Midjourney + Runway Gen 3 Turbo Full Course (Create Lifelike Cinematic AI Videos)

The script introduces the new era of AI in filmmaking with the advent of Runway Gen 3 Turbo, which promises to revolutionize the industry by allowing creators to produce content at the speed of their imagination. The narrator reflects on the progress from the initial use of Gen 2 for AI video production and the strides made towards achieving super photorealism and coherence. They share their recent experiment with a mini AI film to test the capabilities of current AI tools. The video is structured into three parts: storyboarding, transforming storyboards into cinematic shots with consistent style and characters using Mid Journey, and creating cinematic videos with Runway's image-to-video mode, focusing on camera movement and leveraging Runway's powerful engine. The comparison between Gen 3 Turbo and the base model's output quality is also promised.

The second paragraph delves into the process of storyboarding for an AI film. The narrator starts with writing a synopsis and character details, setting the scene in the US Wild West era with a snowy Rocky Mountains backdrop. They utilize large language models like CET GPT Cloud to generate scenes based on the synopsis. The process involves using Dolly for low-fidelity storyboard images, emphasizing consistency in setting, location, and time. The narrator then describes using CAT GPT for its character and style consistency, transforming these storyboards into high-fidelity images using Mid Journey's atomic prompting method. Style references and image prompting are crucial for establishing the tone and quality of the images, with detailed instructions on using image weight and style references to balance the prompts effectively.

In this section, the narrator discusses the transformation of the storyboard's first image from low to high fidelity using image prompting and style references on Mid Journey. They explain the importance of image weight in balancing the influence of image and text prompts, and how to use previously generated images as style references. The paragraph also covers character design, using the storyboard's depiction of a Native American young man to create a more realistic character image. The narrator shares tips on using character weight to maintain consistency in the character's appearance across scenes, adjusting the weight based on whether the character's look needs to change or stay the same. They also demonstrate how to use image prompts effectively to create a series of cinematic shots that maintain character and style consistency.

The fourth paragraph focuses on expanding the cinematic universe by using batch prompting or permutations on Mid Journey. The narrator introduces the concept of using brackets in prompts to generate variations, controlled by the chaos parameter, which influences the diversity of the generated images. They discuss the importance of keeping chaos values low for maintaining prompt adherence while allowing for variety. The narrator also shares a quick tip on using the same image for both style and character references, and the benefits of including shot types in prompts for the image-to-video process. The paragraph concludes with advice on organizing generated images for easy access during the upscaling process, which can be done using Mid Journey's upscalers or other tools like the Magnific upscaler for added realism.

This paragraph discusses the final steps before handing over the generated images to Runway for video creation. The narrator advises liking and organizing images during the Mid Journey generation process for easy filtering and finding later. They mention using the new Mid Journey upscalers or the Magnific upscaler for upscaling images, emphasizing the importance of maximizing resemblance settings. Once in Runway, the narrator describes setting camera movements based on the first frame, which can be used as either the starting point or the destination of the video. They praise the speed of Genry Alpha Turbo and its ability to maintain body coherence and detail accuracy, while also introducing prompt presets for easier camera control and the official prompting guide for text-to-video and image-to-video processes.

The final paragraph provides insights into fine-tuning prompts for image-to-video in Runway, focusing on camera techniques, camera follows X format, and event descriptions. The narrator formalizes a prompt structure based on the official guide, emphasizing the importance of describing camera movements and events rather than the image itself. They offer examples of simple and complex prompt structures, showcasing the flexibility in creating cinematic footage. The paragraph also addresses common problems with Runway, such as slow motion and color changes, suggesting solutions like using speed keywords and maintaining color consistency with specific prompt keywords. The narrator demonstrates Runway's lip sync feature, allowing for the addition of voice and hand gestures to characters. They conclude with thoughts on Runway's pricing, suggesting a low-resolution preview option for better cost control and user satisfaction, and introduce the concept of a fixed seed for creating visually similar generations.