ChinaZ.com, May 23: It's common to come across such posts on the homepage of Xiaohongshu: a simple hand-drawn illustration of a character or animal accompanied by text that resonates with viewers. These posts generally perform well, with numerous likes and collections.
Today, let's analyze how to create such accounts in detail.
Currently, most content of this type is created using AI for both illustrations and captions, characterized by its simplicity and ease of establishing an account.
The production process consists of three steps:
1. Find benchmarking accounts
For example, take the account "A Little Roll." Despite having fewer than 10,000 followers, the data is consistently stable, with numerous likes and collections.
The post with the best performance is "From that day on, I never initiated contact with you again," which garnered 14,000 likes and 10,000 collections.
After finding such benchmarking accounts, we can learn from their layout and popular tags.
By analyzing this account, we can see that the works are mostly simple, typically with a white background, a cartoon character at the bottom, and a large blank area above for text, using some personalized fonts.
Once these elements are determined, we can replicate the model to create similar accounts. The next step is to address the issue of captions.
2. Find captions
For such accounts, in addition to an eye-catching cover image, the most crucial aspect is the caption. Creating captions that strongly resonate with users is key.
Method one: Search for keywords on mainstream short video platforms
For instance, you can search for healing captions, clarity in life, work addiction, etc., on platforms like Douyin and Kuaishou to find many works sharing classic captions. Choose works with many likes and comments and copy the captions.
Method two: Let AI write for you
If you don't want to use the same captions as others, you can feed some of your favorite captions to AI and have it write similar captions in the same style.
3. Create images
After determining the captions, you can create relevant images based on them. There are many AI drawing tools available; choose one you're comfortable with. I'll use Midjourney as an example here.
In the first step, we've determined the painting style and main composition of such works. Then, we can create a prompt template based on these, and just replace the main part according to the captions. Referencing similar works, we find that the images are mostly white with a doodle style, and the characters or animals are quite cartoonish and cute. These elements will become fixed in our prompts, and you only need to modify the main subject.
For example, my image is a little girl holding flowers, so my prompt is as follows:
by Cy Twombly, doodle, Cartoon illustration, modern hand-drawn illustration, Q-version character, a chubby and cute little girl with short black hair, squatting on the ground, On the side, she is wearing a black t-shirt and blue jeans. The girl is holding a red flower in her hand, focusing on the flower. The pure white background is clean, with a large area of white space above her head. The character only occupies half of the picture. The cute anime, soft lighting, focuses on the girl's face --ar3:4--niji6
Roughly translated:
Created by Cy Twombly, doodle, cartoon illustration, modern hand-drawn illustration, Q-version character, a chubby and cute little girl with short black hair, squatting on the ground, wearing a black t-shirt and blue jeans on the side. The girl holds a red flower in her hand, focusing on it. The pure white background is clean, with a large white space above her head. The character occupies only half of the picture. Cute anime, soft lighting, focusing on the girl's face
Note: After fixing a style, you can continuously reuse it. For example, for the above prompt, if I want to draw a boy next time, I only need to change the main subject to a boy, with changes in clothing and scene, while the rest remains fixed. This ensures that your Xiaohongshu page has works in the same style. The prompts for the author's style, painting style, background, etc., can remain unchanged.
After generating images with MJ, the general effect is as shown above. I think the first one is good, so I choose to enlarge it, with the effect as follows:
Since we need to add text above, the blank space at the top of the image is not enough. Here, you can use PS or an image expansion tool to leave space at the top for text.
To add text, you can use PS, Meitu, or Xingtu, depending on personal preference. To highlight the hand-drawn effect, you can download some cute hand-drawn fonts from free commercial font libraries. Here, I use the free commercial font "PF频凡胡涂体," which matches the hand-drawn style of my image.
After completing the creation, you just need to select the most resonant caption as the title among the multiple images, and you can refer to similar popular works for the tags.
