danielrosehill's picture
updated
d6b732a unverified
|
raw
history blame
2.42 kB
metadata
license: apache-2.0
language:
  - en
pretty_name: Test Prompts For Text To Image Model Evaluation

Text To Image Test Prompt Library

A comprehensive collection of evaluation prompts for testing text-to-image AI models across diverse parameters and use cases.

Overview

This repository contains a structured set of test prompts designed to evaluate the capabilities of text-to-image generation models. Rather than focusing on formal evaluation metrics, these prompts are intended for end users who want to test how well a model might perform for their specific use cases.

Purpose

  • Provide a diverse set of real-world test cases for text-to-image models
  • Help users evaluate model performance across different domains and requirements
  • Identify strengths and weaknesses of different models
  • Support informed decision-making when selecting a text-to-image model

Test Categories

This repository includes 32 distinct test categories, with each category containing multiple prompt examples (typically 5 per file). The total collection includes approximately 160 test prompts.

Categories include:

  • Typo Handling: Testing how models handle nonsensical text, obvious typos, and poorly written prompts
  • Marketing Collateral: Prompts for advertisements, billboards, and other marketing materials
  • Lighting Instructions: Testing model response to specific lighting directions
  • Graphic Design: Evaluating capabilities for creating design assets
  • Architectural Renders: Testing architectural visualization capabilities
  • Confusing/Conflicting Prompts: Evaluating how models resolve ambiguous instructions
  • Character Consistency: Testing ability to maintain character attributes
  • Stylistic Following: Evaluating adherence to specific artistic styles
  • Logo Design: Testing capabilities for brand identity creation
  • Emotion/Subtle Emotion: Evaluating ability to convey emotional nuance
  • Very Specific Instructions: Testing precision in following detailed directions
  • And many more specialized categories

How to Use

  1. Browse the category directories to find relevant test prompts for your use case
  2. Use these prompts with your text-to-image model of choice
  3. Compare results across different models using the same prompts
  4. Evaluate which model performs best for your specific requirements