Menu
Single image input for world generation - requirements, best practices, and examples
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
npx skills add https://github.com/CloudAI-X/world-labs-skills --skill world-labs-image-promptThe command stays on one line. Scroll horizontally to inspect it before copying.
Prefer a local copy? Download the files currently available to SkillsMP.
Based on SOC occupation classification
World Labs Marble API integration - authentication, endpoints, models, and media uploads
Chisel 3D blocking tool for world generation - geometry, walls, and reference models
World Labs export formats - Gaussian splats, meshes, images, and engine integrations
Showing SKILL.md
| name | world-labs-image-prompt |
| description | Single image input for world generation - requirements, best practices, and examples |
| allowed-tools | ["Bash","WebFetch"] |
Generate 3D worlds from a single reference image. The model extrapolates the scene to create an immersive 360° environment.
| Requirement | Specification |
|---|---|
| Recommended resolution | 1024px on long side |
| Maximum file size | 20 MB |
| Formats | PNG (recommended), JPG, WebP |
| Aspect ratio | 16:9, 9:16, or anything in between |
| Input Type | Marble 0.1-plus | Marble 0.1-mini |
|---|---|---|
| Standard image | 1,580 | 230 |
| Panorama image | 1,500 | 150 |
Panoramas skip the pano generation step, saving 80 credits.
✅ Clear spatial definition: Images with obvious depth and perspective ✅ Wide shots: Show foreground, midground, and background ✅ Visible ground/floor: Helps establish world orientation ✅ Consistent lighting: Clear, well-lit scenes ✅ Environmental scenes: Landscapes, interiors, architectural spaces
❌ : Lack spatial context for 3D reconstruction ❌ : As main subjects (may cause artifacts) ❌ : Non-representational art ❌ : Decorative edges, watermarks ❌ : Text doesn't render clearly ❌ : 2D illustrations without depth ❌ : Reduces detail inference
First upload your image (see world-labs-api skill), then:
{
"model": "Marble 0.1-plus",
"world_prompt": {
"type": "image",
"image_prompt": {
"source": "media_asset",
"media_asset_id": "550e8400-e29b-41d4-a716-446655440000"
},
"text_prompt": "Optional description to guide interpretation"
}
}
{
"model": "Marble 0.1-plus",
"world_prompt": {
"type": "image",
"image_prompt": {
"source": "uri",
"uri": "https://example.com/my-image.jpg"
},
"text_prompt": "A beautiful mountain landscape"
}
}
is_pano flag)For 360° equirectangular panoramas (2:1 aspect ratio):
{
"model": "Marble 0.1-plus",
"world_prompt": {
"type": "image",
"image_prompt": {
"source": "media_asset",
"media_asset_id": "550e8400-e29b-41d4-a716-446655440000",
"is_pano": true
}
}
}
curl -X POST "https://api.worldlabs.ai/marble/v1/media-assets:prepare_upload" \
-H "WLT-Api-Key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"file_name": "landscape.jpg",
"kind": "image",
"extension": "jpg"
}'
Response:
{
"media_asset": {
"media_asset_id": "550e8400-e29b-41d4-a716-446655440000"
},
"upload_info": {
"upload_url": "https://storage.googleapis.com/...",
"upload_method": "PUT",
"required_headers": {
"x-goog-content-length-range": "0,1048576000"
}
}
}
curl -X PUT "UPLOAD_URL" \
-H "Content-Type: image/jpeg" \
-H "x-goog-content-length-range: 0,1048576000" \
--data-binary @landscape.jpg
curl -X POST "https://api.worldlabs.ai/marble/v1/worlds:generate" \
-H "WLT-Api-Key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "Marble 0.1-plus",
"world_prompt": {
"type": "image",
"image_prompt": {
"source": "media_asset",
"media_asset_id": "550e8400-e29b-41d4-a716-446655440000"
},
"text_prompt": "A dramatic mountain landscape at sunset"
}
}'
import requests
def upload_image_and_generate(image_path: str, api_key: str, prompt: str = None, is_pano: bool = False):
base_url = "https://api.worldlabs.ai/marble/v1"
headers = {"WLT-Api-Key": api_key, "Content-Type": "application/json"}
# Get extension
ext = image_path.lower().split('.')[-1]
if ext == "jpeg":
ext = "jpg"
# Step 1: Prepare upload
prep_response = requests.post(
f"{base_url}/media-assets:prepare_upload",
headers=headers,
json={"file_name": image_path.split('/')[-1], "kind": "image", "extension": ext}
)
prep_data = prep_response.json()
media_asset_id = prep_data["media_asset"]["media_asset_id"]
upload_url = prep_data["upload_info"]["upload_url"]
# Step 2: Upload image
content_types = {"jpg": "image/jpeg", "png": "image/png", "webp": "image/webp"}
with open(image_path, 'rb') as f:
requests.put(
upload_url,
headers={"Content-Type": content_types.get(ext, "image/jpeg")},
data=f.read()
)
# Step 3: Generate world
image_prompt = {"source": "media_asset", "media_asset_id": media_asset_id}
if is_pano:
image_prompt["is_pano"] = True
world_prompt = {"type": "image", "image_prompt": image_prompt}
if prompt:
world_prompt["text_prompt"] = prompt
gen_response = requests.post(
f"{base_url}/worlds:generate",
headers=headers,
json={"model": "Marble 0.1-plus", "world_prompt": world_prompt}
)
return gen_response.json()["operation_id"]
# Usage
operation_id = upload_image_and_generate(
"mountain_vista.jpg",
"your_api_key",
"Add dramatic storm clouds and golden sunset light"
)
Text prompts guide interpretation and can transform the scene:
| Text Prompt | Effect |
|---|---|
| None (omitted) | Auto-caption generated, faithful recreation |
| "At sunset" | Changes lighting/atmosphere |
| "In winter with snow" | Adds seasonal elements |
| "Abandoned and overgrown" | Adds decay/nature reclaim |
| "Futuristic version" | Sci-fi transformation |
When text is omitted, the model auto-generates a caption from the image.
| Issue | Solution |
|---|---|
| Distorted geometry | Use image with clearer depth cues |
| Missing areas | Provide image with more context/edges |
| Wrong scale | Include recognizable objects for reference |
| Artifacts on faces | Avoid images with people as main subject |
| Billboard warping | Objects far from center may warp; center important elements |
world-labs-api - API integration detailsworld-labs-text-prompt - Text prompting best practicesworld-labs-multi-image - Using multiple images with direction controlworld-labs-pano-video - Panorama and video input