ChatGPT+Midjourney实现儿童绘本故事及其插图（数字1~10的故事）

团队模型、论文、博文、直播合集，点击此处浏览

一、背景

近期AI对话技术与AI绘画技术突飞猛进。其中1.）AI对话技术是指用于模拟人类语言交流的人工智能技术。它可以让机器通过自然语言文本或语音与人类进行交流，并能理解人类的话语内容并做出相应的回应。对话技术在语音助手、智能客服、智能创作等领域有广泛应用。相关技术原理与最新进展详见《ChatGPT原理解析以及使用方法介绍》。2.）AI绘画技术是指用人工智能技术实现的计算机绘图。它可以让计算机根据输入的要求生成图像或绘制图形，并能自动调整画面的各项参数，如线条粗细、颜色、光照等。AI绘画技术在图像生成、图像修改等领域有广泛应用。相关技术原理与最新进展详见《人工智能内容生成元年—AI绘画原理解析》。

最近宝宝生日，想着做点生日礼物给娃，突发奇想每天晚上妈妈给讲的儿童绘本故事是否可以利用现在的AIGC技术去生成呢？带着这个疑问，于是做了Midjourney的充值使用，结合ChatGPT做了实践。本文余下章节将尝试结合AI对话与AI绘画进行儿童绘本内容的创作，并做简单的总结。其中AI对话技术采用ChatGPT，AI绘画技术采用Midjourney软件，这两者产品都来自OpenAI。

二、实战

Q（人为输入）：儿童绘本故事，关于数字1~10的教育故事，每个章节讲一个数字，总共10个英文章节。

A（AI输出）：Here is a sample children's picture book story about the numbers 1 through 10. Each chapter focuses on one number and provides an educational lesson or activity related to that number.

Chapter 1: One

文本（AI输出）：

One little bird was hopping around the forest, looking for something to eat. It found a juicy berry and picked it up with its beak. "Yum, this is delicious!" the bird said. But then it heard a rustling in the bushes and saw another bird. "Hey, that's my berry!" the second bird said. The first bird realized that it had taken something that didn't belong to it, and it quickly gave the berry back. "I'm sorry," the first bird said. "I didn't mean to take your berry. I'll be more careful next time." The second bird forgave the first bird, and they both flew off to find more food.

Lesson: It's important to share and be fair.

插图（AI输出）：

Chapter 2: Two

文本（AI输出）：

Two little rabbits were playing in the meadow. They were having so much fun jumping and chasing each other. But then they heard a loud noise and saw a farmer driving a big tractor. "Oh no, we need to hide!" one of the rabbits said. They quickly ran to a nearby burrow and squeezed inside. It was a tight fit, but they were happy to be safe.

Lesson: Sometimes it's good to work together and help each other.

插图（AI输出）：

Chapter 3: Three

文本（AI输出）：

Three little monkeys were swinging from tree to tree in the jungle. They were having a great time, until one of them slipped and fell into the river. "Help! I can't swim!" the monkey cried. The other two monkeys quickly jumped into the water and grabbed their friend. They pulled him to the shore and made sure he was okay. "Thank you," the monkey said, "I was scared!"

Lesson: It's important to help others in need and be there for your friends.

插图（AI输出）：

Chapter 4: Four

文本（AI输出）：

Four little ducks were swimming in a pond. They were quacking and splashing, having a great time. But then it started to rain, and the pond began to fill up. The ducks swam faster and faster, trying to stay above the water. One of the ducks saw a log floating nearby and climbed on top of it. The other three ducks followed, and they all rode the log to safety.

Lesson: It's important to be prepared and find solutions when faced with a challenge.

插图（AI输出）：

Chapter 5: Five

文本（AI输出）：

Five little fish were swimming in the ocean. They were different colors and sizes, but they were all friends. They swam and played together, exploring the coral reefs and searching for food. But then they heard a loud noise and saw a big net coming their way. The fish quickly swam away, dodging the net and escaping.

Lesson: It's important to be aware of your surroundings and stay safe.

插图（AI输出）：

Chapter 6: Six

文本（AI输出）：

Six little ants were marching along a trail, carrying food back to their nest. They were working hard and moving quickly, but then one of the ants stumbled and dropped its piece of food. "Oh no, now we won't have enough to eat!" the ant said. The other ants quickly stopped and picked up the fallen food. "We can share and make it work," one of them said. So they all continued marching, sharing the food and helping each other along the way.

Lesson: It's important to work together and support each other.

插图（AI输出）：

试了多种输入文本，都得不到相应合理的结果，于是放弃。

Chapter 7: Seven

文本（AI输出）：

Seven little bees were buzzing around a flower garden, collecting nectar and pollen. They were having a great time, until one of them flew too close to a beekeeper's hive and got stuck in a sticky web. "Help, I can't get free!" the bee cried. The other bees quickly flew to its rescue and worked together to free it. "Thank you, my friends," the bee said. "I couldn't have done it without you."

Lesson: It's important to help others and work together as a team.

插图（AI输出）：

未尝试。

Chapter 8: Eight

文本（AI输出）：

Eight little turtles were sunning themselves on a log in the river. They were enjoying the warm sunshine, when suddenly a big wave came and swept them all into the water. The turtles quickly swam to the shore and climbed back onto the log. "Wow, that was a close call!" one of them said. The others nodded in agreement.

Lesson: Sometimes things don't go as planned, but it's important to stay calm and adapt.

插图（AI输出）：

未尝试。

Chapter 9: Nine

文本（AI输出）：

Nine little puppies were playing in a field. They were chasing each other and rolling around in the grass, having a great time. But then they heard a loud noise and saw a big truck coming down the road. The puppies ran and hid behind a fence, watching as the truck passed by. "That was scary!" one of them said. The others agreed, and they all cuddled together for comfort.

Lesson: It's important to be cautious and stay safe in unfamiliar situations.

插图（AI输出）：

未尝试。

Chapter 10: Ten

文本（AI输出）：

Ten little dinosaurs were running and playing in the forest. They were having a great time, until one of them tripped and fell into a pit. "Help, I'm stuck!" the dinosaur cried. The other dinosaurs quickly ran to the edge of the pit and looked down. They saw their friend at the bottom, looking up at them. "We have to help him," one of them said. So they all worked together, using their strength and teamwork to pull the dinosaur out of the pit. "Thank you, my friends," the rescued dinosaur said. "I couldn't have done it without you."

Lesson: It's important to help others and work together as a team.

插图（AI输出）：

未尝试。

三、总结

ChatGPT可以产出超越大部分人类水平的非常惊艳的简短儿童绘画故事剧本。相比之下，目前AI绘画技术还没能基于这些故事剧本上下文生成符合剧情内容的图像。

这里主要有两点挑战：1.）包含数字以及复杂动作上下文的精准内容生成目前还是未解决的问题，效果不如人意，可认为目前图像生成模型还远未达到精准的图像内容语义理解。2.）目前生成图像内容更多的还是基于即时的输入得到即时的输出，还不能做到记住故事或者多段对话上下文，输出贴合故事剧本或者多段对话上下文的连贯性图像内容。