Install
openclaw skills install eric-image-craftCurated image generation assistant covering four categories — characters, scenes, products, and style transformation — for generating high-quality stylized 3...
openclaw skills install eric-image-craftImageCraft is a curated image generation assistant that produces high-quality stylized 3D images across four categories: Characters, Scenes, Products, and Style Transformation. It covers 17 distinct styles, each with dedicated prompt templates and execution logic.
<deliver_assets>
<item>
<path>image path</path>
</item>
</deliver_assets>
One <item> block per image; multiple images go inside the same <deliver_assets> block.
| # | Style | Required Input | Optional Input |
|---|---|---|---|
| 1 | Collectible Action Figure | Character photo or illustration | — |
| 2 | Cartoon Portrait | Character photo | — |
| 3 | Chibi Character | Character photo | Action, expression |
| # | Style | Required Input | Optional Input |
|---|---|---|---|
| 4 | City Miniature Diorama | City name | — |
| 5 | Landmark Architectural Render | Landmark name | — |
| 6 | Movie Scene Recreation | Movie name + scene name | — |
| 7 | Isometric Room | Room theme/description | Atmosphere, light source |
| 8 | City Weather Visualization | City name + weather | Date, temperature |
| # | Style | Required Input | Optional Input |
|---|---|---|---|
| 9 | Sticker-Bombed Logo | Logo image or brand name | — |
| 10 | Chibi Brand Store | Brand name | Brand color |
| 11 | Product 3D Render | Product image or product description | — |
| 12 | Product Ad Design | Product image (with model is better) + brand name | Selling points, copy, brand color |
| # | Style | Required Input | Optional Input |
|---|---|---|---|
| 13 | Low-Poly Style | Subject image or description | Color scheme |
| 14 | Meme to 3D | Meme image | — |
| 15 | Glasses-free 3D Effect | Scene description | — |
| 16 | Knolling Arrangement | City name | — |
| 17 | Stylized 3D Character | Character photo | — |
genui-form-wizardgenui-form-wizard for selectionimages_understand to extract featuresgen_images:
reference_files<deliver_assets> format. No explanation.IMPORTANT: Using plain text tables instead of genui-form-wizard for interactive selection is forbidden. Always use genui-form-wizard when the user needs to make a choice.
Trigger conditions:
Execution:
Example selection principles:
Prompt Template:
Create a 1/7 scale commercialized 3D action figure of [CHARACTER DESCRIPTION]. Use a realistic style with accurate proportions and surface details. Place the figure on a circular transparent acrylic base. Use studio-quality lighting, shallow depth of field, and photorealistic materials. High detail, clean composition, professional collectible figure presentation.
Prompt Template:
Create a playful 3D caricature portrait of [CHARACTER]. Blend cartoon-style exaggeration with realistic skin shading. Use an oversized head, stylized hair, and large expressive eyes. Apply soft cinematic lighting with clean, simplified materials. Keep the background minimal with a gentle blur.
Prompt Template:
Create a chibi figurine-style 3D character based on [CHARACTER]. The figure has a big head and small body, made of matte PVC material. [ACTION] pose with [EXPRESSION] expression. Photoreal materials, neutral background, ultra-clean composition.
images_understand to extract features. Identify clothing, hairstyle, expression, pose.gen_images; if reference image exists, put in reference_files. Clear input → 1 image; example mode → 2 images.Prompt Template:
Create a hyper-realistic 3D square diorama of [CITY]. The city appears carved out as a solid block with a visible underground cross-section showing soil, rocks, roots, and earth layers. Above the ground, display a whimsical fairytale-style cityscape featuring iconic landmarks and cultural elements of [CITY]. Use a pure white studio background with soft natural lighting. DSLR photo quality, crisp details, vibrant colors, magical realism style. 1080x1080 resolution.
Prompt Template:
Create a highly detailed isometric 3D rendering of [LANDMARK] in professional architectural visualization style. Show the structure at a 45-degree angle from above. Use photorealistic textures such as stone, glass, metal, and brick. Include a detailed base with tiny people, cars, trees for scale. Clean white background with soft ambient shadows. 1080x1080 resolution.
Prompt Template:
Present a clear 45-degree top-down isometric miniature 3D cartoon scene of [SCENE NAME] from [MOVIE]. Use refined textures, realistic PBR materials, and soft lifelike lighting. Create a raised diorama-style base with the most recognizable elements. Display the movie title at top center in large bold text. 1080x1080 resolution.
Prompt Template:
Create an isometric 3D cube-shaped miniature room with a shallow cutaway. Room description: [ROOM THEME, FURNITURE, DECOR]. Lighting: [ATMOSPHERE], using [LIGHT SOURCES]. Include realistic reflections, soft colored shadows. Camera: slightly elevated isometric three-quarter view, cube centered. Photoreal materials, neutral background. No watermark.
Prompt Template:
Present a clear 45-degree top-down isometric miniature 3D cartoon scene of [CITY]. Feature iconic landmarks. Use soft refined textures, realistic PBR materials. Integrate [WEATHER] conditions into the environment. At top center, place "[CITY]" in large bold text, weather icon, date, and temperature. 1080x1080.
web_search to get iconic elements; movie scene needs to confirm specific scene name.gen_images; default generate 2 variants.Prompt Template:
Create a hyper-realistic 3D physical object shaped like [LOGO/BRAND]. Apply soft studio lighting. Cover the object with a dense sticker-bomb collage in Y2K and retro 90s style. Include acid graphics, bold typography, smiley faces, stars, and vector badges. Stickers wrap naturally around curves with slight peeling edges and high-resolution textures. Isolated black background. Octane render, 8K quality.
Prompt Template:
Create a 3D chibi-style miniature concept store of [BRAND]. Design the exterior inspired by the brand's most iconic product. The store has two floors with large glass windows revealing a cozy interior. Use [BRAND COLOR] as primary color theme, with warm lighting and staff in brand uniforms. Add adorable tiny figures walking, sitting along the street. Include benches, street lamps, potted plants. Render in miniature cityscape style, blind-box toy aesthetic, high detail, soft afternoon lighting. Aspect ratio 2:3.
Prompt Template:
Create a photorealistic 3D render of [PRODUCT]. Use studio-quality lighting with soft shadows. Show material details and textures clearly. Clean white or gradient background. Professional product photography style. 8K quality.
Prompt Template:
An advertising image presents a large [PRODUCT] and a model on a two-toned [BRAND COLOR] background. The backdrop is a gradient of [BRAND COLOR], with a brighter, more light hue at the top, transitioning to a deeper, darker shade at the bottom, creating a subtle, reflective surface.
On the left, a magnified view of a chunky high quality [PRODUCT] in [PRODUCT COLOR] dominates the lower portion of the frame. The [PRODUCT] features intricate lines and details, multiple panels, and textured surfaces. The [PRODUCT] is shiny and smooth. The [PRODUCT] is oriented with its main surface facing the viewer, leaning slightly to the right, rotated upright and standing vertically, creating a dramatic presentation angle.
Leaning against the side of the large [PRODUCT], on the right side of the image, is [MODEL DESCRIPTION - include: age, appearance, skin tone, facing direction, body angle, head position, hair style, clothing description with colors and fabric, arm positions, and note that model wears the same product].
The overall lighting suggests a soft, studio setup, casting minimal shadows and highlighting the subjects against the vibrant [BRAND COLOR].
Include brand logo [BRAND] at top corner. Add headline text '[HEADLINE]' in bold. Add product feature tags: [FEATURES]. Add large decorative brand name text in background. Professional advertising design, 1080x1080.
gen_images; put Logo image in reference_files. Clear input → 1 image; example mode → 2 images.Prompt Template:
A low-poly 3D render of [SUBJECT], constructed from clean triangular facets and shaded in flat [COLOR1] and [COLOR2] tones. Set in a stylized minimalist environment with crisp geometry and soft ambient occlusion. Playful digital diorama with sharp edges and visual simplicity.
Prompt Template:
Turn [MEME DESCRIPTION] into a photorealistic 3D render. Keep composition identical. Convert the character into a plush toy with realistic lighting and materials.
Prompt Template:
An enormous L-shaped glasses-free 3D LED screen at a bustling urban intersection, iconic architectural style like Shinjuku Tokyo. The screen displays a captivating glasses-free 3D animation featuring [SCENE DESCRIPTION]. Characters and objects possess striking depth, extending outward and floating in mid-air. Under realistic daylight, elements cast lifelike shadows onto screen surface and surrounding buildings. Rich detail, vibrant colors, seamlessly integrated with urban setting.
Prompt Template:
Present a clear, directly top-down photograph of [CITY] landmarks as 3D magnets, arranged neatly in parallel lines and right angles, knolling. Realistic miniatures. At top-center, place city name as souvenir magnet, and handwritten post-it note for temperature and weather. Incorporate weather-appropriate items into the knolling. No repeats.
Prompt Template:
Transform the subject into a stylized 3D character with soft clay-like materials. Use rounded sculptural forms, exaggerated facial features, and a pastel plus vibrant color palette. Apply smooth subsurface scattering skin, large cartoon eyes, simplified anatomy. Render on bold blue studio background with soft frontal lighting and subtle shadows. Keep original photo's composition and framing.
images_understand to analyze the original image; identify subject, composition, key elements.gen_images; put original image in reference_files. Default generate 2 variants.| Tool | Purpose | When to Use |
|---|---|---|
images_understand | Analyze reference images | When user provides a photo/image as input |
gen_images | Generate images from prompts | Core generation step for every style |
web_search | Look up real-world information | City landmarks, brand details, movie scenes |
genui-form-wizard | Interactive selection UI | Any time user must choose between options |
genui-form-wizardreference_files when calling gen_images<deliver_assets> XML block