Gen AI Workflow: How to Actually Maintain Creative Control

In this video, I am going to be demonstrating a workflow that is an alternative to a lot of the AI videos you see online that make big claims, but in reality are just simple text-prompt-to-video generation that anyone with a keyboard can create. I will show you a workflow that will give you creative control over the AI to create all the control images you will need to guide AI video generation. We will start with creating a character, training a LoRA, building an environment, and creating a layout. In the end, we will have final start and end frames, as well as a frame-by-frame image sequence of canny and depth control images for our shot. For this workflow, I am using a lot of local AI image diffusion models, which means I am loading the AI models directly into the RAM of my video card and using that for AI image diffusion. While I do have a pretty heavy local setup, you can do everything I am doing in this workflow in the cloud by using Comfy Cloud or renting a GPU online. So there is nothing stopping you from mimicking this workflow for your own creative project. To see more about this workflow and other AI workflows you can go to my portfolio site: https://garrettfry.art/generative-ai