MiniMax is an AI video generator from Hailuo that appeared out of nowhere and quickly gained traction on social media due to its impressive realism and modeling of real-world motion. It is moving to its own dedicated website as the company ramps up global rollout.
This is just the latest foray into generative AI for the Alibaba and Tencent-backed unicorn startup. Its AI companion app Talkie has been downloaded over 15 million times and, like Character.ai, lets you converse with a virtual creation.
The tool is currently only text-to-video but an image-to-video model is coming soon, as well as version two of the AI tool that will be even more realistic, and allow for longer generations and better motion.
The new dedicated platform also comes with prompt enhancement, English by default, and is laid out in such a way that additional features can be added over time. It is in direct competition with fellow Chinese-based AI video platform Kling, which already has many of those features and more.
How well does Hailuo MiniMax work?
New Feature☀️☀️The Prompt Optimizer refines your prompts to boost visual quality and enhance the video’s overall aesthetic, building on your original input. For precise control, you can disable this feature, and the model will strictly follow your input.Ex: 1 & 2 (Before &… pic.twitter.com/YBW9tpm9rPSeptember 16, 2024
Hailuo MiniMax is already what I’d describe as a tier-one model. It is capable of impressive realism in its output. There are some issues, but most are the same issues that plague any new AI product — slow inference and minimal features. But for its core purpose, making AI video, it is exceptional.
I have spent enough time with it to learn some of its quirks, but unlike Runway, Haiper, Luma Labs Dream Machine, or Kling it has no prompting guide to help you steer generations. What I’ve been doing is looking at the “featured videos” prompts and using that as a guide.
With the new release update, Hailuo has added a “magic prompt” feature that uses a text model to make a simple prompt more descriptive. Unfortunately, unlike similar features from Ideogram and others, it doesn’t show you what the prompt has been changed into.
MiniMax has been compared to OpenAI’s flagship and yet-to-be-released AI video model Sora in terms of motion and visual. I’ve even made the comparison myself, but haven’t tried it and compared its output to Sora preview videos — its close but still has some way to go including in video duration.
Prompts to try on MiniMax
I’ve based these prompts on the fact MiniMax can now enhance them to fit its internal prompting guide — which hasn’t been made public. So each is more of a top-level concept rather than the type of detailed prompt I’d share for Runway or Dream Machine. I have also written more descriptive versions.
One thing to note is that you can only generate one video at a time. Unlike other platforms where you leave it to work on multiple videos at a time, here you have to wait until each finishes.
1. Surreal Sky Kingdom in a Glass Globe
(Image credit: Hailou MiniMax/Future AI)
Descriptive prompt: “Within the confines of a massive glass globe, floating high above a vast, turbulent sea, lies a surreal kingdom in the sky. Jagged cliffs rise from the clouds, dotted with ethereal castles connected by delicate, glowing bridges. The sky is a swirl of colours, a sunset that never ends, bathing the world in soft pinks, purples, and gold. Above, mythical creatures like griffins and phoenixes soar gracefully through the air. The glass globe distorts the view slightly, making the scene feel both dreamlike and fragile, as if the entire kingdom could vanish in a wisp of cloud at any moment. The camera pans gently, capturing the serene movement of the creatures and the shimmering light reflecting off the globe’s surface, creating an otherworldly atmosphere of peace and magic.”
2. Rural Village at Sunrise
(Image credit: Hailou MiniMax/Future AI)
Descriptive prompt: “A quiet rural village awakens as the first light of dawn breaks over the horizon. The soft glow of the sun gently illuminates the rooftops of small stone cottages, casting long shadows across fields of golden wheat. Birds begin to chirp, and a thin mist lingers just above the ground, giving the scene a serene, almost magical quality. Farmers prepare for the day, their silhouettes visible as they walk along dirt paths, tools in hand. The camera moves slowly across the landscape, capturing the peaceful transition from night to day, highlighting the simplicity and beauty of life in the countryside.”
3. City Street Market in the Early Morning
(Image credit: Hailou MiniMax/Future AI)
Descriptive prompt: “In the heart of a bustling city, a street market comes to life in the early hours of the morning. Vendors set up their stalls, displaying fresh produce, colourful fruits, vegetables, and local goods. The camera pans across the market, capturing the sounds of conversations, the rustling of crates, and the clinking of coins. Pedestrians, some hurried and others leisurely, weave between the stalls, inspecting items and chatting with the vendors. The atmosphere is vibrant but calm, with a warm morning sun casting soft light over the scene, while the smell of fresh bread and coffee wafts through the air.”
4. Fishing Boats Returning to Shore
(Image credit: Hailou MiniMax/Future AI)
Descriptive prompt: “As the day draws to a close, a fleet of small fishing boats glides across calm waters, returning to shore after a long day at sea. The sky is painted in shades of orange and pink from the setting sun, casting a soft glow on the water. Fishermen, tired but content, begin unloading their catch at a small dock where locals gather to inspect the day’s haul. Seagulls circle overhead, their cries adding to the coastal ambiance. The camera captures the serene beauty of the boats silhouetted against the horizon, with gentle waves lapping against the shore, evoking a peaceful end to the day’s labor.”
5. Metropolitan Intersection During Rush Hour
(Image credit: Hailou MiniMax/Future)
Descriptive prompt: “A busy intersection in the heart of a major city during rush hour, as streams of people flood crosswalks from all directions. The sound of car horns, footsteps, and distant chatter fills the air, while towering skyscrapers loom above, their windows reflecting the soft evening light. The camera focuses on the movement of the crowd, capturing the variety of people—professionals in suits, cyclists weaving through traffic, and pedestrians with shopping bags—all going about their routines. Buses and taxis navigate the congestion, and digital billboards flicker with advertisements. The scene conveys the relentless energy and rhythm of city life, with a subtle focus on the diversity of individuals within the urban sprawl.”
6. Woman in a Café Talking Animatedly
(Image credit: Hailou MiniMax/Future AI)
Descriptive prompt: “Inside a small, cosy café, a woman in her mid-30s sits at a table, talking animatedly to a friend across from her. Her hands gesture expressively, complementing the excitement in her voice, as she leans forward with a smile. Sunlight streams through the large window behind them, casting a warm glow on the wooden table and the steaming mugs of coffee in front of them. The background is alive with the ambient sounds of clinking cups and quiet conversations, while the camera focuses on the woman’s animated expressions and natural gestures, capturing the warmth of human connection in a familiar setting.”
7. Construction Workers on a Skyscraper
(Image credit: Hailou MiniMax/Future AI)
Descriptive prompt: “High above the city skyline, a group of construction workers carefully move across the steel beams of an unfinished skyscraper. Clad in hard hats and reflective vests, they communicate with each other through hand signals and shouted instructions as they weld beams into place. The camera captures close-ups of their focused faces, the sweat glistening on their brows, juxtaposed against the sprawling city below. In the distance, other skyscrapers loom, framed by a clear blue sky. The sound of welding sparks, hammering, and the faint hum of traffic far below creates a rich, industrial atmosphere, while the scene highlights the precision and danger of the workers’ tasks.”
Final thoughts
I ran several of the prompts with the top level idea, leaving MiniMax to refine the idea and turn it into a detailed prompt. I then ran them again with my own properly descriptive prompt to see which turned out better.
What I discovered is that my own descriptive prompts were often more realistic in their vision and motion than the refined prompt version. This may be something to do with MiniMax being trained in Chinese and me prompting in English.
I found that you are better off creating a custom prompt if you are looking for a specific output. If you want a general concept and don’t have a specific idea of how it should look in mind, then using prompt refinement is fine.
