This guide will teach you how to write effective prompts by incorporating cinematographic elements such as camera movements, lighting, atmosphere, and visual style—concepts used by professional videographers and filmmakers. It will also cover prompt structure, tips for refining prompts, and community insights to help you create visually compelling AI-generated videos.
Creating high-quality AI-generated videos requires more than just a basic prompt. The level of detail in your descriptions significantly impacts the final output. Whether you're crafting cinematic sequences, stylized animations, or realistic video clips, understanding how to structure your prompts is key to getting the most out of the model.
Typically, video generation prompts can be made most effective by incorporating cinematographic elements such as camera movements, lighting, atmosphere, composition, and visual style—concepts used by professional videographers and filmmakers. However, every model has its own quirks in how it reacts to prompt structures and elements, so working with Wan 2.1 requires trial and error.
This guide focuses on camera movements in Wan 2.1—how well they work, what needs tweaking, and where the model struggles. We will update this post over time as we continue testing and refining prompts. Be sure to check back often!
A well-structured prompt should be:
Key elements that typically help in creating effective video prompts include:
Describing how the camera moves is crucial in creating dynamic, engaging videos. Examples include:
Lighting can drastically change the feel of a scene:
Setting the right atmosphere ensures that the generated video conveys the desired tone:
Describing how the subject is framed in the shot helps create engaging visuals:
Defining the look and feel of the video is essential for guiding Wan 2.1:
After extensive experimentation, we found that Wan 2.1 does not always respect every camera movement. Some work well, while others are ignored or cause the scene to become static. Below, we detail our findings. For this experiment, we used the 14B text-to-video model and everything was produced in 480p to allow faster iterations.
Example prompt:
"A low angle shot of a jazz pianist in a dimly lit 1920s jazz bar, playing the piano with concentration. He wears a white shirt with suspenders and black trousers, his hands move rapidly on the keys. Camera pans left to low angle shot of a cute girl with pigtails and glasses playing the trumpet."
Example prompt:
"Close up shot of the determined face of a battle-worn samurai. Camera pulls back to reveal him standing alone on a foggy battlefield, gripping his katana. Camera pulls back to reveal fallen warriors behind him. Wind whips through the trees, sending red autumn leaves swirling."
Here, we used a prompt used by the Wan 2.1 team and then attempted to add camera movement.
Starting prompt:
"In the style of an American drama promotional poster, Walter White sits in a metal folding chair wearing a yellow protective suit, with the words "Breaking Bad" written in sans-serif English above him, surrounded by piles of dollar bills and blue plastic storage boxes. He wears glasses, staring forward, dressed in a yellow jumpsuit, with his hands resting on his knees, exuding a calm and confident demeanor. The background shows an abandoned, dim factory with light filtering through the windows. There’s a noticeable grainy texture. A medium shot with a straight-on close-up of the character."
Good prompt with dolly in:
"In the style of an American drama promotional poster, Walter White sits in a metal folding chair wearing a yellow protective suit, with the words "Breaking Bad" written in sans-serif English above him, surrounded by piles of dollar bills and blue plastic storage boxes. He wears glasses, staring forward, dressed in a yellow jumpsuit, with his hands resting on his knees, exuding a calm and confident demeanor. Camera hitchcock zooms in. The background shows an abandoned, dim factory with light filtering through the windows. There's a noticeable grainy texture. A medium shot with a straight-on close-up of the character."
Attempting to create a dolly out effect produced really good dolly ins instead:
Getting a tilt was similarly difficult. We envisioned a scene with a mountaineer who's looking at a formidable mountain ahead of him. The scene was to start off focused on the man and have the camera tilt up to reveal the mountain in front. This may work with more experimentation, but we found that we had to focus the start shot on the feet and then move up. The resulting effect is closer to a boom shot rather than tilt up, but it does achieve a similar look-and-feel.
Example prompt:
"A close up shot of the feet of a man wearing mountaineering gear, standing in a grassy field. He is facing away from the camera. Camera slowly tilts up, revealing the full body of a mountaineer wearing gear. He faces away from the camera. In the distance in front of him majestic rocky mountains tower above."
Example prompt:
"A sprawling cyberpunk metropolis, neon lights reflecting off rain-soaked streets. Pedestrians in futuristic outfits rush by as holographic advertisements flicker in the air. The camera follows a hooded figure in a long tracking shot, weaving through the crowded market. Overhead lights cast a moody glow, while fog drifts through the alleyways. The scene is dark and mysterious, with blue and purple lighting creating a high-tech, dystopian feel."
Example prompt:
"In a large dimly lit midcentury modern room, a man sits with an authoritative and pensive pose on a leather chair. He is wearing a dark suit jacket and grey trousers. He has silver hair. The chair is in the center of the screen. Behind the chair, there is an oak console with a lamp. The wall is made of oak panels. The man looks directly at the camera. Camera rapidly zooms in on the man's face. Then he lets out a slight smirk."
Example prompt:
"Overhead shot of a man asleep on his desk in front of his computer. The room is dark except for the monitor’s glow. The camera rolls in a full 360 motion."
While Wan 2.1 can generate some camera movements effectively, others are inconsistent or non-functional. We recommend structuring prompts carefully, avoiding overly fast camera actions, and iterating based on results. We will continue updating this guide as we refine our findings. Stay tuned!
Generating videos using Wan 2.1 is fun and highly productive but also very resource expensive. If you don't have the right hardware, you can be waiting a long time to generate a clip that is a few seconds long. InstaSD let's you run ComfyUI workflows on powerful GPUs at a very affordable price. All of the following Wan 2.1 workflows are ready to launch on GPUs as powerful as H100s with just one click:
Want to jump right into generating incredible images and videos? We've prepared this and many more workflows to get you started.
Launch This Workflow
