I85PREP
⌘ K
Adobe Certified Professional - Adobe Firefly study materials
Adobe Certified Professional - Adobe Firefly study resources

Adobe Certified Professional - Adobe Firefly practice questions

One question at a time. Build your Adobe Certified Professional - Adobe Firefly understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Generative AI fundamentals and Firefly concepts

Adobe Firefly is best described as which kind of product?

  1. A code repository for version control
  2. A generative AI model family for creating and editing visual and text content Correct answer.
  3. A spreadsheet application for financial modeling
  4. A desktop publishing tool for print layouts

Explanation

Firefly is Adobe's family of generative AI models designed to produce and edit images, text effects, and other creative assets. It is integrated across Adobe applications and offered on the web. It is not a spreadsheet, code repository, or print layout program.

Question 2Generative AI fundamentals and Firefly concepts

What is a prompt in the context of a generative AI tool like Firefly?

  1. A saved preset that fixes the model's output permanently
  2. A license key that unlocks commercial use
  3. A text instruction describing the content you want the model to generate Correct answer.
  4. A file format for exporting generated images

Explanation

A prompt is the natural-language instruction that tells the generative model what to create. Its wording, detail, and structure strongly influence the resulting output. Presets, licenses, and file formats are separate concepts.

Question 3Generative AI fundamentals and Firefly concepts

Which Firefly capability generates decorative lettering and stylized typography from a word or phrase?

  1. Text to image
  2. Generative recolor
  3. Text effects Correct answer.
  4. Generative fill

Explanation

Text effects is the Firefly feature that renders a word or short phrase as stylized, textured lettering based on a prompt. Text to image creates scenes rather than typographic treatments. Recolor and fill modify existing artwork instead of generating lettering.

Question 4Generative AI fundamentals and Firefly concepts

In Firefly, what does the Generative Fill feature do?

  1. It compresses an image to reduce file size
  2. It automatically writes alt text for accessibility
  3. It converts a raster photo into an editable vector graphic
  4. It inserts or replaces content inside a selected area of an image based on a prompt Correct answer.

Explanation

Generative Fill uses a selection as a mask and generates new content inside that region to match a text prompt or surrounding context. It can add objects, remove objects, or extend a scene. It does not vectorize images, author alt text, or handle compression.

Question 5Generative AI fundamentals and Firefly concepts

What is the purpose of a content credential attached to Firefly-generated output?

  1. It encrypts the file so it cannot be opened without a password
  2. It reduces the resolution to a web-safe size
  3. It records provenance information indicating the content was AI-generated Correct answer.
  4. It guarantees the image is free of any copyright claims

Explanation

Content credentials use provenance metadata to signal how and by whom content was created, including whether generative AI was involved. They support transparency rather than encryption or legal guarantees. Resolution and file protection are unrelated to credentials.

Question 6Generative AI fundamentals and Firefly concepts

Which statement accurately describes how a diffusion-based image model typically generates an image?

  1. It stitches together tiles cut from existing copyrighted images
  2. It assembles vector shapes defined by a designer's template
  3. It retrieves the closest matching photo from a stock library
  4. It starts from noise and progressively refines it toward an image matching the prompt Correct answer.

Explanation

Diffusion models begin with random noise and iteratively denoise it, guided by the prompt, until a coherent image emerges. They synthesize new pixel data rather than retrieving stock photos or assembling templates. This process is why small prompt changes can shift the result.

Question 7Generative AI fundamentals and Firefly concepts

A designer wants Firefly to produce several variations from one prompt without rewriting it. Which control is most directly intended for that?

  1. The aspect ratio selector
  2. The number of variations or generated results setting Correct answer.
  3. The download format option
  4. The content type filter

Explanation

Firefly lets you request multiple results per generation so you can compare variations from a single prompt. Aspect ratio, content type, and export format control output shape, category, and file type rather than the count of variations.

Question 8Generative AI fundamentals and Firefly concepts

Why does adding specific details such as subject, setting, lighting, and style to a prompt generally improve Firefly results?

  1. Longer prompts always cost fewer credits
  2. Detailed prompts disable the model's safety filters
  3. Firefly rejects prompts shorter than a fixed word count
  4. The model conditions its output on the descriptive tokens it is given Correct answer.

Explanation

Generative models condition their output on the words in the prompt, so concrete descriptors give the model more to steer toward. Vague prompts leave more to chance. Prompt length is not tied to credit cost or a minimum word count.

Question 9Generative AI fundamentals and Firefly concepts

Which prompt is most likely to yield a controlled, predictable Firefly image?

  1. A ceramic mug on a wooden table, soft window light, shallow depth of field, warm tones Correct answer.
  2. Something nice and modern
  3. Make it amazing
  4. Whatever looks best, surprise me

Explanation

Prompts that name the subject, setting, lighting, and visual qualities give the model concrete attributes to render. Vague or open-ended prompts delegate most decisions to the model, producing less predictable results. Specificity is the key lever for control.

Question 10Generative AI fundamentals and Firefly concepts

A user wants to expand a photo beyond its original borders, generating plausible new scenery around it. Which Firefly workflow fits best?

  1. Vectorize
  2. Generative recolor
  3. Text effects
  4. Generative expand Correct answer.

Explanation

Generative expand enlarges the canvas and generates matching content to fill the new area, effectively outpainted around the original image. Recolor changes colors, text effects renders lettering, and vectorize traces shapes.

Question 11Generative AI fundamentals and Firefly concepts

What does a Firefly 'style reference' or style-matching control primarily influence?

  1. The language of the interface
  2. The file size of the exported image
  3. The number of credits charged per generation
  4. The visual aesthetic of the output, such as color palette and rendering feel Correct answer.

Explanation

Style references guide the aesthetic qualities of generated output, such as palette, texture, and rendering character. They do not govern export size, credit pricing, or interface language. Style controls shape look while the prompt shapes content.

Question 12Generative AI fundamentals and Firefly concepts

Why might a designer use a structural or composition reference when generating with Firefly?

  1. To guide the layout and arrangement of elements while allowing new content Correct answer.
  2. To convert the image to an audio waveform
  3. To remove all color from the result
  4. To lock the output to a single unchangeable pixel grid

Explanation

Composition references steer where elements sit and how the scene is arranged, while the prompt and model still generate new content. They are guidance, not a hard pixel lock. Color removal and audio conversion are unrelated operations.

Question 13Generative AI fundamentals and Firefly concepts

What is the role of a seed value in some generative image workflows?

  1. It selects the output file format
  2. It sets the commercial license tier
  3. It defines the image's copyright owner
  4. It helps reproduce or make controlled variations of a prior result Correct answer.

Explanation

A seed initializes the random noise used in generation, so reusing it with similar settings can reproduce or subtly vary a prior result. It does not determine licensing, ownership, or file format. Seeds support iteration and consistency.

Question 14Generative AI fundamentals and Firefly concepts

A Firefly model is described as trained on licensed and public domain content. What is the main practical benefit emphasized for commercial users?

  1. Generated files cannot be edited by others
  2. The training approach is designed to reduce IP risk for commercial use Correct answer.
  3. Outputs are guaranteed to be unique worldwide
  4. The model runs without any internet connection

Explanation

Firefly was positioned around training on licensed and public domain material specifically to reduce intellectual property risk for commercial use. This does not guarantee uniqueness or prevent others from editing files. Connectivity is a deployment detail, not a training claim.

Question 15Generative AI fundamentals and Firefly concepts

In Adobe's generative credit system, what does a generative credit typically represent?

  1. A measure of image resolution in pixels
  2. A password for accessing cloud storage
  3. A license to redistribute Adobe software
  4. A unit consumed when performing a generative operation such as a text-to-image generation Correct answer.

Explanation

Generative credits meter usage of generative features, with operations consuming credits according to the feature and plan. They are not software redistribution licenses, pixel measures, or credentials. Plans allocate a monthly credit allowance.

Question 16Generative AI fundamentals and Firefly concepts

Which task is Firefly's text-to-image capability designed to perform?

  1. Transcribe speech into written text
  2. Generate a new image from a written description Correct answer.
  3. Translate a document into another language
  4. Compress a video for streaming

Explanation

Text-to-image generates pictorial content from a natural-language description. Speech transcription, video compression, and document translation are different functions outside this capability. The prompt describes the scene to be rendered.

Question 17Generative AI fundamentals and Firefly concepts

A user selects a person in a photo and types a prompt to replace their jacket with a red coat. Which Firefly feature is being used?

  1. Generative expand
  2. Text effects
  3. Auto-tone
  4. Generative fill Correct answer.

Explanation

Generative fill operates inside a selection, generating replacement content that matches the prompt and surrounding image. Expanding extends the canvas, text effects renders lettering, and auto-tone adjusts exposure and color.

Question 18Generative AI fundamentals and Firefly concepts

What is a practical limitation of generative image models that users should plan for?

  1. They cannot accept text prompts
  2. They always produce identical results for identical prompts
  3. They can produce artifacts such as distorted hands or inconsistent details Correct answer.
  4. They can only output black-and-white images

Explanation

Generative models can render anatomy, text, and fine details inconsistently, producing visible artifacts that require review and retouching. They do produce color output, accept text prompts, and vary between runs unless a seed is fixed. Planning for review is part of a sound workflow.

Question 19Generative AI fundamentals and Firefly concepts

Why is human review still recommended before publishing Firefly-generated content?

  1. Generated output can contain errors, bias, or unintended elements that require judgment Correct answer.
  2. The model refuses to run without a second user
  3. Firefly output cannot be saved until a reviewer signs in
  4. Review is required to convert the file to PNG

Explanation

Automated generation can introduce factual errors, biased depictions, or unintended details, so a human should evaluate appropriateness and accuracy. Review is not a technical gate for saving or exporting files, and no second user is required. Judgment remains a human responsibility.

Question 20Generative AI fundamentals and Firefly concepts

A marketing team wants on-brand images quickly without commissioning photography. Which Firefly use case most directly supports this?

  1. Generating concept imagery from brand-described prompts for review Correct answer.
  2. Managing email campaign schedules
  3. Auditing website accessibility compliance
  4. Reconciling accounting ledgers

Explanation

Firefly can rapidly generate concept imagery guided by descriptive prompts, letting teams explore directions before committing to production. Accessibility auditing, campaign scheduling, and accounting are unrelated business functions. The value is fast visual ideation.

Question 21Generative AI fundamentals and Firefly concepts

How does Firefly's integration across Adobe applications primarily benefit a designer?

  1. It lets generative features be used within existing creative workflows and documents Correct answer.
  2. It removes the need for any design software
  3. It forces all work into a single file format
  4. It prevents sharing files with collaborators

Explanation

Firefly features are embedded in Adobe creative tools so generation happens alongside existing editing and layout work. This preserves workflow continuity rather than forcing a single format or eliminating design software. Collaboration remains supported.

Question 22Generative AI fundamentals and Firefly concepts

What does 'commercially safe' generally mean when describing Firefly's design intent?

  1. The tool may only be used for non-profit projects
  2. Output is automatically trademarked to the user
  3. Training and tooling are intended to reduce the risk of infringing third-party rights in commercial work Correct answer.
  4. Generated content cannot be modified afterward

Explanation

The phrase refers to designing the model and its training around licensed and public domain content to lower IP risk for business use. It does not grant trademarks, restrict use to non-profits, or freeze content from editing. Legal review may still be prudent for specific uses.

Question 23Generative AI fundamentals and Firefly concepts

Which factor most directly determines whether a generative feature consumes more or fewer credits per operation?

  1. The number of fonts installed on the computer
  2. The specific feature and settings used for the generation Correct answer.
  3. The color of the user's interface theme
  4. The day of the week the generation occurs

Explanation

Credit consumption depends on the generative feature invoked and its settings, as defined by the plan's terms. Interface theme, installed fonts, and calendar timing do not change credit cost. Users should check plan details for specific rates.

Question 24Generative AI fundamentals and Firefly concepts

A user generates an image that includes garbled, unreadable text in a sign. What is the most appropriate interpretation?

  1. The file is corrupted and must be discarded
  2. The prompt was rejected by safety filters
  3. The model has run out of credits
  4. Rendering legible text is a known weakness, so the area should be revised or retouched Correct answer.

Explanation

Generative models often struggle to render coherent text within images, producing garbled lettering. The practical response is to adjust the prompt, regenerate, or retouch the area. Corruption, filter rejection, and credit exhaustion would present differently.

Question 25Generative AI fundamentals and Firefly concepts

What distinguishes Firefly's approach to generating content from simply searching a stock photo library?

  1. Firefly synthesizes new pixels rather than returning a pre-existing photograph Correct answer.
  2. Firefly cannot produce images of objects
  3. Firefly requires a scanner to capture input
  4. Firefly only returns images that already exist online

Explanation

Generative models synthesize novel pixel data conditioned on the prompt, whereas stock search retrieves existing photographs. This distinction underlies both the flexibility and the review needs of generative output. No scanner or online image return is involved.

Question 26Generative AI fundamentals and Firefly concepts

Which statement best describes the relationship between a generative AI model and the data it was trained on?

  1. The model only works with data that was uploaded by the user in the current session.
  2. The model stores and retrieves exact copies of training images at generation time.
  3. The model requires a live internet search of the training data for every prompt.
  4. The model learns statistical patterns from training data and uses them to produce new outputs. Correct answer.

Explanation

Generative models are trained to capture patterns, relationships, and structures in their training data, then use those learned patterns to synthesize new content. They do not store training examples as a searchable database or retrieve them verbatim during generation.

Question 27Generative AI fundamentals and Firefly concepts

In the context of Firefly, what does the term 'prompt' refer to?

  1. A text description that guides the model toward the desired output. Correct answer.
  2. A saved preset that controls the application's export settings.
  3. A licensing flag that marks an image as commercially safe.
  4. A numerical value that fixes the random seed for a generation.

Explanation

A prompt is the input instruction, usually text, that describes what the user wants the generative model to create. It is the primary way users communicate subject, style, and other desired attributes to the model.

Question 28Generative AI fundamentals and Firefly concepts

A user wants to generate an image of a 'futuristic city at sunset' but the first result looks like a daytime scene. Which adjustment is most likely to bring the output closer to the intended time of day?

  1. Add explicit lighting and time-of-day terms to the prompt. Correct answer.
  2. Reduce the number of generation steps to speed up the process.
  3. Increase the output resolution to the maximum available.
  4. Switch the file format from PNG to JPEG before generating.

Explanation

Prompts guide the model, so making the desired lighting and time of day explicit helps steer the output toward the intended scene. Resolution, file format, and step count affect quality or speed rather than the semantic content described in the prompt.

Question 29Generative AI fundamentals and Firefly concepts

Which statement best explains why Firefly's training on licensed and public domain content matters to a business user?

  1. It reduces legal risk because the model was designed with commercial use in mind. Correct answer.
  2. It ensures the model never produces content similar to any existing work.
  3. It guarantees that every generated image is unique and has never existed before.
  4. It allows users to bypass all review before publishing generated assets.

Explanation

Training on licensed and public domain content is intended to make outputs safer for commercial use by reducing intellectual property concerns. It does not guarantee uniqueness or eliminate the need for human review.

Question 30Generative AI fundamentals and Firefly concepts

A designer is using Firefly to create a product mockup and wants to keep the product's shape and edges intact while changing the background. Which approach is most appropriate?

  1. Use a generative fill or inpainting workflow with a mask around the background area. Correct answer.
  2. Use a higher seed value to lock the product's shape during generation.
  3. Use a text-to-image prompt that describes the product and the new background together.
  4. Use a style reference image that matches the desired background aesthetic.

Explanation

Mask-based editing restricts generation to a selected region, which preserves unselected areas such as the product. Text-to-image regenerates the whole scene, and style references influence appearance rather than protecting specific pixels.

Question 31Generative AI fundamentals and Firefly concepts

What is the primary purpose of a negative prompt in generative image workflows?

  1. To specify elements that should be excluded from the generated image. Correct answer.
  2. To lower the resolution of the output for faster rendering.
  3. To reverse the colors of the generated image for artistic effect.
  4. To reduce the number of credits consumed by a generation.

Explanation

A negative prompt tells the model which concepts, objects, or styles to avoid, helping reduce unwanted elements in the output. It does not affect billing, color inversion, or resolution.

Question 32Generative AI fundamentals and Firefly concepts

A user notices that Firefly sometimes renders hands with an incorrect number of fingers. What is the most accurate explanation for this behavior?

  1. The model's training data contained limited or inconsistent examples of hands. Correct answer.
  2. The output resolution was too high for the model to render correctly.
  3. The user's prompt lacked a seed value to stabilize the anatomy.
  4. The model has a deliberate artistic style that always distorts anatomy.

Explanation

Generative models reproduce patterns present in their training data; when a subject is underrepresented or highly variable, errors such as malformed hands can occur. This is a data-driven limitation rather than a deliberate style or a resolution issue.

Question 33Generative AI fundamentals and Firefly concepts

Which scenario best illustrates the concept of 'outpainting' in Firefly?

  1. Extending an image beyond its original boundaries with generated content. Correct answer.
  2. Applying a color grade to match a reference photograph.
  3. Replacing a selected object inside an image with a newly generated object.
  4. Removing a background from a subject to create transparency.

Explanation

Outpainting generates plausible content outside the existing image frame, effectively expanding the canvas. Inpainting replaces content inside a selection, and background removal or color grading are separate editing operations.

Question 34Generative AI fundamentals and Firefly concepts

A user wants to create a consistent character across multiple Firefly generations. Which practice is most likely to help?

  1. Generating at the lowest possible resolution to reduce variation.
  2. Avoiding any mention of the character's physical traits in the prompt.
  3. Changing the seed value randomly for each generation.
  4. Reusing a detailed character description and a consistent style reference. Correct answer.

Explanation

Consistency is improved by providing stable descriptive details and a consistent visual reference that the model can follow. Random seeds and sparse prompts increase variation rather than reduce it.

Question 35Generative AI fundamentals and Firefly concepts

Which statement best describes how a generative model's output relates to its training data?

  1. Outputs are independent of training data and rely only on the prompt.
  2. Outputs are exact reproductions of the most similar training image.
  3. Outputs are synthesized from learned patterns and may resemble, but not copy, training examples. Correct answer.
  4. Outputs are assembled by pasting together fragments of training images.

Explanation

Generative models synthesize new content by combining learned patterns; outputs can resemble training examples without being direct copies. They are not assembled from pasted fragments, and they depend heavily on training data.

Question 36Generative AI fundamentals and Firefly concepts

A team wants to use Firefly to create variations of a logo concept while keeping the layout close to an approved sketch. Which Firefly capability is most directly relevant?

  1. Applying a blur filter before generation to simplify the sketch.
  2. Increasing the number of generated variations without any reference.
  3. Using a negative prompt to exclude all other logo styles.
  4. Using a structural or composition reference to guide the layout. Correct answer.

Explanation

A structural reference communicates composition and layout to the model, helping generated variations stay close to an approved arrangement. Negative prompts and variation counts do not convey the specific layout.

Question 37Generative AI fundamentals and Firefly concepts

Which factor most directly determines whether a Firefly operation consumes more or fewer generative credits?

  1. The length of the file name assigned to the output.
  2. The number of layers in the source document.
  3. The type of operation and the model or settings used. Correct answer.
  4. The color profile selected for the document.

Explanation

Credit consumption is tied to the generative operation performed and the model or settings chosen, such as resolution or feature type. File naming, color profiles, and layer counts do not determine credit cost.

Question 38Generative AI fundamentals and Firefly concepts

A Firefly-generated image includes a sign with text that reads as meaningless symbols. What is the most appropriate interpretation?

  1. The output resolution was too low for any text to appear.
  2. The prompt contained a negative prompt that disabled text rendering.
  3. The model struggles to render legible text and the result should be reviewed or corrected. Correct answer.
  4. The model intentionally encrypts text to prevent copyright issues.

Explanation

Generative image models often produce imperfect or illegible text because text rendering requires precise structure that is difficult to learn from image data. Users should review such outputs and correct or regenerate as needed.

Question 39Generative AI fundamentals and Firefly concepts

A designer needs to generate an image that matches a specific brand color palette. Which Firefly control is most directly suited to this requirement?

  1. A seed value entered as a hexadecimal number.
  2. A negative prompt listing unrelated objects.
  3. A color and tone reference or palette control. Correct answer.
  4. A higher generation step count.

Explanation

Color and tone references or palette controls are designed to steer the generated image toward specified colors. Seeds, negative prompts, and step counts do not directly enforce a color palette.

Question 40Generative AI fundamentals and Firefly concepts

Which statement best explains why a generative model can produce a plausible but factually incorrect depiction of a historical event?

  1. The model only generates content that has been verified by human editors.
  2. The model requires a fact-checking database to be connected before generating.
  3. The model prioritizes visual plausibility learned from patterns over factual accuracy. Correct answer.
  4. The model deliberately falsifies history to avoid controversial content.

Explanation

Generative models are optimized to produce outputs that resemble their training distribution, not to verify facts. They can therefore generate convincing imagery that is historically or factually inaccurate.

Question 41Text-to-image generation and prompt crafting

In a text-to-image prompt, what is the primary role of the descriptive subject text?

  1. It controls the account's credit consumption rate
  2. It determines how many images the model returns
  3. It sets the file format the model will export
  4. It defines what the main content of the generated image should be Correct answer.

Explanation

The subject portion of a prompt tells the model what to depict, forming the core semantic content the generator renders. Formatting, batch count, and credit usage are set by interface controls rather than by descriptive prompt wording.

Question 42Text-to-image generation and prompt crafting

Which prompt element most directly influences the visual style of a generated image?

  1. Style descriptors such as medium and artistic movement Correct answer.
  2. The length of the prompt in characters
  3. The number of subjects named
  4. The order in which colors are listed

Explanation

Style words such as 'watercolor,' 'oil painting,' or 'art deco' steer the rendering medium and aesthetic. Subject count, color ordering, and raw character length do not by themselves define a style.

Question 43Text-to-image generation and prompt crafting

A user wants a specific composition with the subject on the left and open sky on the right. Which prompt addition best supports this?

  1. Adding a long list of unrelated adjectives
  2. Naming a popular camera brand
  3. Increasing the image resolution setting
  4. Describing the spatial arrangement explicitly Correct answer.

Explanation

Layout is best controlled by stating where elements should appear, such as 'subject on the left, open sky on the right.' Camera brands, adjective lists, and resolution settings do not communicate spatial placement.

Question 44Text-to-image generation and prompt crafting

Why does adding irrelevant or contradictory details to a prompt often reduce output quality?

  1. Contradictions force the model into black-and-white output
  2. The model rejects prompts above a fixed word count
  3. The interface silently truncates the first half of the prompt
  4. Extra terms dilute focus and can conflict with the main subject Correct answer.

Explanation

Each term influences the result, so unrelated or conflicting descriptors compete for emphasis and blur the intended subject. There is no universal word limit that rejects prompts, and contradictions do not force monochrome output.

Question 45Text-to-image generation and prompt crafting

Which prompt is the clearest example of specifying a lighting condition?

  1. A castle lit by warm golden-hour sunlight Correct answer.
  2. A castle on a hill
  3. A castle in a wide aspect ratio
  4. A castle rendered at high resolution

Explanation

Lighting descriptors name the quality, direction, or time of light, such as golden-hour warmth. The other options describe subject, resolution, or aspect ratio rather than lighting.

Question 46Text-to-image generation and prompt crafting

What does the aspect ratio setting in a text-to-image generation primarily control?

  1. The rendering style of the image
  2. The number of subjects in the scene
  3. The color palette of the result
  4. The proportions of the output canvas Correct answer.

Explanation

Aspect ratio defines the width-to-height proportions of the generated canvas, such as square or widescreen. It does not determine subject count, palette, or style.

Question 47Text-to-image generation and prompt crafting

A designer needs a prompt for a vertical social media story. Which adjustment is most appropriate?

  1. Add more style adjectives to the prompt
  2. Switch the aspect ratio to a tall portrait orientation Correct answer.
  3. Reduce the prompt to a single word
  4. Request a square canvas and crop later

Explanation

A vertical story format calls for a tall portrait aspect ratio so the composition fits the target placement. Adding adjectives, shortening the prompt, or generating square and cropping do not directly produce a native vertical composition.

Question 48Text-to-image generation and prompt crafting

What is the main purpose of a negative prompt?

  1. To specify the output file type
  2. To describe elements the user wants excluded Correct answer.
  3. To set the generation seed value
  4. To list the elements the user wants emphasized

Explanation

A negative prompt lists attributes or objects the user wants the model to avoid. Emphasis, file type, and seed values are handled by other prompt phrasing or settings.

Question 49Text-to-image generation and prompt crafting

A user keeps getting text-like artifacts in an image that should have no lettering. What is the most direct prompt-based remedy?

  1. Add more color adjectives to the positive prompt
  2. Add words like 'text' and 'letters' to the negative prompt Correct answer.
  3. Increase the aspect ratio to widescreen
  4. Add 'watermark' to the positive prompt

Explanation

Placing unwanted elements such as text, letters, or watermarks in the negative prompt signals the model to avoid them. Adding those words positively or changing aspect ratio and colors does not target the artifact.

Question 50Text-to-image generation and prompt crafting

Why is iterating on a prompt in small steps considered good practice?

  1. It isolates which wording changes affect the output Correct answer.
  2. It guarantees the first result will be usable
  3. It avoids consuming any generation credits
  4. It removes the need for a negative prompt

Explanation

Changing one element at a time lets the user attribute differences in output to specific wording, making refinement systematic. Iteration does not guarantee usability, eliminate credit use, or remove the value of negative prompts.