Manus' popular MCP enables Claude to automate 3D modeling with just one sentence, netizen: True AI+application

qubit

follow with interest

2025-03-15

0 comments

0 views

0 favorites

13 minutes

  There are four main career paths for product managers: professional line, management line, project line, and independent entrepreneurship. Management line refers to transitioning to a management position and leading a team

Recently, Manus' popular MCP (Model Context Protocol) protocol has attracted widespread attention. Through MCP, AI tools such as Claude can seamlessly integrate with professional software (such as Blender) to achieve automated operations, such as converting 2D images into 3D models, and even building interactive web pages in minutes. This technology not only greatly improves work efficiency, but also opens new doors for the application of AI in more fields. This article will delve into the working principle of MCP, practical application cases, and its potential impact on the future development of AI.

One sentence prompt: Claude automatically opens Blender to convert 2D images into 3D models.

The whole process flows smoothly.

And it can also use only one prompt word to build interactive web pages based on this scenario.

The key behind it is the recently popular MCP (Model Context Protocol) - an important trick for replicating Manus.

By integrating this protocol with Blender, the above effect can be obtained.

The modeling work that used to take several hours to complete manually has now been reduced to a few minutes without the need for human intervention.

How can B2B product managers grow rapidly?

Product and business architecture mainly involves layering and organizing the entire business workflow, abstracting individual requirements, mapping business requirements to products in a reasonable and reasonable manner, and ultimately enabling business data to flow, execute, record, and use within the product.

View details>

The open-source project BlenderMCP has reached 3.8k GitHub stars in just 3 days since its launch.

And the modeling effect it produces is trustworthy. Someone personally tested it and said that if it were to design Martian terrain, Claude could handle errors and problems on his own and inform humans.

No wonder someone exclaimed after reading it: humans no longer need design tools much, amazing！

AI+application tools are becoming increasingly powerful.

It is worth mentioning that this solution can be replicated on other open-source professional tools.

For example, someone has already implemented MCP+QGIS (Geographic Version "PS"), using Claude automation for sensing mapping.

Blender/Cursor can both be MCP

Simply put, BlenderMCP connects Blender to Claude, allowing Claude to directly interact with Blender and control it.

Many things can be accomplished based on BlenderMCP.

For example, creating a dungeon scene with a dragon guarding a can of gold.

Tip words:

Create a low poly scene in a dungeon, with a dragon guarding a pot of gold.

During this process, the instruction adherence was effective.

Special emphasis was placed on low poly, and the final result showed that both the dragon and the jar were round and plump.

You can also create realistic beach scenes.

Tip words:

Create a beach vibe using HDRIs, textures, and models like rocks and vegetation from Poly Haven.

This instruction requires the use of HDRIs, textures, rocks, vegetation, etc. from Poly Haven to model the beach.

Poly Haven is a free and open-source 3D resource website where Claude can directly download and use the resources himself.

Other abilities that can be tried include:

Paint this car red with a metallic texture

Create a sphere and place it above the cube

Set the lighting to the effect of a studio

Aim the camera at the scene and maintain an equidistant perspective

The author introduces in the project page that BlenderMCP's capabilities include creating, modifying, and deleting 3D objects; Use and modify materials and colors; Scene inspection and code execution.

This system mainly consists of two parts, Blender Addon and MCP Server.

The former is a Blender plugin that can create a server in Blender that accepts and executes commands. The latter is used to implement MCP.

The specific installation method has been fully open sourced by the author on GitHub.

In addition to integrating MCP into Blender, netizens are also trying to upgrade various tools using it.

Even AI programming software will become more automated after using MCP.

Someone used the MCP protocol on cursor to simultaneously access Slack and GitHub, completing a new feature development.

After configuring the plugin and completing authentication, cursor automatically reads the requirements document from Slack through MCP, then pulls the code from GitHub and automatically completes the writing and uploading of new features.

This operation utilizes an MCP service provided by an organization called Compsio, which can be directly configured through a link in the cursor.

GitHub, Google search, email, maps... have all been turned into MCP services by Composio.

In addition to Compio, MCP enthusiasts have also established their own MCP community, providing a vast amount of open-source server and client resources.

For example, this MCP service can retrieve papers from arXiv, and after configuring according to the tutorial, you can directly find papers in the Claude client.

Interestingly, large models can also be "MCP serviced", such as allowing servers to call other models through OpenAI compatible APIs.

Even integrating DeepSeeker R1 into Claude is not a problem.

Why is MCP so powerful?

MCP is a communication protocol proposed by Anthropic, which is now compared to the Type-C interface of AI applications.

And Anthropic has planned to take the lead in promoting the MCP protocol as an industry open standard.

Seamless integration between large model applications and external data sources and tools helps AI obtain the necessary contextual data and generate higher quality, task relevant answers.

MCP mainly addresses a common pain point faced by global application players - data isolation.

It is like a bridge between AI systems and data sources, allowing developers to establish bidirectional connections between data sources and AI tools.

MCP adopts a client server architecture, where multiple services can connect to any compatible client. The client can be Claude Desktop, IDE, or other AI tools, while the server acts as an adapter, exposing the data source.

Its advantage lies in the fact that in the future, whether accessing local resources (databases, files, services) or remote resources (such as Slack, GitHub APIs), the same protocol can be used.

And the supported data formats are very diverse, including file content, database records, API responses, real-time system data, screenshots and images, log files, etc., covering almost all types.

The MCP server also has built-in security mechanisms that allow the server to control resources on its own without having to hand over API keys to the large model.

According to the service source, MCP mainly uses communication mechanisms, with standard input/output used for local communication and SSE used for remote communication.

Both communication methods use JSON format for message transmission, which enables standardization of MCP communication process and brings scalability.

It seems that MCP can call many and complex services, but in reality, the development process is very simple.

At the time of release, the official announcement clearly stated that the latest Claude 3.5 Sonnet was already very skilled at setting up MCP servers and completing closed loops directly.

With powerful calling capabilities, convenient development processes, backed by Anthropic, and also gaining attention from the open source community, MCP seems to have the potential to become a future AI standard as envisioned by Anthropic.

But can it really be like this?

There are actually quite a few people who hold a wait-and-see attitude or a pessimistic attitude.

Recently, the well-known open-source big model framework LangChain also conducted an official vote on X.

40.8% of people believe that MCP is the future standard, while more people feel that we need to take another look.

There have also been some disagreements within LangChain.

The CEO believes that MCP has lowered the threshold for Agent access tools.

The founding engineer believes that at the engineering level, there will be many customized requirements, and in many cases, MCP cannot fully play its role.

MCP needs to become like OpenAI's GPTs in order to match its popularity, but in reality, GPTs do not seem to be as popular.

What's your opinion? Will MCP be a flash in the pan?

Welcome to leave comments and discuss in the comment section~

GitHub address:

https://github.com/ahujasid/blender-mcp?tab=readme -ov-file

Reference link:

[1] https://x.com/bilawalsidhu/status/1900240156826939560

[2] https://x.com/bilawalsidhu/status/1900632591516008599

[3] https://x.com/mattpocockuk/status/1898789901824590328

[4] https://x.com/KaranVaidya6/status/1898439847322525963

[5] https://blog.langchain.dev/mcp-fad-or-fixture/

Minmin Cressy from Wafeisi Quantum Bit | official account QbitAI

This article is authored by everyone who is a product manager [Qbit]. The WeChat official account is [Qbit]. The original/authorized release is made by everyone who is a product manager. Reproduction without permission is prohibited.

The title image is from Unsplash, based on the CC0 protocol.

Manus带火的MCP，让Claude一句话自动化3D建模，网友：真·AI+应用