What the Production of AI Drama “后西游记” Reveals About Long-Form Visual Creation

China’s first AI long-form drama, “后西游记,” arrived in Hunan Satellite TV’s prime-time schedule on August 31, according to an official recap published by Volcano Engine. The series has a planned length of 60 episodes, with each episode running for approximately 40 minutes. It was produced by Mango TV and made by Bojing Culture through the Mango Lingchuang platform.

The source says the project moved from approval in May to broadcast in only half a year, while the production itself took 3 months. No camera was used for filming the drama, and 100% of its video generation was completed with Seedance. One week after launch, the full episodes on Mango TV had surpassed 1.5 billion views, according to the recap.

The production account also gives a cost and scheduling comparison. Chief director Li Dongshen used one action sequence as an example: the team spent 10 days producing it, at a cost of approximately several hundred thousand yuan. A comparable live-action shoot could potentially require 3 million to 4 million yuan, the source says. This is presented as a production comparison rather than a general cost guarantee for all AI-generated scenes.

For visual creators, the most useful detail is the workflow implied by these figures. The project was organized as a long-form narrative with a defined episode count, broadcast target, platform support, and production window, rather than as a collection of isolated AI clips. That suggests AI video production at this scale depends not only on generation quality, but also on planning, continuity management, scene design, and coordination between creative and technical teams.

AiPix analysis: “后西游记” is a useful case study because the source connects a specific model, Seedance, to an end-to-end drama production process and reports both schedule and cost comparisons. However, the recap does not establish that the same timeline, budget, or audience performance can be reproduced by other teams. Creators evaluating similar projects should treat these details as source-reported project results, then assess their own requirements for consistency, revisions, rights, and delivery before choosing a production workflow.