NEW YEAR SALE: Get 15% Extra Credits, up to $150.Top Up Now!
vision-language
Molmo2-4B Prompt Optimizer: Enhance prompts for image and video generation with intelligent restructuring, style guidance, and context-aware improvements. Open-source vision-language model. Ready-to-use REST API, no cold starts, affordable pricing.
Idle
A vibrant, swirling painting in the style of Vincent van Gogh's "Starry Night." The night sky is filled with dynamic, swirling patterns in shades of blue, yellow, and white, with a prominent crescent moon in the upper right corner. Below, a small town with blue and green buildings, including a church with a tall steeple, is nestled among rolling hills and mountains. The scene is illuminated by warm yellow lights from the windows, creating a striking contrast against the cool night sky. The overall composition is rich in texture and movement, capturing the essence of van Gogh's post-impressionist style.
Your request will cost $0.003 per run.
For $1 you can run this model approximately 333 times.
ExamplesView all
README
Molmo2 Prompt Optimizer
Enhance your prompts for better AI generation results with Molmo2 Prompt Optimizer. This intelligent tool analyzes your input — whether text or image — and generates optimized prompts tailored for different styles and output modes. Perfect for improving text-to-image and text-to-video generation quality.
Why It Works Great
- Dual input modes: Analyze images or text to generate optimized prompts.
- Multiple styles: Optimize for artistic, photographic, technical, anime, or realistic output.
- Image & video support: Generate prompts optimized for either image or video generation.
- Instant results: Fast processing for seamless workflow integration.
- Ultra-affordable: Just $0.003 per optimization — 333 runs for $1.
- Quality boost: Get better results from your generation models.
Parameters
| Parameter | Required | Description |
|---|---|---|
| image | No | Reference image to analyze and describe. |
| text | No | Text prompt to enhance and optimize. |
| style | No | Output style: default, artistic, photographic, technical, anime, or realistic. |
| mode | No | Target generation type: image or video. Default: image. |
How to Use
From Image
- Upload an image — the image you want to describe or recreate.
- Select style — choose the aesthetic direction.
- Select mode — image or video generation target.
- Run — get an optimized prompt describing the image.
From Text
- Enter your prompt — your basic idea or description.
- Select style — choose the aesthetic direction.
- Select mode — image or video generation target.
- Run — get an enhanced, detailed prompt.
Pricing
Flat rate per optimization.
| Output | Cost |
|---|---|
| Per optimization | $0.003 |
| 100 optimizations | $0.30 |
| 1,000 optimizations | $3.00 |
Style Guide
| Style | Description | Best For |
|---|---|---|
| default | Balanced, general-purpose optimization | Any content type |
| artistic | Creative, expressive, painterly language | Art, illustrations, creative work |
| photographic | Camera, lens, and lighting terminology | Photos, portraits, products |
| technical | Precise, detailed specifications | Technical diagrams, precise output |
| anime | Japanese animation style keywords | Anime characters, manga art |
| realistic | Photorealistic, lifelike descriptions | Realistic renders, simulations |
Mode Options
| Mode | Description |
|---|---|
| image | Optimizes prompts for text-to-image models |
| video | Optimizes prompts for text-to-video models with motion descriptions |
Best Use Cases
- Image Captioning — Generate detailed prompts from reference images.
- Prompt Enhancement — Upgrade basic prompts to detailed descriptions.
- Style Conversion — Reframe prompts for different aesthetic styles.
- Cross-model Optimization — Adapt prompts for image vs video generation.
- Batch Processing — Optimize many prompts affordably at scale.
- Learning Tool — Understand what makes effective generation prompts.
Pro Tips for Best Results
- Use image input to reverse-engineer prompts from existing artwork.
- Match style to your target model's strengths (anime for anime models, etc.).
- Switch mode to "video" when targeting video generation for motion-aware prompts.
- Combine: upload image + add text for context-aware optimization.
- At $0.003 per run, experiment freely to find optimal prompt styles.
Notes
- Provide either image or text (or both) as input.
- Processing is near-instant for rapid iteration.
- Video mode adds motion and temporal descriptions to prompts.
- Results can be directly used with generation models on WaveSpeed.