AI Short Drama Production Pipeline: Script → Blender Gray Model Directing → Video Generation → Editing
Creating a few-second clip with generative AI is easy, but making a short drama of tens of seconds to several minutes is difficult. This article introduces a pipeline where directing is decided in Blender and image creation is left to generative models.
While it has become easy to create "plausible few-second clips" using generative AI, it is difficult to create something that can be watched through as a short drama of tens of seconds to several minutes. Specific challenges include character positions changing every cut, camera directions not connecting, and the length being unpredictable.
This article introduces a pipeline currently being tested: "Decide the directing in Blender, and leave the image creation to the generative model." The overall process consists of [1] Script/Shot list, [2] Directing with Blender gray models (gray bo...), video generation, and editing.
For steps requiring a GPU, such as rendering and video generation, CloudGPU instances are rented by the minute.
This item is a summary. The full, checkable version is at the original source:
Read the original at Zenn (ComfyUI) ↗