Generate "Non-Model" Looks with ChatGPT Image 2.5?
Normally, when you ask ChatGPT to generate an image, it produces AI-style handsome men and beautiful women.
The prompt is simply "Make a beautiful woman."
The prompt is simply "Make a handsome man."
You probably see these often on your timeline recently.
I posted something like this the other day:
https://x.com/1banana2546/status/2098894423174090788
https://x.com/1banana2546/status/2097421065689710799
Thank you very much for the interest from many people.
By the way,
You can also make things like this.
Caution with Real Locations
If the background is a real location, ChatGPT often draws it based on imagination, so locals will immediately know it's fake, but it still feels realistic.
Unlike the typical AI-generated handsome/beautiful faces, these might feel more relatable?
There is definitely demand for these kinds of images. In some scenarios, these are more useful than highly polished faces.
For example, they could work well as the "Before" in beauty product descriptions.
Depending on how they're used, they could unfortunately be used for scams, which is scary. But just like scissors or knives, the tool itself isn't bad; it depends entirely on the user.
Please use them responsibly.
Using this method, you can generate:
Prompt
Here is the long-awaited prompt.
⇩⇩⇩⇩ Prompt starts here ⇩⇩⇩⇩
Draw one adult Japanese woman.
Do not make her look like a model or celebrity with overly perfect features. Instead, create a natural, approachable, ordinary person you would see on the street or in daily life.
Render it with the texture of a photorealistic smartphone photo, expressed as a casual snapshot taken naturally during daily life. Do not give it the polished atmosphere of an advertising shoot, fashion shoot, or studio shoot.
Do not excessively beautify the face. Keep common facial features, natural asymmetry, and individual differences. Retain healthy, natural skin texture including pores, fine unevenness, slight redness, and color variations. Do not use excessive skin smoothing or retouching.
Hair should have fine flyaways and slight disarray, kept in a natural state rather than perfectly styled. The body type should be natural, whether thin or heavy. Do not make her a "beauty queen."
Clothing should be simple casual wear that an ordinary adult woman would wear in daily life. Dress her in natural everyday clothes appropriate for the current season.
Assume the photo was taken handheld with a standard smartphone. The composition should be slightly off-center, and the camera should be tilted slightly. Do not use perfectly calculated composition; aim for natural framing as if casually photographed.
Use only natural light or indoor light existing at the location. Do not use studio lighting, ring lights, or strong dramatic lighting. Leave natural contrasts and light variations on the face and background.
The background should be a natural place seen in daily life in Japan. Residential areas, rooms, cafes, parks, or station surroundings are suitable. If using a real location, refer to actual images for the latest state and do not draw from imagination.
Avoid excessive background blur, cinematic effects, advertising-style color correction, CG-like textures, or AI illustration styles.
Overall, prioritize making it look like:
"A natural, slightly rough daily photo of an ordinary adult woman casually taken with a smartphone."
n=5
+++++++++++++++++++++++++++++
Sorry if you already know this, but the last line "n=5" means "generate 5 images." If you want 8, change it to "n=8"; if you want 3, change it to "n=3."
It's handy to remember this.
Oh, I haven't introduced myself yet. My name is Banana.
Recently, I've been sharing information about monetizing AI-generated images.
While earning up to 3 million yen per month through freelance client work, I'm also working on AI image monetization as a separate stream.
Here is a video explaining exactly how I made 450,000 yen in 3 months.
https://x.com/1banana2546/status/2098335744108741085
That's all.
