Gen 3 Alpha is Runway’s newest text-to-video model, producing high-fidelity, controllable video.
In the evolving landscape of video creation, Runway Gen 3 stands out as a revolutionary tool for generating cinematic frames. This advanced AI-powered platform leverages structured prompting and a vast library of keywords to help creators bring their artistic visions to life with unparalleled precision and creativity.
To use Gen 3, head to the text or image-to-video tool and select Gen 3 from the dropdown.
You can create a 5-second or 10-second video. Generation times are about 60 seconds for 5 seconds and 90 seconds for 10 seconds.
In the prompt box, add a prompt like “closeup of a smiling young woman with curly hair on a dusky beach, the ocean blurs in the background bathed in deep blue twilight, creating a serene atmosphere.”
Prompt: “Dynamic motion: tracking a woman galloping on a black horse through a field of red poppies. The camera follows her, gradually narrowing the frame until a close-up of the horse’s head. Camera Movement: Start with a wide angle, then slowly zoom in to a close-up of the horse’s head.”
Understanding the Basics of Gen 3 Alpha
Runway Gen 3 Alpha utilizes a structured prompting system designed to generate high-quality video content consistently. This system breaks down the details of the scene, subject, and camera movement into clear, distinct sections. The basic format of a prompt is:
- [camera movement]: [establishing scene]. [additional details].
For example, to create a scene featuring a woman in a tropical rainforest, you might use a prompt like:
- Low angle static shot: The camera is angled up at a woman wearing all orange as she stands in a tropical rainforest with colorful flora. The dramatic sky is overcast and gray.
This structured approach ensures clarity and focus, making it easier to achieve the desired visual output. It’s essential to keep your prompt focused on what should be in the scene rather than what should not be, as positive descriptions yield better results.
Runway Gen 3 was trained on highly descriptive, temporally dense captions. The more detailed the text prompt, the better Gen 3 performs. The description can include subject, scene, lighting, camera movements, and more. Additionally, Gen 3 can be prompted for subject action, camera action, speed, transitions, and more.
For example, you can prompt, “A high-speed wide FPV shot approaches the rocky seaside cave, enters the cave, and emerges in an Arctic landscape with glaciers and snow-capped mountains hyperlapse.”
Gen 3 works with various prompt structures, from simple to complex. For inspiration, try separating your prompt into a visual description and a camera motion description.
For example, “A pillow fort in a cozy living room, made from quilts, fabrics, and pillows. Handheld camera smoothly zooms into the entrance of the pillow fort, revealing an ancient castle interior.”
Crafting Effective Prompts
To effectively use Gen-3 Alpha, it’s crucial to understand how to construct detailed and specific prompts. Here are a few examples to illustrate different approaches:
Seamless Transitions
Continuous hyperspeed FPV footage: The camera seamlessly flies through a glacial canyon to a dreamy cloudscape.
Detailed Camera Movement
A glowing ocean at night time with bioluminescent creatures underwater. The camera starts with a macro close-up of a glowing jellyfish and then expands to reveal the entire ocean lit up with various glowing colors under a starry sky. Camera Movement: Begin with a macro shot of the jellyfish, then gently pull back and up to showcase the glowing ocean.
Dynamic Text Title Cards
A title screen with dynamic movement. The scene starts at a colorful paint-covered wall. Suddenly, black paint pours on the wall to form the word “Runway”. The dripping paint is detailed and textured, centered, superb cinematic lighting.
These examples highlight how different elements of a scene can be described in detail to create compelling and visually striking videos.
Leveraging Prompt Keywords
Keywords play a crucial role in refining your prompts and achieving specific styles in your output. Ensuring that keywords are cohesive with your overall prompt enhances their effectiveness. Here are some key categories and examples of keywords to experiment with:
Camera Styles
Low angle, high angle, overhead, FPV, hand held, wide angle, close up, macro cinematography, over the shoulder, tracking, establishing wide, 50mm lens, SnorriCam, realistic documentary, camcorder.
Lighting Styles
Diffused lighting, silhouette, lens flare, back lit, side lit, colored gel lighting, Venetian lighting.
Movement Speeds
Dynamic motion, slow motion, hyperspeed, timelapse.
Movement Types
Grows, emerges, explodes, ascends, undulates, warps, transforms, ripples, shatters, unfolds, vortex.
Style and Aesthetic
Moody, cinematic, iridescent, home video VHS, glitchcore.
Text Styles
Bold, graffiti, neon, varsity, embroidery.
By integrating these keywords into your prompts, you can more precisely control the visual and stylistic elements of your video. For instance, if you’re aiming for a dramatic and cinematic look, you might use keywords like “low angle,” “diffused lighting,” and “moody.”
Advanced Techniques for Prompting
Beyond the basics, there are advanced techniques you can employ to get the most out of Gen-3 Alpha. Here are some tips to help you master the art of prompting:
Clarity and Specificity
Be clear and specific about what you want in the scene. Vague descriptions can lead to unpredictable results. For example, instead of saying “a beautiful landscape,” specify “a lush, green valley surrounded by towering mountains at sunset.”
Positive Descriptions
Focus on what should be present in the scene rather than what should not. Positive descriptions tend to yield better results. For instance, say “a bright, sunny day” instead of “not a cloudy day.”
Experimentation
Don’t be afraid to experiment with different structures and keywords. The more you practice, the better you’ll understand how to manipulate the tool to achieve your desired outcomes. Try different combinations of camera angles, lighting styles, and movement types to see what works best.
Reinforcement
Reinforce key ideas by mentioning them in different sections of your prompt. This helps ensure that these elements are emphasized in the final output. For example, if the color red is important, mention it in the description of the subject and the background.
Contextual Cohesiveness
Ensure that all elements of your prompt are cohesive and make sense within the context of the scene. Inconsistent details can lead to confusing results. For example, a prompt that mixes modern and medieval elements might create an incoherent scene.
