Have you ever had one ...
Have you ever had one of those moments – your mind is overflowing with images, you're desperate to draw them out, but the moment your hand touches the pen, what comes out is so embarrassing you can't even bear to look at it?
I have. And more often than I'd like.
A while ago, a scene suddenly popped into my head – a Shiba Inu in a suit sitting on the moon drinking coffee, with a deep starry sky in the background. A mix of cyberpunk and British gentleman style – I thought it was pretty cool myself. Then I opened my drawing software, fiddled for ages, and what I produced looked like a glowing potato. Right then, I thought: am I just destined to be cut off from visual expression for life?
Later, a friend sent me a link, saying, "Try this – you can paint your dream with just one sentence."
I clicked on it – Dream Diary – the name sounded artsy, like some kind of journaling app. But it turned out to be an AI painting platform, jointly produced by Westlake University's Deep Learning Lab and Westlake Xinchen. To be honest, I signed up with the attitude of "can a tool from academia really be any good?"
Then I got schooled. H...
Then I got schooled. Hard.
I typed in "a Shiba Inu in a suit drinking coffee on the moon," and hit generate. In less than a second, an image appeared.
At that moment, I stared at the screen for several seconds.
The Shiba Inu's deadpan expression, the folds of the suit, the craters on the moon's surface, even the steam rising from the coffee cup – it was about 80–90% of what I had pictured. A person who can turn a stick figure into a fracture patient, with just one sentence and under a second, had produced an image good enough to be an avatar.
Do you know what that means? It means the wall between "I imagine" and "I present" has been torn down.
Of course, there were ...
Of course, there were failures too. For example, once I tried to generate a scene of "an ancient‑style swordsman practicing swordplay in a bamboo forest," and the AI gave me an Ultraman in Hanfu holding a glowing stick. The dissonance was like hearing heavy metal in a tea house. Later I figured out that the more specific the description, the better the result – don't write "a beautiful woman," write "an anime‑style girl with double ponytails, wearing a white dress, standing under a cherry blossom tree." In short, it's about learning to "speak human" to the AI, and the more detail you give, the better it understands you.
Another thing that pleasantly surprised me – it supports over 20 painting styles and over 10 artist styles. Anime, paper‑cut, oil painting, watercolour, CG rendering – you name it. You can also imitate the styles of masters like Xu Beihong, Monet, Feng Zikai, and Picasso. Want Van Gogh to paint a cat for you? No problem. Want Xu Beihong to paint a horse? Also possible. It feels like having dozens of different master artists living in your phone, available at your beck and call.
It also has a ControlNet canvas that supports advanced features like pose recognition and line‑art colouring. You sketch a rough draft, and the AI colours and refines it for you. You want the character in the image to strike a "heart" pose? Just type it in. The path from "clueless amateur" to "semi‑professional creator" is very clear.
And here's something I particularly care about – the copyright of generated images belongs to you. Haven't we seen enough cases of people being sued for using images grabbed from search engines? Having a tool you can trust with confidence is better than anything else.
Now, you might be asking: does it cost money?
New users get 100...
New users get 100 free "Dream Stars," and daily check‑ins give you more for free. Each generation costs 2 stars for free users. I'm a light user, and I haven't spent a single cent so far. If you need more, you can subscribe to VIP – monthly for ¥50, quarterly for ¥128, or yearly for ¥388. One cup of milk tea is enough to keep you going for a good while.
Finally, let me give you some honest advice, purely as a friend –
If you're not a professional illustrator but often need images for social media, covers, inspiration, or simply want to visualise your imagination, don't hesitate – just go try it. It supports web, WeChat mini‑program, and mobile app – you can use it on both computer and phone. My own habit is to generate high‑res images on my computer and play around with smaller ideas on my phone – seamless switching.
The only thing I need to warn you about – don't get hooked late at night like I did, and end up watching the sunrise. I did that last week – I just wanted to "try a couple of images and go to bed," and before I knew it, the birds outside were chirping. I went to work with two big dark circles under my eyes, and when my colleague asked if I had been working overtime, I was too embarrassed to say I had been "painting all night" – and what's worse, it wasn't even me who painted them.
After all, imagination shouldn't be imprisoned by drawing skills. If you can't draw, it's not your brain's fault – it's just that you haven't found the right tool yet. You handle the big ideas, and leave the rest to the AI.
Go search for "Dream D...
Go search for "Dream Diary," give it a try. And then you'll discover that the "magic brush" is now within everyone's reach.