Copy-ready prompt
Generate a full-color Chinese comic strip, the main character is this person.{argument name="character" default="Boyuan Chen, OpenAI Researcher"} (First image) He is working on improving the text rendering capabilities of the ChatGPT Image 2 model in the upcoming release. (The background shows bubble tea, and a banana is stuck to the wall with a single strip of tape). He is trying to generate text about his hometown on his computer screen.{argument name="location" default="Wuxi"} When creating a hand-drawn style multilingual infographic poster, the model demonstrated its ability to render extremely small Chinese characters. His efforts paid off, and the team was amazed by the model's excellent multilingual text performance, seeing the variety of languages it could write. While resting with his phone in one hand, he received a translated text message from Sam Altman (see the second image for his profile picture). Sam asked him to check the rendered multilingual text in a newly generated image congratulating the team, since Sam only understood English. However, let's add a touch of humor: at the end of the comic, Boyuan (in typical comic style) is driven crazy because he sees the phrase "稳稳地接住你" (wěn wěn dì jiē zhù nǐ) prominently displayed in the center of the image generated by Sam, while the rest of the image is rendered perfectly. This is because the phrase has become a meme on the Chinese internet, considered an unnatural but funny Chinese phrase that GPT likes to use. Boyuan should angrily shout: "Oh my god! It's learned to 'catch' again!" (Next to him are the shrunken profile pictures of his teammates, sweating and saying: "We're working on fixing it!"). At the very bottom of the comic, add a tiny Chinese footnote: "Note: The entire comic, including this footnote and the picture-in-picture, was generated in one go by GPT Image 2, without editing or multi-step processing." Additional note: Uses a 1440x2560 vertical image layout. The first line shows the researcher working diligently; the second line shows him {argument name="location" default="Wuxi"} The first line should showcase the team's multilingual achievements on the project; the second line should display the team's excitement; the third line should be divided into two parts, with the left side showing messages received on his phone during a break and the right side showing Sam's text message content; the fifth line should show Sam's photo and Chen Boyuan's reaction. No narration should be used except for the first line. Avoid showing a map of China. All characters should be in a cartoon style. The banana background should only appear in the first panel, and the tape should be a single strip, not a cross. The banana and tape decorations should be small and inconspicuous Easter eggs. The OpenAI logo should only appear on Chen Boyuan's clothing and not elsewhere. Mugs should not appear in the scene, as there is already bubble tea. Sam should only appear in the text message panel. The entire cartoon should be presented as a professional photograph of a physical comic book page. Add a line of small text in the bottom right corner of the poster that reads "Even the smallest Chinese characters are clearly readable:", along with an even smaller Chinese paragraph that begins with "(This is a test of the smallest font) Wuxi is the author's hometown, so I made this poster, and the Chinese characters are finally fixed. I haven't been home for many years, and I really want to eat hairy crabs!" (in the smallest font).
Prompt breakdown
Generate a full-color Chinese comic strip, the main character is this person.{argument name="character" default="Boyuan Chen, OpenAI Researcher"} (First image) He is working on improving the text rendering capabilities of the ChatGPT Image 2 model in the upcoming release.
(The background shows bubble tea, and a banana is stuck to the wall with a single strip of tape).
He is trying to generate text about his hometown on his computer screen.{argument name="location" default="Wuxi"} When creating a hand-drawn style multilingual infographic poster, the model demonstrated its ability to render extremely small Chinese characters.
His efforts paid off, and the team was amazed by the model's excellent multilingual text performance, seeing the variety of languages it could write.











