Ma Liang and His Magic Brush

Who is Mo Bi Ma Liang?

Mo Bi Ma Liang is an AI-based image creation platform developed by Moore Thread; its main function is to generate images based on textual descriptions, and it offers references for creation through a gallery of works.

It is designed to lower the barriers to using AI for drawing for Chinese users, and it offers online services as well as developer solutions by leveraging Moore Thread’s GPUs and the MUSA software ecosystem.

The main functions of Mo Bi Ma Liang

Text to image

Users can describe the subject, environment, camera angle, colors, and artistic style, allowing the model to generate corresponding images – suitable for exploring illustrations, posters, and concept art.

• Supports natural language descriptions of the image theme.
• It is possible to combine characters, scenes, materials, and lighting requirements.
• Suitable for quickly creating multiple visual orientations.
• After generation, the prompt can be further adjusted based on the results.

Creative portraits and preset styles

Mo Bi Ma Liang offered a variety of preset styles for creative portraits, enabling users to carry out image transformations and create themed portraits with minimal settings.

Portrait generation involves facial features and personal information; before uploading photos of oneself or others, it is necessary to ensure that proper authorization has been obtained, and sensitive material should be handled with care.

Work gallery and identical creations

The official website’s gallery displays works in various styles along with relevant keywords; users can examine the composition, materials, and keyword combinations used in outstanding examples.

• Check out the works on the platform to gain inspiration.
• Refer to the format of the public prompt.
• Reuse the creative approach through the same entry point.
• Compare the effects of Chinese descriptions and English-style words.

Assistance with Chinese prompts

The official SDK includes a module for translating prompts from Chinese to English, which helps reduce the effort required for Chinese users to understand the terminology of the English-based model.

Translation is merely a tool to assist in creation; the final outcome will still be influenced by the model, randomness, parameters, and versions of third-party models.

APIs, SDKs, and open-source status

MoorThread has made the MobiMaliangSDK available on GitHub, providing a reference for local deployment for users who have MoorThread’s hardware and development environment.

Generation tasks supported by the SDK

• Stable Diffusion for text-to-image and image-to-image generation
• ControlNet control based on edges, pose, and lines
• Load LoRA and Textual Inversion models
• Deriving image labels using DeepDanbooru
• Use Real-ESRGAN for image upscaling.
• Translate Chinese prompts into English prompts.

Deployment environment requirements

The deployment process for the official repository is designed for Ubuntu, Moulard’s MUSA drivers, container-based execution environments, and compatible hardware; it is not a one-click installation solution for ordinary computers.

Developers also need to download models such as Stable Diffusion, ControlNet, and LoRA on their own, and comply with the licensing terms of each model and resource.

Explanation of the open-source scope

The MobiMaliangSDK is licensed under the BSD-3-Clause license, but this does not mean that the online platforms, cloud-based models, training data, and generated content are all part of the open-source category.

When planning for secondary development or commercial use, it is necessary to verify the licensing terms for the SDK, the base models, the extended models, and the input materials separately.

How does Ma Liang use his writing brush?

Regular users can first learn about the structure of prompt phrases by reviewing the content on the official website, and then use the prompts displayed on the current page to access the available AI creation tools.

1. Open the Mo Bi Ma Liang page and check the available entry points at the moment.
2. Log in to your account or follow the prompts to access the WeChat mini-program.
3. Select available modes such as text-to-image and creative portraits.
4. Specify the subject, scene, style, lighting, and frame requirements.
5. Generate images and check details of characters, text, and structure.
6. Adjust the prompt, then regenerate or save the desired artwork.

How to write prompts more effectively

It is recommended to first write down the main subject and actions, and then add details such as scene, composition, camera angle, color, lighting, materials, and style, in order to avoid having conflicting requirements all at once.

• Subject: people, animals, products, or buildings
• Scene: indoor, street, natural, or fantasy space
• Composition: close-up, half-body, full shot, or top-down view
• Style: photography, illustration, traditional Chinese style, or 3D rendering
• Quality: details, clarity, and direction of light and shadow

What use cases are suitable?

• Content creators: Create images for articles and visual drafts for social media.
• Designers: Explore poster, illustration, character, and brand visual directions
• E-commerce operation: Quickly conceiving product scenarios and marketing materials
• Students and teachers: Create course presentation slides and materials for creative exercises
• Developer: Testing the generative image workflow in the Mole Thread environment

Product advantages

• For Chinese users, understanding prompts and accessing creation tools is much easier.
• The official website’s gallery offers a large number of artworks and suggestions for keywords.
• Integration of online platforms with the Moore Thread GPU ecosystem
• The official SDK is made available to the public, facilitating technical users in understanding the implementation process.
• The SDK covers control generation, model extension, super-resolution, and word inversion.

Usage restrictions and precautions

• AI-generated images may exhibit errors in hands, text, perspective, or object structure.
• The entry points, quotas, and functions of the official website and mini-programs may change depending on product updates.
• The public page does not display fixed membership packages, and it is not possible to estimate the prices based on previous information.
• Authorization must be obtained before uploading a portrait, and attention should be paid to the protection of personal information.
• Commercial works require verification of models, materials, fonts, and portrait rights.
• The local SDK has specific requirements regarding Moore Thread’s hardware and the MUSA environment.

Prices and packages

As of this verification, Mobi Ma Liang’s official website does not disclose any fixed membership packages, prices for points, or costs associated with API calls; the actual fees shall be those indicated on the page after logging in.

Ordinary creators mainly experience these tools through online pages or WeChat mini-programs; whereas developers who deploy them locally have to bear the costs associated with compatible hardware, model downloads, and maintenance.

Supported platforms

• Web page version
• WeChat Mini Program
• Moor Thread Developer Online Services
• Official SDKs for Ubuntu and MUSA environments

Basic information

Recommendation score

Recommendation score: 4.2/5.

Mo Bi Ma Liang is suitable for users who wish to try AI-generated art in Chinese as quickly as possible; the artwork gallery and the tools for creating such art can help beginners develop ideas for their prompts.

Technical users can also explore the official SDK, but the barriers to local deployment are significantly higher than those for online creation; it is necessary to verify the hardware, licenses, and the status of the relevant services before using it.

Frequently Asked Questions

What is Mo Bi Ma Liang?

Mo Bi Ma Liang is an AI-based image creation platform launched by Moore Thread, offering Chinese users tools for generating images based on prompts, as well as examples and interfaces for creating content.

Can Mo Bi Ma Liang be used for free?

The official website does not disclose any fixed membership packages or prices for points; whether the service is free and what the available quota is depend on the information displayed on the website after logging in or on the WeChat mini-program page.

Does Mo Bi Ma Liang support Chinese prompt inputs?

The official SDK offers the capability to translate prompts between Chinese and English; online products are also designed for use in Chinese-language creation tasks. However, for complex descriptions, it is still advisable to specify the subject, style, and visual requirements clearly.

Can Mo Bi Ma Liang be used on a computer?

The official website allows users to view the works and access the AI creation page; however, during verification, a message is displayed suggesting that one should use the WeChat mini-program to experience the mobile version of the service. The actual entry point shall be determined by the page currently being viewed.

Is Mobi Ma Liang an open-source tool?

The online platform is not a fully open-source product; however, Moore Threads has made the MobiMaliangSDK available, and its repository is licensed under the BSD-3-Clause license.

What tasks can the official SDK perform?

The official SDK supports tasks such as text-to-image generation, image-to-image generation, ControlNet, LoRA, super-resolution, inference of prompts from images, and translation of Chinese prompts.

Are the images generated using Mo Bi Ma Liang reliable?

The generated results may contain spelling errors, structural issues, or inaccuracies in the details; before using them for commercial purposes, it is also necessary to check the licensing conditions regarding portraits, trademarks, training data, and third-party models.