Components & Elements of a Prompt
Sample Prompt & Result:
A sample prompt can be as straightforward as a single word, phrase or a proper sentence.
Result:
However, writing better instructions will lead to better results, so let’s look at how to improve your prompt writing.
💡Pictro AI breaks down the keywords in a prompt into components, each component being an instruction for an element in the desired image. The elements are compared to the training data of the Pictro AI Model and then used to generate your requested image. A well-structured and organized prompt will help you create beautiful and unique images. Below are the components for a good basic prompt including their elements. Not all elements need to be included in every prompt; choose the ones that matter most to you.
Quality Keywords / Creating Better Quality Images
Quality: The quality of the images from the dataset used to generate your requested image.
Resolution: The resolution of the images from the dataset used to generate your requested image.
Detail Level: The requested amount of details in your generated image.
Subject / Deciding the Central Focus of an Image
Subject: Your chosen topic for the image: character, animal, location, object, or a scene. Be creative with your subject, combining topics to create interesting, thought-provoking, and inspiring artworks.
Adjective Examples for Character Subject: Keywords that describe the details of your character, starting with general info like sex and age. Continue with physical traits such as hair color & hairstyle, eye color, and facial expressions. Then describe body traits such as body type & pose, and continue adding descriptions for things like clothing & accessories.
Adjective Examples for Topic Subject: Think of things like size, color, position, patterns, anything specifically describing your subject rather than the overall image you’re creating.
Environment Keywords / Setting the Image Appearance
Environment: The place or surroundings you want your image to show. Examples: forest, medieval castle, or city street.
Background: The scenery or objects you want behind the main subject of your image. Examples: sunset, bookshelves, or a starry sky.
Time of Day: When you want your image to display, Examples: during the day or at night.
Season: The time of year you want your image to display. Examples: spring or winter.
Era: The time period you want your image to display. Examples: ancient Egypt, or the future.
Mood: The feeling you want your image to display. Examples: happy or mysterious.
Colors: The main colors you want your image to contain. Examples: bright or warm colors.
Lighting: How you want your image to be lit. Examples: with soft light or spotlight.
Weather: The type of weather you want your image to display. Examples: sunny or snowy.
Atmosphere: The overall vibe or ambiance of the scene in your image. Examples: peaceful, mysterious, or energetic.
Art Medium: The type of art format you want your image to look like. Examples: a painting or photograph.
Cultural References: Styles or themes from a culture or time period you want to display in your image. Examples: Disco or Grunge.
Art Movement: A famous art style you want your image to look like. Examples: Impressionism or Cubism.
Theme: The main concept, message, or story you want your image to express. Examples: friendship, adventure, or growth.
Platforms: Art styles or trends from websites or online platforms you want in your image. Examples: DeviantArt or Artstation.
Artist: Artists whose style you want your image to look like using “By [Artist name].” Examples: “By Picasso” or “By van Gogh.”
Art Techniques: Ways of making art you want in your image. Examples: using dots or brush strokes.
Viewpoint: The angle you want your image to be seen from. Examples: from above or close-up.
Camera Device: The type of camera you want your image to look like it was taken with. Examples: DSLR or Canon EOS 5D Mark IV.
Layout: The organization and positioning of elements within the image, emphasizing how these elements relate to each other spatially. Examples: Rule of Thirds or Symmetry.
Composition: The overall visual arrangement of elements in your image. Examples: Balanced or Rhythmic.
Textures: The feel or look of things in your image. Examples: smooth or rough.
Patterns: Repeating shapes or colors you want in your image. Examples: tiles or tribal.
Photographic Effects: Visual effects you want in your image. Examples: blur, glow, or Bokeh.
Emphasis Syntax / More Precision in Your Instructions
Syntax: Used to provide clearer instructions to the AI on what to prioritize in a prompt.
( ) Parenthesis: Instruct the AI to place a higher priority on the section within. You can use methods such as (((keyword))) or (keyword:1.4) to increase the specific weight an instruction should have within your prompt. Available is between 1.1 and 1.9 in most models. Don’t go above 1.4 though for V5 models.*
[ ] Brackets: Instruct the AI to place a lower priority on the section within. You can use methods such as [[[ keyword ]]] or [keyword:0.4] to decrease the specific weight an instruction should have within your prompt. Available is between 0.1 and 0.9 in most models.
{ } Brackets: Instructs the AI to randomize the keywords within this section. For example, {color} will create different colors in every generation for the specified element in your prompt.
, Comma: Separates the elements of a prompt. For example, "blue eyes, black hair" tells the AI these are two different elements. Models of V4.1 or below prefer more structured, comma-separated prompts, while V5 models and XL models process commas more strictly. So it’s best not to use too many commas with these models unless you really want to emphasize that the AI should start a new element in the prompt.
. Dot: Ends an element within a prompt. Similar to the comma, but more definitive. Where the comma keeps the elements connected, the dot will instruct the AI that the two elements are not related. For example: "A woman in a red dress. A man in a black tuxedo." Separated by a comma, the two are likely to interact with each other, while separated by a dot, it’s more likely they take separate places in the image.
Prompting Insights
Find tips and tricks to write the best prompts:
Balancing Prompt Length and Details:
Prompts can be concise yet effective. A single word can suffice. However, shorter prompts give the AI Model more creative freedom in filling in unspecified details. This means you might end up with undesired elements in your created image if left too much up to chance. Include the essential elements in your prompt. Fewer details mean greater variety but less control.
Careful Word Selection:
Selecting the right words is crucial. Specific synonyms often yield better results. Instead of "big," consider "tiny," "huge," "gigantic," "enormous," or "immense." Use a thesaurus online to look up synonyms and explore vocabulary choices.
Keyword Order Matters:
The priority of the keywords is specified by the order of the prompt, so the placement of these elements is crucial to organizing the instructions.