深习堂
课程价格指南博客

开放课程Skill DevelopmentAI Courses

AI Image Generation Basics

Describe the image you want, guide the result, and learn what to edit and check

开始学习
全部章节
14
Locale
English
课程类型
开放课程

Turn an idea into clear image requirements, then guide AI with prompts, reference images, and settings. Learn to identify what needs changing when a result misses the mark, and check image quality, usage rights, and risks before using it. No drawing, design, or image-model background is required.

适合人群

  • Beginners with no background in drawing, design or image-generation models
  • People whose AI-generated images often miss their requirements
  • Content creators making covers, posters, social media graphics or product scenes
  • Professionals who want to save useful image-generation methods for future projects

解决的问题

  • You can picture an image but cannot write clear instructions for generating it
  • When an image misses the mark, you are unsure whether to change the prompt, reference, settings, or part of the image
  • Characters, products or styles change when you move to a new scene
  • An image looks good, but you are unsure whether you can publish it or use it commercially

    • 学完收获

    • Specify the image’s purpose and subject, what must stay the same, and what may change
    • Use prompts, composition, lighting and color to describe the image you want
    • Choose reference images and settings to control variation and check consistency across a series
    • Decide whether a problem calls for a targeted edit, outpainting, layered editing, or a new image
    • Build quality, licensing, and publication checks into the process from request to delivery

    课程亮点

    • Fourteen lessons cover tool selection, editing, troubleshooting and delivery checks
    • Learn how to make decisions across image-generation tools, without depending on one product
    • Practice guiding results by comparing images, revising prompts, and diagnosing problems
    • Handle real products, exact text and brand assets separately from generated content

    全部章节

    14 个章节

    Lesson 1: What You Can Control in AI Image Generation

    Choose between text-to-image, image-to-image, inpainting, outpainting, and layered editing based on the available assets, protected content, and permitted scope of change, then explain how text, images, masks, settings, and human judgment work together to control the result.

    Lesson 2: Choose an Image Tool from the Task

    Choose a primary AI image workflow for a specific task based on starting assets, control requirements, revision methods, delivery requirements, and total effort, then explain which finishing tools should complement it.

    Lesson 3: Why the Same Prompt Produces Different Images

    Explain generation differences through model capability, current conditions, random sampling, and Seed, then choose a reasonable control action for excessive variation, ignored instructions, composition drift, and similar failures.

    Lesson 4: Turn a Vague Idea into a Visual Brief

    Turn a vague image request into a Visual Brief that covers purpose, platform, audience, core message, subject, visual direction, fixed and flexible content, and output limits, while separating generation, finishing, and acceptance information.

    Lesson 5: Write a Prompt the Model Can Follow

    Extract the information needed for the current generation from a Visual Brief, then write a clear, editable prompt using subject, action, scene, composition, camera, lighting and color, style and materials, and only the necessary constraints.

    Lesson 6: Control Composition, Lighting, and Color

    Choose subject hierarchy, composition, shot size, camera angle, lighting, color, and materials for an image's purpose, then translate an abstract mood into visible conditions that reinforce one another.

    Lesson 7: Iterate from the First Image to a Usable Direction

    Compare candidates through exploration, selection, convergence, and finalization; record protected elements, problems, and one change for the next round; and decide whether to continue, branch, or restart.

    Lesson 8: Use Reference Images Correctly

    Choose references for content, composition, pose, style, color, people, products, or structure, then define each image's purpose, protected elements, flexible elements, and exclusions.

    Lesson 9: Keep Characters, Products, and Styles Consistent

    Separate fixed and flexible variables for a character, person, product, or brand series; build a reusable consistency card; and choose corrections based on the level of drift.

    Lesson 10: Control the Range of Change with Settings

    Distinguish content problems from setting problems, then choose canvas ratio, Seed, randomness, style strength, reference weight, variation strength, quality, and output settings for exploration, convergence, and finalization.

    Lesson 11: Edit Images with Inpainting, Outpainting, and Canvas Tools

    Choose regeneration, variation, inpainting, outpainting, or layered canvas editing for whole-image, directional, local, boundary, and compositing problems, then define masks, protected regions, and edit order.

    Lesson 12: Troubleshoot Image Problems

    Describe image problems as observable facts, locate them at the content, visual, technical, or delivery layer, and choose a cost-effective correction and stopping condition across prompts, references, settings, and editing methods.

    Lesson 13: Check Copyright and Safety Before Commercial Use

    Evaluate AI image risk across inputs, tools and models, generation, outputs, and publication; distinguish tool permission, copyright protection, and commercial responsibility; and plan disclosure, recordkeeping, and professional review.

    Lesson 14: An AI Image Workflow from Request to Delivery

    Connect the Visual Brief, generation boundaries, tool workflow, prompt, references and settings, candidate selection, editing, quality review, risk review, and delivery records into an explainable AI image decision chain with clear rollback points.