Skip to main content
Glama
mikeysrecipes

BlenderMCP

generate_hyper3d_model_via_images

Create 3D models with materials from images and import them into Blender for 3D modeling workflows. Adjust model dimensions using bounding box parameters.

Instructions

Generate 3D asset using Hyper3D by giving images of the wanted asset, and import the generated asset into Blender. The 3D asset has built-in materials. The generated model has a normalized size, so re-scaling after generation can be useful.

Parameters:

  • input_image_paths: The absolute paths of input images. Even if only one image is provided, wrap it into a list. Required if Hyper3D Rodin in MAIN_SITE mode.

  • input_image_urls: The URLs of input images. Even if only one image is provided, wrap it into a list. Required if Hyper3D Rodin in FAL_AI mode.

  • bbox_condition: Optional. If given, it has to be a list of ints of length 3. Controls the ratio between [Length, Width, Height] of the model.

Only one of {input_image_paths, input_image_urls} should be given at a time, depending on the Hyper3D Rodin's current mode. Returns a message indicating success or failure.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
input_image_pathsNo
input_image_urlsNo
bbox_conditionNo

Implementation Reference

  • The handler function decorated with @mcp.tool() that implements the logic for generating a Hyper3D model from input images (local paths or URLs). It validates inputs, processes images, handles bbox_condition, and sends a 'create_rodin_job' command to the Blender connection.
    @mcp.tool()
    def generate_hyper3d_model_via_images(
        ctx: Context,
        input_image_paths: list[str]=None,
        input_image_urls: list[str]=None,
        bbox_condition: list[float]=None
    ) -> str:
        """
        Generate 3D asset using Hyper3D by giving images of the wanted asset, and import the generated asset into Blender.
        The 3D asset has built-in materials.
        The generated model has a normalized size, so re-scaling after generation can be useful.
        
        Parameters:
        - input_image_paths: The **absolute** paths of input images. Even if only one image is provided, wrap it into a list. Required if Hyper3D Rodin in MAIN_SITE mode.
        - input_image_urls: The URLs of input images. Even if only one image is provided, wrap it into a list. Required if Hyper3D Rodin in FAL_AI mode.
        - bbox_condition: Optional. If given, it has to be a list of ints of length 3. Controls the ratio between [Length, Width, Height] of the model.
    
        Only one of {input_image_paths, input_image_urls} should be given at a time, depending on the Hyper3D Rodin's current mode.
        Returns a message indicating success or failure.
        """
        if input_image_paths is not None and input_image_urls is not None:
            return f"Error: Conflict parameters given!"
        if input_image_paths is None and input_image_urls is None:
            return f"Error: No image given!"
        if input_image_paths is not None:
            if not all(os.path.exists(i) for i in input_image_paths):
                return "Error: not all image paths are valid!"
            images = []
            for path in input_image_paths:
                with open(path, "rb") as f:
                    images.append(
                        (Path(path).suffix, base64.b64encode(f.read()).decode("ascii"))
                    )
        elif input_image_urls is not None:
            if not all(urlparse(i) for i in input_image_paths):
                return "Error: not all image URLs are valid!"
            images = input_image_urls.copy()
        try:
            blender = get_blender_connection()
            result = blender.send_command("create_rodin_job", {
                "text_prompt": None,
                "images": images,
                "bbox_condition": _process_bbox(bbox_condition),
            })
            succeed = result.get("submit_time", False)
            if succeed:
                return json.dumps({
                    "task_uuid": result["uuid"],
                    "subscription_key": result["jobs"]["subscription_key"],
                })
            else:
                return json.dumps(result)
        except Exception as e:
            logger.error(f"Error generating Hyper3D task: {str(e)}")
            return f"Error generating Hyper3D task: {str(e)}"
  • A prompt strategy that guides the usage of the tool, recommending when to use generate_hyper3d_model_via_images for asset creation.
    def asset_creation_strategy() -> str:
        """Defines the preferred strategy for creating assets in Blender"""
        return """When creating 3D content in Blender, always start by checking if integrations are available:
    
        0. Before anything, always check the scene from get_scene_info()
        1. First use the following tools to verify if the following integrations are enabled:
            1. PolyHaven
                Use get_polyhaven_status() to verify its status
                If PolyHaven is enabled:
                - For objects/models: Use download_polyhaven_asset() with asset_type="models"
                - For materials/textures: Use download_polyhaven_asset() with asset_type="textures"
                - For environment lighting: Use download_polyhaven_asset() with asset_type="hdris"
            2. Sketchfab
                Sketchfab is good at Realistic models, and has a wider variety of models than PolyHaven.
                Use get_sketchfab_status() to verify its status
                If Sketchfab is enabled:
                - For objects/models: First search using search_sketchfab_models() with your query
                - Then download specific models using download_sketchfab_model() with the UID
                - Note that only downloadable models can be accessed, and API key must be properly configured
                - Sketchfab has a wider variety of models than PolyHaven, especially for specific subjects
            3. Hyper3D(Rodin)
                Hyper3D Rodin is good at generating 3D models for single item.
                So don't try to:
                1. Generate the whole scene with one shot
                2. Generate ground using Hyper3D
                3. Generate parts of the items separately and put them together afterwards
    
                Use get_hyper3d_status() to verify its status
                If Hyper3D is enabled:
                - For objects/models, do the following steps:
                    1. Create the model generation task
                        - Use generate_hyper3d_model_via_images() if image(s) is/are given
                        - Use generate_hyper3d_model_via_text() if generating 3D asset using text prompt
                        If key type is free_trial and insufficient balance error returned, tell the user that the free trial key can only generated limited models everyday, they can choose to:
                        - Wait for another day and try again
                        - Go to hyper3d.ai to find out how to get their own API key
                        - Go to fal.ai to get their own private API key
                    2. Poll the status
                        - Use poll_rodin_job_status() to check if the generation task has completed or failed
                    3. Import the asset
                        - Use import_generated_asset() to import the generated GLB model the asset
                    4. After importing the asset, ALWAYS check the world_bounding_box of the imported mesh, and adjust the mesh's location and size
                        Adjust the imported mesh's location, scale, rotation, so that the mesh is on the right spot.
    
                    You can reuse assets previous generated by running python code to duplicate the object, without creating another generation task.
    
        3. Always check the world_bounding_box for each item so that:
            - Ensure that all objects that should not be clipping are not clipping.
            - Items have right spatial relationship.
        
        4. Recommended asset source priority:
            - For specific existing objects: First try Sketchfab, then PolyHaven
            - For generic objects/furniture: First try PolyHaven, then Sketchfab
            - For custom or unique items not available in libraries: Use Hyper3D Rodin
            - For environment lighting: Use PolyHaven HDRIs
            - For materials/textures: Use PolyHaven textures
    
        Only fall back to scripting when:
        - PolyHaven, Sketchfab, and Hyper3D are all disabled
        - A simple primitive is explicitly requested
        - No suitable asset exists in any of the libraries
        - Hyper3D Rodin failed to generate the desired asset
        - The task specifically requires a basic material/color
        """
  • Helper function used by the tool to normalize the bbox_condition input.
    if all(isinstance(i, int) for i in original_bbox):
        return original_bbox
    if any(i<=0 for i in original_bbox):
        raise ValueError("Incorrect number range: bbox must be bigger than zero!")
    return [int(float(i) / max(original_bbox) * 100) for i in original_bbox] if original_bbox else None

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A4.2/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the behavioral disclosure burden. It usefully mentions built-in materials, normalized sizing that may require re-scaling, that the asset is imported into Blender, and that a success/failure message is returned. However, it does not disclose whether generation is asynchronous, whether a job ID is produced, or what side effects occur beyond the import, which could be significant for an agent deciding how to proceed.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is compact and well-structured: a clear one-line purpose, a few meaningful behavioral notes, and a bullet-like parameter section. Every sentence adds value without redundancy, and the most important invocation constraints are placed prominently.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

The description covers the key parameters and gives relevant behavioral context, but it does not explain how an agent can determine the current Hyper3D Rodin mode, what image requirements exist, or what the success/failure message contains. Since there is no output schema and sibling polling tools exist, the lack of async/job details leaves some ambiguity for correct invocation.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The input schema has 0% description coverage, so the description fully compensates. It explains that input_image_paths must be absolute paths, that single values must still be wrapped in a list, that the three-element bbox_condition controls Length/Width/Height ratio, and that the two input parameters are mutually exclusive based on the current mode. This is strong parameter-level guidance.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description states a specific verb and resource: 'Generate 3D asset using Hyper3D by giving images' and 'import the generated asset into Blender.' It clearly distinguishes itself from the sibling text-based generation tool by emphasizing image input, so an agent can identify when this tool is relevant.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides concrete guidance on parameter selection: input_image_paths is for MAIN_SITE mode, input_image_urls is for FAL_AI mode, and exactly one must be provided. It does not explicitly name alternative generation tools or state when not to use this tool, but the image-vs-text distinction is clear enough to infer the primary use case.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.