Midjourney + Runway Gen 3 Turbo Full Course (Create Lifelike Cinematic AI Videos)
TLDRThe era of AI films is dawning with Runway Gen 3 Turbo, enabling filmmakers to ideate at the pace of their imagination. This video offers a deep dive into creating a mini AI film, showcasing the process from storyboarding to generating ultra-realistic cinematic shots with consistent style and character using Midjourney. It guides viewers through creating cinematic videos with Runway's image-to-video mode, controlling camera movements, and comparing output quality between Runway Genry Turbo and the base model, ultimately providing practical tips for leveraging AI in filmmaking.
Takeaways
- 😀 The era of AI films has begun, with Runway Gen 3 Turbo making film-making more accessible and faster than ever.
- 🎥 Runway Gen 3 Turbo is pushing towards super photorealism and excellent coherence, improving upon Gen 2.
- 🚀 The video creator published a mini AI film to test the limits of current AI tools and will share practical ideas and prompting tips.
- 📒 The process involves three main parts: storyboarding, transforming storyboards into cinematic shots with Midjourney, and creating cinematic videos with Runway's image-to-video mode.
- 🧭 Prompting tips are based on the official Runway Genry prompting guide and are crucial for achieving consistency in character and style.
- 🖼️ Storyboarding starts with a synopsis, character details, and setting, which helps in generating scenes with DALL-E and maintaining consistency.
- 🖌️ Midjourney is used to create ultra-realistic cinematic images from the storyboard, focusing on style exploration and image prompting.
- 🎬 Runway's image-to-video mode is guided by camera movement, making the most of its powerful engine to create cinematic sequences.
- 📈 A comparison of Runway Genry Turbo with the base model shows improved speed without a significant drop in quality, especially in body coherence.
- 💡 The video emphasizes the importance of using character and style references effectively in Midjourney to maintain consistency across scenes.
- 🔧 Runway's lip-sync feature and fixed seed option provide additional creative control, allowing for synchronized speech and stylistic consistency.
Q & A
What era does the video suggest has begun with the advent of AI films?
-The era of AI films has started, which is referred to as cinematic Singularity.
What is the significance of Runway Gen 3 Turbo in the context of film making?
-Runway Gen 3 Turbo signifies the democratization of film making, allowing for ideation at the pace of one's imagination and moving closer to super photorealism with excellent coherence.
What was the purpose of the mini AI film published by the speaker?
-The mini AI film was published to push the boundaries and see what's possible with AI tools today.
What are the three main parts of the process covered in the video?
-The video covers three main parts: storyboarding for AI film, transforming storyboard into cinematic shots with consistent style and character using Midjourney, and creating cinematic videos using Runway's image to video mode.
How does the speaker use CET GPT or large language models in the ideation phase?
-In the ideation phase, the speaker uses CET GPT or large language models to generate scenes based on a brief synopsis draft.
Why is specifying the setting, location, and time important when using DALL-E for storyboarding?
-Specifying the setting, location, and time is important to ensure that the images produced are consistent in time periods and do not result in inconsistent outputs.
What is the purpose of creating visual descriptions of each storyboard image?
-The visual descriptions are used to guide the image generation process on Midjourney, helping to create ultra-realistic cinematic images for the AI video.
How does the speaker use the atomic prompting method on Midjourney?
-The speaker uses the atomic prompting method by crafting detailed prompts and attaching style references along with the storyboard image to establish the right tone for the images.
What is the significance of image weight when converting images to video on Midjourney?
-Image weight is significant as it balances the influence of the image prompt with the text prompt, equally affecting the final image. A value of 0.5 is recommended for a balanced approach.
How does the speaker ensure character consistency across scenes?
-The speaker ensures character consistency by using character reference weight and including previously generated images as character references, allowing for more flexibility in changing hairstyles and outfits while preserving facial features.
What is the role of shot types in the prompts for image to video process on Runway?
-Shot types in the prompts enhance creative control by providing a well-defined first or last frame, which is beneficial for the image to video process.
How does the speaker address the issue of slow motion in Runway outputs?
-The speaker suggests using keywords like 'normal speed' or 'high speed' to accelerate the footage, as there is currently no speed control setting in Runway.
What is the new pricing model for Runway Gen 3 Turbo?
-The new model is priced at 5 credits per second of video, which is half the cost of the original Gen 3 Alpha that was priced at 10 credits per second.
Outlines
🎬 The Dawn of AI Cinematography
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.
📝 Storyboarding and Image Prompting
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.
🖼️ Crafting Cinematic Shots with Character Consistency
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.
🌟 Creating a Cinematic Universe with Batch Prompting
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.
🚀 Upscaling and Camera Movement in Runway
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.
🎥 Fine-Tuning Prompts and Solving Common Issues
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.
Mindmap
Keywords
💡AI films
💡Runway Gen 3 Turbo
💡Midjourney
💡Storyboarding
💡Cinematic coherence
💡Image to video mode
💡Prompting tips
💡Character reference weight
💡Style reference
💡Upscaling
💡Lip sync feature
Highlights
The era of AI films has begun, with Runway Gen 3 Turbo leading the democratization of film making.
Runway Gen 3 Turbo allows ideation at the pace of imagination, lighting fast.
The video demonstrates progress towards super photorealism and excellent coherence in AI-generated films.
A mini AI film 'A Blast from the Past' was created to test the limits of current AI tools.
Practical ideas and prompting tips from the official Runway Genry prompting guide will be shared.
The video is divided into three parts: storyboarding, creating cinematic shots with Midjourney, and generating videos with Runway's image-to-video mode.
A rough draft of a synopsis is essential for the storyboarding process.
Character 'Ilas' and the setting of the US Wild West era in a snowy Rocky Mountains region were chosen for the film.
CET GPT cloud or other large language models can generate scenes based on a synopsis draft.
16 scenes were written as a brief for storyboarding with Midjourney.
CAT GPT is used for storyboarding due to its character and style consistency.
Visual descriptions of each storyboard image are crucial for the next steps in the process.
The atomic prompting method on Midjourney is used to create ultra-realistic cinematic images.
Style exploration is necessary to establish a consistent tone throughout the series of images.
Image prompting and style reference are key in transforming low-fidelity images to high-fidelity cinematic shots.
Character reference weight in Midjourney is crucial for directing scenes with character consistency.
Inclusion of shot types in prompts benefits the image-to-video process by providing a well-defined first or last frame.
Batch prompting on Midjourney allows for the creation of multiple image generations with a single prompt.
Upscaling images with MID Journey UPS scalers or other tools like Magnific is part of the workflow.
Camera movement in Runway is directed based on the first frame, which can be set as either the starting point or the destination of the video.
Runway Genry Turbo is praised for its speed and satisfactory output quality, especially in body coherence.
Prompt presets in Runway assist beginners in controlling camera movements more comfortably.
For image-to-video, focus on camera movement, character, and scene movement rather than describing the image itself.
Common mistakes include describing the image for image-to-video, which is unnecessary.
Using 'camera follows X' format simplifies prompting for image-to-video in Runway.
Runway's lip sync feature allows for the addition of dialogue and hand gestures to generated characters.
Runway's pricing has been updated with Genry Alpha Turbo, now costing five credits per second of video.
A low-resolution preview before full-resolution generation is recommended for better cost control.
Fixed seed on Runway creates visually similar generations, acting as a style reference.