Skip to content
ImagePrompt.cv
BETA

A reference you love. A prompt you can make your own.

Try Image to Prompt

The field guide

Better prompts begin with better questions.

Practical lessons in visual language. Choose a guide and start creating with more intention.

Start here4 min read

Your first image-to-prompt workflow

Turn a reference into a prompt you can actually refine.

Read the guide →
  1. Choose a clear reference. Start with one subject and readable light. Avoid a collage for your first attempt.
  2. Choose General and Balanced. This gives you an editable natural-language starting point.
  3. Generate and read critically. Check the subject, setting, light direction, and visual style against the image.
  4. Make one deliberate change. Swap the setting or palette while keeping the composition.
  5. Copy and test. Use the prompt in your chosen image generator, then refine what changed.
Put it into practice →
Prompt craft6 min read

Build a prompt around what matters

Give each visual choice a clear job.

Read the guide →

A useful order is subject → setting → composition → lighting → material → mood. You do not need every field for every image.

Instead of “beautiful cinematic product photo,” try: “A brushed steel desk lamp on an oak surface, close three-quarter view, soft window light from the left, warm neutral background, quiet editorial mood.”

Remove words that do not change the picture. If the subject is drifting, put its defining features first and simplify the surrounding detail.

Put it into practice →
Workflow5 min read

General, structured, or JSON?

Choose the output that fits your next step.

Read the guide →

General reads naturally and is easy to revise in a creative tool. Structured separates the image into labeled components, useful when you want to change only lighting or composition. JSON is useful for a workflow that expects named values.

Keep a stable set of fields when comparing references. Review the content before handing it to another tool; a neatly formatted result can still contain a mistaken visual detail.

Put it into practice →
Visual language7 min read

Describe light with confidence

Replace vague mood words with visible lighting decisions.

Read the guide →

Ask four questions: Where is the light? How large is the source? How strong is the contrast? What color is it?

“Soft window light from camera left, gentle shadow falloff, cool ambient fill” is more actionable than “professional lighting.” A small source tends to create harder shadows; a large source tends to soften their edges.

When refining a prompt, change only the light description first. Keeping the subject and framing stable makes the effect easier to judge.

Put it into practice →
Composition5 min read

Direct the frame, not just the subject

Use viewpoint, scale, and negative space to guide attention.

Read the guide →

Name the viewpoint: eye-level, low angle, overhead, or close three-quarter view. Then name the framing: close-up, medium shot, or wide scene.

Describe where the subject sits and what surrounds it: “A small figure in the lower-right corner, wide open sky, a distant ridge crossing the lower third.” This gives the scene a structure instead of a list of objects.

If you need room for a headline, specify the side and amount of clean negative space. Add text later in a design tool when exact typography matters.

Put it into practice →
Good practice3 min read

Work thoughtfully with visual references

Keep ownership, accuracy, and privacy in your workflow.

Read the guide →

Use images you are authorized to process. Before uploading, remove confidential text, personal documents, and unnecessary identifying details.

AI analysis is an interpretation. Confirm names, objects, materials, and any text before relying on the result. A generated prompt does not establish copyright ownership or permission to reuse a visual.

Recent prompt history is stored in your browser. Clear it when using a shared device, and read the privacy policy for the current handling of image uploads.

Put it into practice →