An AI image description prompt tells a vision model what kind of description you actually need. “Describe this image” can work, but it gives the model very little direction. A better prompt defines the purpose, the level of detail, the evidence to focus on, and what the model should avoid guessing.
That matters because the same image may need very different outputs:
- a one-sentence caption;
- a detailed visual inventory;
- composition and lighting notes;
- a product reference description;
- a screenshot explanation;
- an alt-text draft;
- a prompt-ready visual breakdown.
This guide gives you 30 reusable prompts for those jobs.
If you want the description itself rather than a template, try the free AI Image Describer. It analyzes the selected image locally in your browser and can give you a structured starting point.
A reliable image-description prompt formula
A strong request usually contains four parts:
Task + focus + evidence rules + output format
Example:
Describe this image in detail. Focus on the main subject, setting, composition, lighting, colors, materials, visible text, and important relationships between objects. Describe only what is reasonably visible. If something cannot be confirmed, label it as uncertain. Organize the answer into short sections.
This works better than stacking vague adjectives because it tells the model what to inspect and how to report uncertainty.
For the observation method behind these prompts, read How to Describe an Image for AI.
1. General detailed description
Describe this image in detail. Identify the main subject, scene, important objects, visible action, composition, lighting, dominant colors, materials, style, and readable text. Explain where important elements are positioned relative to each other. Describe only visible evidence and mark uncertain details clearly.
Use this when you want a broad but disciplined first pass.
2. Short image description
Describe this image in one or two clear sentences. Mention the main subject, setting, and one or two visually important details. Avoid speculation, unnecessary adjectives, and details that are too small to confirm.
Good for quick notes, catalogs, or image libraries.
3. Very detailed visual inventory
Analyze this image systematically. List the main subject, secondary objects, foreground, middle ground, background, spatial relationships, visible actions, materials, textures, colors, lighting, shadows, reflections, text, and any details that are unclear. Separate direct observation from interpretation.
Use this when missing a small detail would matter.
4. Composition analysis prompt
Describe the composition of this image. Explain orientation, framing, camera distance, viewpoint, subject placement, balance, symmetry or asymmetry, negative space, foreground/background structure, depth, leading lines, and the visual focal point. Do not invent camera settings that are not visible.
Useful for designers, photographers, and prompt writers.
5. Lighting analysis prompt
Analyze only the lighting in this image. Describe the apparent light direction, softness or hardness, brightness, contrast, shadow direction, highlight behavior, color temperature, backlighting or rim light if visible, and how the light separates the subject from the background. Mark anything uncertain.
6. Color palette description prompt
Describe the visible color palette. Identify the dominant colors, important accent colors, warm-versus-cool balance, saturation, brightness relationships, and where the main colors appear in the frame. Do not invent exact HEX codes unless they are measured separately.
7. Style and medium analysis
Describe the visual medium and style of this image. State whether it appears to be a photograph, illustration, render, screenshot, collage, diagram, painting, or another medium. Then describe visible characteristics such as texture, realism, line quality, depth of field, grain, flat shapes, rendering style, or retouching. Avoid naming a specific living artist as a shortcut.
8. Product photo description prompt
Describe this product image for a commercial creative brief. Focus on the exact product shape, proportions, visible materials, color, hardware, label placement, camera angle, crop, surface, background, lighting, reflections, shadows, and scale cues. Do not invent specifications, features, brand details, or included accessories that are not clearly visible.
For prompt creation after this analysis, see Product Photography Prompts.
9. Ecommerce listing image prompt
Analyze this ecommerce product image. Describe the product itself, viewing angle, framing, background, contact shadow, visible packaging, label areas, material finish, color accuracy cues, and anything that could affect how a buyer understands the item. Flag unreadable text instead of guessing it.
10. Food photo description prompt
Describe this food photograph accurately. Identify the visible dish, ingredients that can be reasonably recognized, plating, portion appearance, garnish, plate or container, camera angle, surface, background, lighting, texture, and color. Do not add ingredients or claim a specific recipe when it cannot be confirmed.
For food-generation work, use AI Food Photography Prompts.
11. Restaurant menu photo prompt
Describe this restaurant dish photo as a menu-production reference. Focus on the food actually visible, portion, item count, plating, garnish, container, camera angle, crop, lighting, background, and consistency cues. Note any detail that must be verified before the image is used commercially.
12. Portrait description prompt
Describe this portrait without identifying the person. Focus on visible pose, expression, gaze direction, hairstyle, clothing, framing, background, lighting, color palette, and photographic treatment. Do not infer identity, ethnicity, religion, profession, health, personality, or other sensitive attributes from appearance.
13. Professional headshot prompt
Analyze this headshot for a professional-photo brief. Describe framing, head position, facial direction, expression, clothing, background, light direction, shadow softness, color balance, crop, and separation from the background. Do not identify the person or invent professional credentials.
For editing ideas, continue with AI Portrait Editing Prompts.
14. Group photo description prompt
Describe this group photo factually. State how many people are clearly visible, how they are arranged, their general poses, where they are looking, the setting, framing, lighting, and visible event or background elements. Do not guess relationships between people, identities, or affiliations.
15. Room or interior description prompt
Describe this interior image. Identify the room type if reasonably visible, major furniture and fixtures, their positions, layout, architectural features, camera viewpoint, lighting sources, materials, dominant colors, and open space. Separate visible facts from design-style interpretation.
16. Real-estate exterior prompt
Describe this property exterior factually. Focus on visible building form, number of clearly visible levels, roof shape, windows, doors, driveway, landscaping, camera angle, surrounding context, weather or sky conditions, and light. Do not infer price, neighborhood quality, ownership, or hidden features.
17. Landscape description prompt
Describe this landscape image. Identify the major natural or built elements, foreground, middle ground, background, horizon position, depth, weather or atmospheric conditions, light direction, dominant colors, and overall composition. Avoid guessing the exact location unless text or unmistakable evidence supports it.
18. Travel photo prompt
Describe this travel photograph using only visible evidence. Explain the main subject, setting, architecture or landscape, people if present, composition, lighting, colors, readable signs, and details that might help someone understand the scene. Do not guess the exact city, country, date, or event without clear evidence.
19. Street photo prompt
Analyze this street photograph. Describe visible people without identifying them, vehicles, buildings, signs, road or sidewalk structure, camera viewpoint, depth, lighting, weather, and important interactions. Separate observable action from assumptions about motives or relationships.
20. Screenshot description prompt
Describe this screenshot as a user-interface document. Identify the application or page type only if it is clear, then describe the major layout regions, navigation, buttons, cards, charts, tables, fields, alerts, and readable text. Preserve the order of the interface and do not invent text that is too small to read.
21. Analytics dashboard prompt
Analyze this analytics dashboard. Describe the visible metric cards, charts, tables, filters, date ranges, trends that can be directly seen, and layout. Copy exact values only when legible. Do not infer causes for increases or decreases unless the image itself provides that explanation.
22. Chart description prompt
Describe this chart accurately. Identify the chart type, axes, labels, units, categories, series, legend, highest and lowest visible values, major trends, and any annotations. Distinguish what the chart directly shows from any interpretation or causal explanation.
23. Diagram or flowchart prompt
Describe this diagram step by step. Identify the main boxes, labels, arrows, branches, decision points, direction of flow, repeated structures, and relationships between elements. Preserve the visible sequence and mark text as unclear when it cannot be read reliably.
24. Code screenshot prompt
Describe this code screenshot without pretending to read text that is illegible. Identify the editor layout, visible language clues if clear, file tree, terminal or console panels, function or class structure that can be read, errors or warnings that are legible, and important UI state. Quote code only when it is clearly readable.
25. Artwork analysis prompt
Describe this artwork based on visible characteristics. Explain medium or apparent medium, subject matter, composition, palette, shapes, line quality, texture, depth, perspective, lighting treatment, and mood created by those visual choices. Do not claim a specific artist, date, movement, or meaning unless there is clear supporting evidence.
26. Fashion image prompt
Describe this fashion image as a visual styling reference. Focus on garment type, silhouette, visible construction, fabric appearance, color, layering, accessories, pose, framing, background, and lighting. Do not infer brand, price, size, or personal attributes that are not visible.
For generation briefs, see Fashion AI Image Prompts.
27. YouTube thumbnail analysis prompt
Analyze this YouTube thumbnail visually. Describe the focal subject, facial expression if visible, object placement, background, contrast, dominant colors, empty space, text position, title readability, and what creates the strongest visual hierarchy. Do not judge click-through performance from the image alone.
For creation prompts, use YouTube Thumbnail Prompts.
28. Alt-text draft prompt
Write a concise alt-text draft for this image based on the page purpose: [insert page context]. Mention the information a user would need to understand the image in that context. Avoid redundant phrases such as “image of,” avoid decorative detail that does not help the page purpose, and do not guess unclear information. Then list any uncertainty separately for human review.
AI-generated alt text should always be reviewed in context before publishing.
29. Image-to-prompt bridge
First describe this image factually: subject, scene, composition, viewpoint, lighting, colors, materials, important details, and uncertainty. Then create a separate generation-ready visual brief that preserves those confirmed details without claiming it is the original prompt. Clearly label the factual description and the creative prompt as two different outputs.
This is the safest way to move from observation into generation. You can also use our Image-to-Prompt Generator.
30. Verification-first description prompt
Analyze this image with a verification-first approach. Create three sections: Confirmed visible details, Likely but uncertain details, and Details that cannot be determined from the image alone. Include subject, scene, composition, lighting, colors, text, and relevant objects. Do not fill gaps just to make the description sound complete.
This prompt is especially useful when accuracy matters more than creativity.
How to make any image-description prompt better
Add the purpose
Instead of:
Describe this image.
Use:
Describe this image for a product catalog.
or:
Describe this screenshot so a developer can understand the interface state.
The purpose helps the model decide what deserves attention.
Ask for evidence, not confidence
A model can sound confident while being wrong. Asking for “what is visible” and “what is uncertain” usually produces a more reviewable output than asking it to be “100% accurate.”
Request relationships
Descriptions improve when the model explains where things are:
- left or right;
- foreground or background;
- above or below;
- behind or beside;
- centered or off-center;
- large or small relative to the frame.
This is particularly useful if the description will later become an AI image prompt.
Keep factual and creative outputs separate
If you want both an image description and a generation prompt, ask for two labeled sections. Otherwise creative language can leak into the factual description.
For examples of what strong finished descriptions look like, see AI Image Description Examples.
Prompt template you can reuse
Copy this and change the bracketed parts:
Describe this image for [purpose]. Focus on [details that matter]. Explain the main subject, setting, composition, lighting, colors, materials, visible text, and relationships between important elements. Describe only what is reasonably visible. Do not guess [details that must not be inferred]. If something cannot be confirmed, label it as uncertain. Format the result as [short paragraph / bullet list / labeled sections / table].
Example for a product team:
Describe this image for an ecommerce creative brief. Focus on product geometry, materials, color, camera angle, lighting, reflections, surface, background, label placement, and scale. Describe only visible evidence. Do not invent specifications, included accessories, or unreadable label text. Format the result as labeled sections.
When a plain “describe this image” prompt is enough
A simple prompt is often enough when:
- the image is straightforward;
- you only need a rough overview;
- nothing important depends on exact wording;
- the result will be manually reviewed.
Use a structured prompt when:
- small visual details matter;
- the image contains data or text;
- the description will become alt text;
- you are creating a commercial product brief;
- you plan to turn the description into a generation prompt;
- unsupported guesses could cause problems.
Frequently Asked Questions
What is the best prompt to ask AI to describe an image?
A reliable general prompt asks for the main subject, scene, composition, lighting, colors, style, visible text, important relationships, and uncertainty, while telling the model to stick to visible evidence.
How do I ask AI to describe an image in detail?
Ask for a structured visual inventory: subject, foreground, background, object relationships, composition, lighting, color, materials, text, and uncertainty. Specify that the answer should separate observation from interpretation.
What prompt should I use for an image description for accessibility?
Include the page context and ask the model to describe only information that matters to the purpose of the image. Treat the result as a draft and review it against accessibility guidance before publishing.
Can an image-description prompt recreate the original AI prompt?
No. A model can describe visible characteristics and create a new prompt inspired by them, but it cannot reliably recover the exact original generation prompt from pixels alone.
Should image-description prompts include negative instructions?
Yes when accuracy matters. Instructions such as “do not guess identity,” “do not invent unreadable text,” or “mark uncertainty” help define the boundary of the task.