WanXiangYingZao

Overview

Wanxiang Creation (万相营造) is an AI-native creative production environment from Taobao AGI, developed for constructing and managing multimodal digital content. The name reflects the product’s core purpose: helping users build “ten thousand forms” of visual and conceptual assets through a structured, AI-assisted process. It is positioned as an end-to-end workspace that connects ideation, generation, review, and delivery, so that both professional teams and individual creators can move from natural-language briefs to near-production-ready outputs without switching between fragmented tools.

At the core of Wanxiang Creation is a multimodal generation and processing pipeline that leverages large language models and generative foundation models to interpret user instructions, generate coherent scenes or asset sets, and support iterative refinement. Users can express an idea as a written prompt, adjust stylistic parameters, and then modify specific elements through targeted feedback and editing instructions. The system is designed with human-in-the-loop principles, meaning that generated artifacts are treated as drafts to be reviewed, compared, and improved rather than as final black-box results. Version history and side-by-side comparison help teams track creative decisions and avoid losing promising directions during exploration.

Beyond single-asset generation, Wanxiang Creation aims to support production-scale workflows. This includes organizing generation tasks into reusable pipelines, applying consistent style or brand constraints across a series of outputs, and enabling batch generation for A/B testing and variant exploration. Collaboration features such as review, annotation, and approval allow stakeholders to participate at the right moment in the creative process, while final assets can be exported in standard formats for downstream integration into e-commerce pages, campaign sites, presentations, video production, and other publishing channels.

The practical value of Wanxiang Creation is strongest in scenarios where creative iteration speed, visual variety, and production efficiency matter. E-commerce marketers can quickly explore campaign concepts and product display ideas; brand teams can maintain visual consistency across multiple content pieces; and creative agencies can prototype directions for clients before committing to expensive production. By abstracting complex generation capabilities into a manageable environment, Wanxiang Creation makes advanced AI-assisted creation more accessible to non-specialist roles without removing the controls that experienced artists and designers need.

Core capabilities

Natural-Language Brief Interpretation

Turn written descriptions into structured initial drafts, using large language models to understand scenes, subjects, and style requirements.

Scene and Composition Control

Arrange subjects, backgrounds, and visual elements using prompt-level and parameter-level controls for composition, lighting, and style.

Iterative Editing and Refinement

Modify generated assets through targeted instructions, allowing fine-grained changes while preserving the intended composition.

Style and Brand Consistency

Maintain reusable style presets or reference-based constraints to keep outputs aligned with a defined visual identity.

Batch and Variant Exploration

Generate multiple variants of a concept in parallel, supporting faster creative discovery and A/B-style evaluation.

Version History and Comparison

Track iterations, revisit previous versions, and compare outputs side-by-side during the review process.

Reusable Workflow Pipelines

Combine generation, review, and export steps into repeatable workflows that standardize the team’s production process.

Export and Handoff

Deliver final assets in standard formats for use in downstream design, marketing, and development workflows.

Use cases

Campaign concept exploration and mood board creation

E-commerce product page visuals

Multi-variant ad creative generation

Brand-style content production at scale

Internal design prototyping and idea communication

Visual asset localization and adaptation

Training and demonstration materials for AI-driven workflows