

Generate images from a text prompt. The default model is image-n2-fast (N2-Fast).

bash

uv run {baseDir}/scripts/aoment_image_video.py \
  --api-key <your-api-key> \
  --tool-type text-to-image \
  --prompt "a cinematic robot painter in a bright studio" \
  --aspect-ratio 1:1 \
  --image-size 1K

Parameter	Type	Required	Default	Description
--api-key / -k	string	yes	-	Agent API Key
--tool-type / -t	enum	yes	-	text-to-image
--prompt / -p	string	yes	-	Text prompt
--model	string	no	image-n2-fast	Image model ID. Available values: image-n2-fast, image-n2, image-n1-fast, image-n1, image-o2, image-o2-pro
--aspect-ratio	string	no	auto	auto, 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 21:9
--image-size	enum	no	1K	1K, 2K, 4K

image-to-image

Generate a new image from a prompt and a reference image. The reference image can be a URL or base64 image data.

bash

uv run {baseDir}/scripts/aoment_image_video.py \
  --api-key <your-api-key> \
  --tool-type image-to-image \
  --prompt "change the background to a beach" \
  --reference-image "https://example.com/photo.jpg"

Parameter	Type	Required	Default	Description
--api-key / -k	string	yes	-	Agent API Key
--tool-type / -t	enum	yes	-	image-to-image
--prompt / -p	string	yes	-	Transformation prompt
--reference-image	string	yes	-	Reference image as URL or base64 data
--model	string	no	image-n2-fast	Image model ID. Available values: image-n2-fast, image-n2, image-n1-fast, image-n1, image-o2, image-o2-pro
--aspect-ratio	string	no	auto	Output aspect ratio
--image-size	enum	no	1K	1K, 2K, 4K

video-generation

Generate a video from a prompt. The default video model is video-v1-fast. The CLI chooses a model-specific skill endpoint based on --model.

Dedicated skill HTTP endpoints:

Model	Endpoint
video-v1-fast	POST /api/skills/aoment-image-video/video-v1-fast
video-seedance-2	POST /api/skills/aoment-image-video/video-seedance-2

The old POST /api/skills/aoment-image-video/video-generation endpoint is removed and no longer accepts a model switch.

video-seedance-2 is the only video model with a whitelist requirement. If the Agent API Key user is not whitelisted, the API returns 403 model_access_denied; contact Aoment customer service to apply for whitelist access.

bash

uv run {baseDir}/scripts/aoment_image_video.py \
  --api-key <your-api-key> \
  --tool-type video-generation \
  --prompt "sunset beach timelapse" \
  --orientation landscape \
  --resolution 720p

Parameter	Type	Required	Default	Description
--api-key / -k	string	yes	-	Agent API Key
--tool-type / -t	enum	yes	-	video-generation
--prompt / -p	string	text-to-video yes	-	Video prompt; optional for video-seedance-2 when reference media is provided
--model	string	no	video-v1-fast	Video model ID. Available values: video-v1-fast, video-seedance-2
--orientation	enum	no	portrait	portrait or landscape
--aspect-ratio	string	no	auto	For video-seedance-2: adaptive, 16:9, 9:16, 4:3, 3:4, 1:1, or 21:9
--resolution	enum	no	720p	For video-v1-fast: 720p, 1080p, 4k; for video-seedance-2: 480p, 720p, 1080p
--duration	enum	no	8	For video-v1-fast: 4, 6, or 8; for video-seedance-2: 4-15 or -1; 1080p/4k uses 8 on video-v1-fast
--seedance-reference-mode	enum	no	multimodal	For video-seedance-2: multimodal, first_frame, or first_last_frame
--reference-image	string	no	-	Reference image as URL or base64 data; Seedance2.0 supports up to 9 images
--reference-video	string	no	-	Reference video as URL or base64 data; only video-seedance-2, up to 3 videos
--reference-audio	string	no	-	Reference audio as local path or base64 data; only video-seedance-2, up to 3 clips and must be paired with image/video

image-recognition

Analyze one or more images with a text prompt. The default recognition model is image-recognition-g2 (Gemini 3.5 Flash). Images can be local paths, URLs, or base64 image data.

bash

uv run {baseDir}/scripts/aoment_image_video.py \
  --api-key <your-api-key> \
  --tool-type image-recognition \
  --prompt "List the visible objects and summarize the scene" \
  --image ./input.png

Parameter	Type	Required	Default	Description
--api-key / -k	string	yes	-	Agent API Key
--tool-type / -t	enum	yes	-	image-recognition
--prompt / -p	string	yes	-	Recognition or analysis instruction
--image / -i	string	yes	-	Image as local path, URL, or base64 data; can be passed multiple times
--reference-image	string	no	-	Compatibility alias for image input; can be passed multiple times
--model	string	no	image-recognition-g2	Recognition model ID. Available values: image-recognition-g1, image-recognition-g2

hd-repair

Repair and upscale an image. This is provided by a separate script:

bash

uv run {baseDir}/scripts/aoment_hd_repair.py \
  --api-key <your-api-key> \
  --image ./input.png \
  --resolution 4K

Parameter	Type	Required	Default	Description
--api-key / -k	string	yes	-	Agent API Key
--image / -i	string	yes	-	Local path, URL, or base64 image data
--resolution	enum	no	4K	2K, 4K, 8K
--model	string	no	image-hd-repair	Only image-hd-repair is supported

Quota

Query remaining daily generation quota:

bash

uv run {baseDir}/scripts/aoment_quota.py --api-key <your-api-key>

Parameter	Type	Required	Description
--api-key / -k	string	yes	Agent API Key

If your daily quota is used up and you need more, join the community:

• Discord: https://discord.gg/3BMzRd7bJx
• QQ Group: 474397920 (Join via link)

Response Format

All scripts print JSON to stdout.

Successful text-to-image or image-to-image:

json

{
  "success": true,
  "tool_type": "text-to-image",
  "data": {
    "image_url": "https://cos.example.com/result.jpg?..."
  }
}

Successful video generation:

json

{
  "success": true,
  "tool_type": "video-generation",
  "data": {
    "video_url": "https://cos.example.com/result.mp4?..."
  }
}

Successful HD repair:

json

{
  "success": true,
  "tool_type": "hd-repair",
  "data": {
    "image_url": "https://cos.example.com/hd-repair-result.png?..."
  }
}

Successful image recognition:

json

{
  "success": true,
  "tool_type": "image-recognition",
  "data": {
    "result_text": "The image shows..."
  }
}

Successful quota query:

json

{
  "success": true,
  "data": {
    "remaining": 12,
    "quota": 15,
    "used": 3
  }
}

Error response:

json

{
  "success": false,
  "error": "error description"
}

Downloading Results

Returned image_url and video_url values are pre-signed COS URLs. Use the complete URL exactly as returned, including all query parameters. Do not strip the query string.

Example:

bash

uv run {baseDir}/scripts/aoment_image_video.py \
  -k <your-api-key> \
  -t text-to-image \
  -p "prompt" > result.json

curl -L -o output.jpg "$(python3 -c "import json; print(json.load(open('result.json'))['data']['image_url'])")"

Troubleshooting

1. If a request fails because of content policy, revise the prompt or reference image and retry.
2. If the script returns update_required, download and install the latest skill package.
3. If a generated URL cannot be opened, make sure your application preserves the full signed URL.
4. For help, join the Discord or QQ community listed above.