generate-image – skill Thiết kế & UI cho agent AI

Generate Image

Generate and edit images through OpenRouter's Image API, which reaches Gemini, Seedream, Recraft,

GPT-Image, Riverflow, and other models behind one request shape.

When to use

Use this skill for: photos and photorealistic images, illustrations and artwork, concept art,

presentation and poster visuals, logos and vector marks, image editing, and compositing from

reference images.

Use scientific-schematics instead for: flowcharts, circuit diagrams, biological pathways,

system architecture diagrams, CONSORT diagrams, and other technical schematics.

API key

Generation requires an OpenRouter key. The script resolves it in this order:

1. --api-key

2. the OPENROUTER_API_KEY environment variable

3. OPENROUTER_API_KEY= in a .env file, searching the working directory upward, then the

script's own directory

If none is present the script exits with setup instructions. Keys: https://openrouter.ai/keys

--list-models, --model-info, and --dry-run need no key.

Quick start

# Generate
python scripts/generate_image.py "A beautiful sunset over mountains"

# Edit an existing image
python scripts/generate_image.py "Make the sky purple" -i photo.jpg -o edited.png

Run these commands from the skill directory; input and output paths resolve from the working

directory. Output defaults to generated_image.<ext>, using each image's returned media type

(or .bin when unknown). An explicit output suffix is kept, with a warning on mismatch; the

script does not transcode. OpenRouter and upstream costs are printed separately when supplied.

Then look at the image. Read the file back and check it before using it anywhere: composition,

aspect ratio, and any text are all things models get wrong silently.

Choosing a model

Default: google/gemini-3.1-flash-image.

| Need | Model |

| --- | --- |

| General quality, prompt adherence | google/gemini-3.1-flash-image |

| Highest Gemini tier | google/gemini-3-pro-image |

| Cheap iteration | google/gemini-3.1-flash-lite-image (1K only), openai/gpt-image-1-mini |

| Photoreal control, seeded generation | bytedance-seed/seedream-4.5 |

| Several images per request | bytedance-seed/seedream-4.5, openai/gpt-image-2 (up to 10) |

| Vector / SVG output | recraft/recraft-v4.1-vector |

| Transparent background | openai/gpt-image-1 with --background transparent |

| Legible text inside the image | recraft/recraft-v4.1, sourceful/riverflow-v2.5-pro — see the caveat below |

[references/models.md](references/models.md) carries the reviewed capability and pricing snapshot. The live listing is authoritative and free:

python scripts/generate_image.py --list-models            # every model and its allowed values
python scripts/generate_image.py --list-models gemini     # filtered by substring
python scripts/generate_image.py --model-info openai/gpt-image-1   # one model, plus pricing

Parameter support varies by model

This is the main thing to get right. Models advertise different parameter sets **and different

allowed values**. The CLI conservatively rejects unadvertised options; individual providers may

ignore or normalize some fields. The generic guide can lag the discovery records.

The script checks the model catalogue and requires at least one provider endpoint to accept the

complete combination before generation. Model capabilities are a union across endpoints, not a

promise that every provider supports them. Unsupported requests fail locally with legal values:

$ python scripts/generate_image.py "abstract pattern" -m openai/gpt-image-2 --background transparent
Error: Request rejected before billing (1 problem):
  - background=transparent is not allowed; this model accepts: auto, opaque

Current examples of capability differences (reviewed 2026-09-30):

• --resolution: Gemini 3.1 Flash accepts 512, 1K, 2K, 4K; Flash Lite and Krea

accept only 1K. Seedream 5.0 Lite accepts 2K/4K, while 5.0 Pro accepts 1K/2K.

• --output-format: Riverflow 2.5 Pro accepts png/jpeg/webp; its Fast variant accepts

only jpeg. Recraft vector models accept svg. OpenAI models do not advertise this flag.

• --quality: GPT Image 2.5 Flare/Sunburst also accept xhigh/max; Grok 2.0 accepts

only low/medium. Inspect the specific model before selecting a quality value.

• --background transparent is not available on gpt-image-2 or gpt-5.4-image-2.

It is advertised on GPT Image 1 and 2.5. Do not combine transparency with JPEG; when no

output format is selectable, inspect the returned file for an actual alpha channel.

• --aspect-ratio: GPT-5 Image and Mini now accept 1:1, 3:2, 2:3, auto;

GPT-5.4 Image 2 also accepts wider ratios. None accept a resolution flag in this snapshot.

• --n is an upper bound, not a guarantee: providers may return fewer images. Seedream 4.5

caps at 10, Seedream 5.0 Lite at 4, and 5.0 Pro at 1. Krea does not advertise n.

• --seed is available on Seedream, Krea, Qwen and FLUX, but not Gemini or OpenAI.

Keep the model/provider, prompt and parameters fixed; a seed is not a cross-version guarantee.

--dry-run performs free discovery and prints the request without generating. If discovery

fails, the command stops; --no-preflight explicitly bypasses it. Combining both flags is an

offline payload inspection, not validation of provider support.

Writing the prompt

Prompt quality decides output quality more than model choice does. Name, in one sentence each:

1. Subject — what is in frame, and how much of it. "A single pipette tip above a 96-well plate."

2. Medium and style — photograph, watercolour, 3D render, flat vector, scientific illustration.

3. Lighting and palette — "soft diffuse lighting, cool blue and white palette."

4. Composition — "wide shot, subject left of centre, empty space on the right for a title."

5. What to avoid — "no text, no labels, no watermark."

Asking for empty space where a caption or title will go is the single most useful compositional

instruction for posters and slides.

Iterate cheaply: draft on gemini-3.1-flash-lite-image, then regenerate the wording you settled on

with the model you actually want. To refine rather than restart, feed the last output back as a

reference (-i out.png) and describe only the change.

Editing and reference images

-i/--input is repeatable and accepts local paths, HTTP(S) URLs, or data URLs. Local files are

base64-encoded and sent as input_references.

# Single-image edit
python scripts/generate_image.py "Add sunglasses to the person" -i portrait.png

# Composite several references
python scripts/generate_image.py "Blend these two styles" -i style_a.png -i style_b.jpg -o blend.png

# Reference an image already on the web
python scripts/generate_image.py "Restyle as a watercolor" -i https://example.com/photo.jpg

Reference ranges include both a minimum and a maximum: OpenAI accepts up to 16, Gemini 3 and

Seedream up to 14, Riverflow Pro up to 10, and MAI 2.6 up to 5. Recraft v4 Styles requires

1–10 references; its regular v4/v4.1 models accept 0–1. Ming Design Layer requires exactly one,

while Ming Design and Recraft Flash are text-only. The CLI checks required and zero-reference

limits too. Local formats supported by the helper: PNG, JPEG, GIF, WebP; provider decoding rules

still apply. Local files and data URLs send their bytes; URL references are fetched by the service.

Do not use private or credential-bearing URLs. Riverflow v2 bills $0.20 per reference image.

Worked examples

The -o paths are destinations the script creates, not bundled files. These are generation

examples, not authenticated end-to-end results from this review. Their request parameters were

checked against public discovery, and local encoding/response handling was tested with mocks.

# Wide hero image for a poster, with space reserved for the title
python scripts/generate_image.py \
  "Laboratory with modern equipment, photorealistic, well-lit, wide shot, \
   equipment on the left, empty wall on the right, no text" \
  --aspect-ratio 21:9 --resolution 2K -o poster/hero.png

# Conceptual illustration for a manuscript — illustrative, never presented as data
python scripts/generate_image.py \
  "Stylised illustration of immune cells surrounding a tumour cell, scientific illustration, \
   cool palette, no text" \
  --resolution 2K -o figures/immunotherapy_concept.png

# Vector logo
python scripts/generate_image.py \
  "Minimal geometric fox logo, two colors" \
  -m recraft/recraft-v4.1-vector --output-format svg -o assets/logo.svg

# Slide background with a transparent alpha channel
python scripts/generate_image.py \
  "Abstract molecular pattern, subtle, blue and white, no text" \
  -m openai/gpt-image-1 --background transparent -o slides/bg.png

# Four variations in one request
python scripts/generate_image.py \
  "Stylized neuron network ill

… (nội dung đầy đủ có trong gói tải về)