A worked example: plan the image before choosing a style
Suppose the image is for a founder’s profile, not a fictional fashion campaign. Start with the job the image must do: show a real person as approachable and credible in a square social profile. Keep the subject description factual, choose one visual style, and use lighting and lens controls to support that purpose. The builder puts those decisions into a reusable prompt; it does not inspect a reference photo or guarantee the image model will follow every instruction.
Build and refine in separate passes
Generate once with the default subject and controls. Check whether the face, framing and background communicate the intended role.
Change only one control, such as soft window light to neutral studio light. Generate again and compare the effect.
When the composition works, add one concrete detail in “Extra direction,” such as a plain navy jacket and a softly blurred office background.
Paste the prompt into the image tool you use. Review the actual image for identity, hands, text, logos and other errors before publishing.
What the output can and cannot do
Camera and lighting phrases are creative directions, not technical guarantees. Different image systems interpret “85mm” or “shallow depth of field” differently. The prompt builder runs in your browser and does not send your entries to TrendPromptLab. The chosen image service may have its own data and content rules, so check those separately before uploading a face photo or using an image commercially. Use a real person’s likeness only with permission.
Use this sequence when a portrait prompt misses its purpose. Each pass changes one decision so the result is easier to judge; these are review steps, not a claim that a particular image model will obey the prompt.
Pass
Keep fixed
Change
Inspect in the result
Baseline
Founder, approachable profile, square crop
Use the current style, light and camera choices
Can a viewer recognize the person at thumbnail size?
Light pass
Subject, framing, style
Change only soft window light to overcast daylight
Does the face remain clear without flattening the background?
Framing pass
Subject, light, style
Change only the crop or camera distance
Is there enough space around the head for the intended crop?
Record the setting and the visible difference after each generation. If two controls change together, you cannot tell which instruction helped. The prompt is a creative brief; inspect the chosen output for identity, anatomy, text and rights before using it.
Reproducible example · October 24, 2026
Exact input, actual draft, editorial review
This fictional brief was run through this page’s current text builder. The output below is captured from the builder, not a result from an external AI service or a customer campaign.
See all input settings
Subject
a ceramic teapot on a wooden table
Style
cinematic editorial portrait
Lighting
soft window light at 45 degrees
Camera
85mm portrait lens, shallow depth of field
Ratio
vertical 4:5
Extra
plain cream glaze, no decorative pattern
Captured output
Create an original cinematic editorial portrait of a ceramic teapot on a wooden table. Use soft window light at 45 degrees, with a 85mm portrait lens, shallow depth of field look. plain cream glaze, no decorative pattern. Natural detail, coherent anatomy, intentional composition, accurate materials and restrained retouching. No logos, no readable text, no celebrity likeness, no copyrighted characters. Output as vertical 4:5.
What the result shows
The teapot, material and framing are explicit. Camera terminology describes a look; it does not control an actual lens.
What to change before publishing
If the model adds a decorative pattern, remove the style phrase first and repeat the plain-glaze constraint. Inspect the silhouette and spout before using the picture.