InAI image generationincreasingly powerful today.Generate a "photo", seems to have become a thing that everyone can do.
But many newbie hacks will find thatThe tools are readily available, the hard part is how to ask for them."I want a street modeling photo" may seem clear, but the actual AI ends up generating results that are so sky high that drawing cards is more of a waste of time than the actual photo shoot.
How to tell it to AI? I've chosen a few popular graphing tools for comparison testing this time.
You will see a set of generic and efficientPhotographsVincent FigureCue word structure and will provide you with meta cue words. No knowledge of photography is required, and you don't have to be an AI expert; just type it into the AI and a simple description can be expanded into a specialized cue phrase.
Let's you say goodbye to card draws, all of them are good pictures.
Generic perspectives for describing photos can be roughly summarized in six areas.
Next I'll try 5 examples using a couple of old and commonly used chart-generating tools.
separately fromPhotographs of people, scenery, and objects on displayin, get a feel for the cue words for this structure in thei.e. Dreams, Midjourney and GPT4oThe effects are generated in all three models.
Among them, i.e., Dream and GPT4o use Chinese cue words, and Midjourney is generated using the English results of Google Translate, with an additional scale parameter (-ar 3:2).
Example 1｜Street Portrait Realistic Style
- An Asian male in his 20s, wearing a light gray sweatshirt and jeans, stands on a street in Shibuya, Tokyo, with a relaxed and smiling expression. In the background are blurred street scenes and pedestrians, with reflections of rainwater on the ground. Natural light from the side and back at sunset creates a warm glow. Using a shallow depth of field, the people are clear and the background is blurred, and the photo is close to street documentary photography, with a slightly grainy film feel.
Dream
Midjourney
GPT4o
Example 2｜Interior Living Photo Style
- A 30-year-old white woman sits in a wooden chair by the window, wearing a beige sweater and holding a book. Outside the window is a clear blue sky, and sunlight pours in through the white gauze curtains, illuminating the side of her face. In the background is a row of wooden bookshelves and a small coffee table. The composition is in thirds, the camera is horizontal and slightly overhead, and the overall style is warm and quiet, simulating a portrait of life in natural light.
Dream
Midjourney
GPT4o
Example 3｜Natural Landscape Style
- In a valley in the morning mist, a small stream runs through a dense pine forest, and the sunlight pours through the mist, creating a golden beam of light. The composition is diagonal, with a moss-covered rock at the bottom of the picture. In the distance there is a deer with its head lowered to drink, the style tends to be dreamy and realistic, the natural light is soft, the colors are cold, and there is a slight soft focus.
Dream
Midjourney
GPT4o
Example 4｜Animal Realistic Style (Pet Photography)
- A golden retriever is sitting on the grass in a park, looking right at the camera with his tongue naturally lolling out and a friendly expression. In the background are blurred green trees and dappled shadows from the sunlight. The sunlight is shining from the back left, creating a warm glow around the edges of the fur. Using a shallow depth of field, the dog is clear and the background is softly bokeh. The overall style is warm and realistic, similar to a family pet photography magazine cover.
Dream
Midjourney
GPT4o
Example 5｜Still Life Product Photography (Food)
- Different types of desserts and cakes on different china plates, the desserts are beautifully made. In the background is a wooden dining table with a few small flowers and a milk jug scattered around. Natural light comes in through the window on the right, creating soft shadows and patches of light. The top view composition is neat and symmetrical. The style is fresh and Japanese, suitable for food magazines or social platform food sharing.
Dream
Midjourney
GPT4o
Photo Vincennes - Meta Cue Words
If you say, "I still can't write.
Then here's a cue word that's sure to help.With just a short description, the AI can help you expand it into a full and rich paragraph of Vince's graphic cue words.
# Role
You are a professional AI image cue expansion assistant, good at automatically expanding into complete and detailed photo-based image cues based on users' short Chinese descriptions for text-to-image (text-to-image) tools.
## Expand according to the following 6 dimensions
- Photographing the subject (e.g., people, animals, objects, etc.)
- Environmental context (e.g., location, spatial features, etc.)
- Time light (e.g. sunset, night, soft light, etc.)
- Angular compositions (e.g., overhead, panoramic, thirds, etc.)
- Photographic style (e.g. filmic, sharp modern, bokeh, etc.)
- Detailed additions (actions, expressions, weather, elements, etc.)
## Output Requirements
The output should contain two versions: Chinese prompts and English prompts.
Chinese prompts use descriptive sentences that flow naturally.
The English prompts are styled to follow the Midjourney / SD / DALL-E photography class structure, as close as possible to actual photography descriptions.
## User Requirements
XXX
Let's also look at an example below.
Let's say my user requirement is: I want a photo overlooking a table of digital products.
this timeI use GPT4o, DeepSeek, and Beanbag, respectively, and all use the same meta-prompt words to generate the literate graph instructions. Then use Instant Dreams, Midjourney, and GPT4o to generate images and compare to see how it works.
GPT4o generates prompt words
an image of a dream come true
Midjourney Raw Chart
GPT4o Raw Chart
DeepSeek generates cue words
an image of a dream come true
Midjourney Raw Chart
GPT4o Raw Chart
Beanbag Generation Prompt Words
- The English generation is incomplete, but there is no great impact in terms of Midjourney's results.
an image of a dream come true
Midjourney Raw Chart
GPT4o Raw Chart
To summarize.
If you are a newbie getting started, it is recommended to useBean Bag Expansion + Instant Dream Life PictureIt's just basically enough.
If you are in a position to useGPT4o, then consider using it to expand the Vince's chart prompt chart. It works better.
If a certain raw map effect is sought, as of today, theMidjourneyIt's still worth the rush.
It's not the tools, but the expression that really closes the gap. Your ability to accurately state the picture you have in mind determines whether the AI can paint the world you want.
I hope that this set of cue word structure can become your most reliable "photographic assistant" every time you open the AI drawing tool, helping you to step in the pit less and produce more pictures.