L1
·
Quiz
·
Lab
L2
·
Quiz
·
Lab
L3
·
Quiz
·
Lab
L4
·
Quiz
·
Lab
Module Test
Module 4 · Lesson 1

The Landscape: Major AI Image Tools

Midjourney, DALL·E, Stable Diffusion, Firefly — what each one actually does and who uses it
How did a handful of tools born in 2022 change what it means to make a picture?

When Midjourney opened its Discord beta in April 2022, digital artist Greg Rutkowski discovered that his name had become one of the most-typed phrases in AI image prompts — users invoked it thousands of times daily to conjure his distinctive oil-painting fantasy style. Rutkowski had not consented, had not been paid, and could not easily stop it. His situation spotlighted a question the entire industry was unprepared to answer: what is the relationship between an AI tool's outputs and the human work it was trained on?

That question remains contested, but the tools themselves kept shipping. Within eighteen months, four distinct platforms had established clear identities, each attracting different types of creative professionals.

Four Tools, Four Different Bets

Each major platform made a distinct architectural and product decision that shapes everything about how you use it today.

Midjourney launched in April 2022 via Discord and deliberately avoided releasing a public API or open-source weights. Its founder David Holz described the goal as producing images that feel "aesthetically interesting" rather than photorealistic. By 2023 Midjourney was the most-cited AI image platform in professional design contexts, with an estimated one million daily active users by mid-2023.

DALL·E 3 (OpenAI, released October 2023) took the opposite bet: tight integration with ChatGPT so that natural language could drive image creation without prompt engineering. OpenAI also built in automatic content filtering and, crucially, introduced a policy declining to generate images in the style of living artists by name — a direct response to the Rutkowski controversy.

Stable Diffusion (Stability AI, released August 2022) chose open-source distribution. Anyone can download the model weights, run them locally, and fine-tune on their own images. This decision made Stable Diffusion the foundation for thousands of derivative tools, LoRA fine-tunes, and community models, and also made content moderation essentially impossible to enforce at scale.

Adobe Firefly (public beta March 2023, general availability September 2023) was built on a fundamentally different training philosophy: Adobe claims the model was trained exclusively on licensed Adobe Stock images, openly licensed content, and public domain works. This "commercially safe" positioning was specifically aimed at professional designers who need to use outputs in client work without IP liability.

A Quick Tool Comparison
Tool Access Model Best Known For Primary User Base
Midjourney v6 Subscription, Discord/web Painterly, high-aesthetic outputs; strong composition Concept artists, illustrators, game studios
DALL·E 3 ChatGPT Plus / API Following complex text instructions; legible text in images Marketers, content creators, non-designers
Stable Diffusion (SDXL) Open-source / local Fine-tuning on custom data; full control, no censorship Developers, researchers, technical artists
Adobe Firefly Creative Cloud subscription IP-safe outputs; Photoshop/Illustrator integration Commercial designers, ad agencies, brand teams
How These Tools Are Actually Being Used

In 2023, game studio Riot Games publicly disclosed it was exploring AI-generated concept art for early ideation phases, using it to produce rough directional images before human artists refined the final assets. The company's creative director noted that the tools cut early concepting time significantly but that human artists remained essential for character-defining work.

Meanwhile, the advertising industry moved quickly. Coca-Cola's "Masterpiece" campaign (2023) used AI image tools alongside traditional CG to animate classic paintings, generating considerable press coverage about what AI-assisted commercial art looks like in practice. The campaign explicitly credited human directors and studios — a disclosure approach that became a template others followed.

On the independent side, illustrator Refik Anadol and his studio used large-scale image model outputs as raw material for data-sculpture installations at MoMA in 2023, demonstrating that AI image tools can function as one layer in a multi-step artistic process rather than as end-to-end production systems.

Key Insight

The tool you choose shapes your creative workflow more than the prompt you write. Midjourney rewards aesthetic intuition; Stable Diffusion rewards technical setup; Firefly rewards workflow integration; DALL·E 3 rewards clear prose description. Match tool to task before writing a single word.

Key Terms
Diffusion modelA generative architecture that learns to reverse a noise-adding process. Given a text prompt, it gradually de-noises random pixels into a coherent image. Stable Diffusion and DALL·E 3 both use diffusion architectures.
LoRA (Low-Rank Adaptation)A fine-tuning technique that lets users train a small supplementary model on their own images and attach it to a base Stable Diffusion model to produce outputs in a specific style or featuring specific subjects.
Commercially safe training dataTraining data sourced from licensed, public domain, or consensually contributed images — Adobe Firefly's core differentiator for professional use cases.
Module 4 · Quiz 1

The Landscape: Major AI Image Tools

5 questions — select the best answer for each
1. Which AI image tool was specifically trained on licensed Adobe Stock, openly licensed, and public domain content to address commercial IP concerns?
Correct. Adobe Firefly was explicitly designed around "commercially safe" training data sourced from Adobe Stock, licensed works, and public domain content.
Not quite. Adobe Firefly was the tool built specifically to address commercial IP liability concerns through its training data sourcing.
2. Greg Rutkowski's name becoming a top Midjourney prompt in 2022 illustrated which core concern about AI image tools?
Correct. Rutkowski's case became a key example of how AI tools trained on artists' work could be used to generate content in their style without consent or payment.
Incorrect. The concern was specifically about replicating artistic style without consent or compensation — a core IP and ethics issue.
3. What architectural decision made Stable Diffusion the foundation for thousands of derivative tools?
Correct. Stability AI's decision to release weights publicly meant anyone could download, modify, and build on the model, enabling the vast ecosystem of fine-tunes and derivative tools.
Incorrect. It was the open-source release of model weights that enabled the derivative ecosystem around Stable Diffusion.
4. DALL·E 3's primary product differentiator versus Midjourney was its:
Correct. OpenAI built DALL·E 3 into ChatGPT specifically so users could describe images in natural prose without prompt engineering, the tool's main differentiator.
Not quite. DALL·E 3's key differentiator was its deep ChatGPT integration and strong instruction-following capability.
5. Refik Anadol's 2023 MoMA installation demonstrated which approach to AI image tools?
Correct. Anadol's studio used large-scale model outputs as one layer in data-sculpture installations — a multi-step process in which AI was a material, not a replacement for artistic vision.
Incorrect. Anadol used AI outputs as raw material within a larger artistic process, not as an end-to-end production system.
Module 4 · Lab 1

Tool Selection Practice

Use the AI assistant to work through real tool-choice scenarios

Lab Goal

Given a professional creative scenario, you'll practice selecting the most appropriate AI image tool and justifying your choice based on the tool's specific strengths, training data policies, and access model.

Have a conversation with the assistant about at least three different creative scenarios. Ask it to help you reason through which tool fits each use case and why.

Starter prompt: "I'm a freelance graphic designer and a client wants to use AI-generated images in a national print ad campaign. Which tool should I recommend and why?"
AI Tool Selector — Lab 1
Image Tools Expert
Welcome to Lab 1. I'm here to help you think through AI image tool selection for real creative scenarios. Tell me about a project — the medium, the client type, the IP concerns, and what the image needs to do — and we'll work out which tool fits best and why.
Module 4 · Lesson 2

Prompt Engineering for Images

The specific language structures that produce consistent, professional results across different tools
Why does "a painting of a city at night" produce wildly different results than a professional prompt — and what exactly is the difference?

When Midjourney released v5 in March 2023, its community of roughly 13 million Discord members immediately began documenting which prompt structures produced the most reliable outputs. The unofficial "Midjourney Prompt Craft" guides that circulated within days identified a consistent pattern: prompts that specified a medium, a lighting condition, an aspect ratio, and a reference aesthetic consistently outperformed vague descriptive prompts by a wide margin in community voting on the platform's #showcase channels. The pattern held across v5, v5.1, v5.2, and v6 — the structural logic of good image prompts proved more stable than the model versions themselves.

The Anatomy of an Effective Image Prompt

Research by practitioners and platform documentation consistently identifies the same core components for high-quality outputs. Understanding each component gives you a repeatable system rather than a guessing game.

Subject + Action: The primary content. Be specific — "a middle-aged woman reviewing architectural blueprints" outperforms "a woman working."

Medium / Style: The visual language. "Oil painting," "editorial photography," "vector illustration," "gouache on paper." This single component often has the largest effect on output quality because it anchors the model in a specific visual tradition.

Lighting: "Golden hour backlight," "overcast studio lighting," "neon-lit night scene," "harsh midday sun." Lighting descriptions are read almost literally by diffusion models and have a large effect on mood and realism.

Composition / Framing: "Wide establishing shot," "close-up portrait," "bird's-eye view," "Dutch angle." These cinematographic terms are reliable because models were trained on massive corpora of captioned photography and film stills.

Technical / Quality Modifiers: Platform-specific boosters: "--v 6 --ar 16:9" in Midjourney; "photorealistic, 8k, sharp focus" in Stable Diffusion; "high detail, professional" in DALL·E. These don't always do what users believe, but some demonstrably shift outputs toward higher-fidelity results.

Prompt Comparison: Vague vs. Structured
Prompt TypeExampleTypical Issues
Vague "A city at night" Generic skyline, flat lighting, unclear style, low cohesion
Structured "Tokyo neon district, rain-slicked streets, editorial photography, wide angle, shallow depth of field, f/1.8 bokeh, overcast night —ar 16:9 —v 6" Minimal — specific enough to avoid most ambiguity
Style-anchored "Isometric illustration, cozy coffee shop interior, warm amber lighting, vector flat art, clean lines, Dribbble aesthetic" May over-lean into one aesthetic; use for deliberate style lock
Negative Prompting and Exclusion

Most tools accept some form of negative guidance — telling the model what to exclude. In Stable Diffusion this is a dedicated "negative prompt" field. In Midjourney it is the --no parameter. In DALL·E 3 it works through natural language: "do not include any text or watermarks."

Effective negative prompts address the model's common failure modes: "no extra fingers, no distorted faces, no text artifacts, no chromatic aberration" addresses typical anatomical and photographic failures. "no over-saturation, no lens flare" addresses stylistic defaults the model would otherwise apply.

In 2023, concept artist Karla Ortiz — one of the illustrators who filed a class action lawsuit against Stability AI and Midjourney — publicly discussed how she had studied the models' failure patterns specifically to understand what training data biases they revealed. Her analysis showed that models defaulted to specific "Artstation-ized" aesthetics unless explicitly prompted away from them, confirming that negative prompting is a form of bias correction as much as quality control.

Practical Rule

Write your image prompt in this order: Subject → Medium → Lighting → Composition → Quality modifiers → Negatives. This sequence mirrors how human art directors brief photographers and illustrators — and diffusion models appear to have internalized that same briefing logic from their training data.

Iteration Strategy: Seed Locking and Variation

Professional use of AI image tools relies heavily on iteration. In Midjourney, the seed parameter (--seed [number]) locks the random starting noise so you can make incremental prompt changes while preserving the underlying composition. In Stable Diffusion, seed control is even more granular — you can run hundreds of variations on a single seed to find the best combination of prompt and randomness.

The practical workflow used by many studios: generate a grid of four to eight variations at low quality settings, select the best composition seed, then re-run that seed at full quality with refined prompts. This approach was described publicly by design studio Ueno in their 2023 writeup on integrating AI tools into their client workflow — they reported cutting initial concepting time by roughly 60% while still producing outputs that required significant human art direction.

Key Terms
Seed valueA number that initializes the random noise a diffusion model starts from. Using the same seed with the same prompt produces identical outputs; changing only the prompt while keeping the seed produces variations of the same underlying composition.
Negative promptText describing content, styles, or artifacts to avoid in the output. Functions as a bias-correction and quality-control mechanism.
CFG scale (Classifier-Free Guidance)A Stable Diffusion parameter controlling how strictly the model follows the prompt versus how creative it is. High CFG = rigid adherence; low CFG = loose interpretation. Sweet spot for most professional work is 7–9.
Module 4 · Quiz 2

Prompt Engineering for Images

5 questions — select the best answer for each
1. Which prompt component typically has the largest single effect on an AI image's overall look and visual language?
Correct. Medium and style specifications anchor the model in a specific visual tradition and typically have the largest single effect on the overall look of an output.
Not quite. Medium and style specifications — "oil painting," "editorial photography," etc. — typically have the largest effect on overall visual language.
2. In Stable Diffusion, CFG scale controls what?
Correct. CFG (Classifier-Free Guidance) scale is a slider between strict prompt adherence and creative freedom. Higher values produce outputs that literally interpret the prompt; lower values allow more model creativity.
Incorrect. CFG scale controls the balance between following the prompt strictly and allowing creative deviation — not resolution or step count.
3. What does "seed locking" allow a professional to do during iterative image generation?
Correct. Using a fixed seed number means the model starts from the same random noise each time, so prompt changes produce variations of the same underlying composition rather than entirely new images.
Incorrect. Seed locking preserves the underlying compositional structure so you can refine prompts incrementally without losing a composition you like.
4. Artist Karla Ortiz's analysis of AI image models' default aesthetics revealed that negative prompting functions as:
Correct. Ortiz's analysis showed models defaulted to "Artstation-ized" aesthetics from training data biases — meaning negative prompts that push away those defaults are correcting a bias, not just improving quality.
Not quite. Ortiz's analysis showed that negative prompting corrects training data biases — for example, the model's default toward a specific "Artstation aesthetic" — making it a bias-correction tool as much as a quality tool.
5. Design studio Ueno's 2023 report on AI tool integration found that the main productivity gain came from:
Correct. Ueno reported approximately 60% faster initial concepting, but noted that human art direction was still required to produce outputs suitable for clients.
Incorrect. Ueno reported faster concepting — roughly 60% — while emphasizing that human art direction remained essential throughout the process.
Module 4 · Lab 2

Prompt Structure Workshop

Build and refine structured image prompts with expert feedback

Lab Goal

Practice building properly structured image prompts using the Subject → Medium → Lighting → Composition → Modifiers → Negatives framework. The assistant will critique your prompts and help you improve them.

Share at least three prompt attempts — starting vague and progressively more structured — and ask for analysis of each component.

Starter prompt: "Here's my first attempt: 'a forest in autumn.' How do I make this into a professional image prompt for Midjourney?"
Prompt Engineer — Lab 2
Prompt Critique Mode
Welcome to Lab 2. Share any image prompt — even a rough one — and I'll break it down component by component: what's working, what's missing, and how to restructure it for more consistent, professional results. We can work through Midjourney, DALL·E 3, or Stable Diffusion syntax specifically.
Module 4 · Lesson 3

Style, Ownership, and Ethics

The real legal and ethical landscape surrounding AI-generated visual art in 2024
When a machine learns from a million human paintings, who owns what it creates — and does the question even have a clean answer?

In February 2023, the U.S. Copyright Office issued a decision in the case of Kristina Kashtanova's graphic novel Zarya of the Dawn — the first major U.S. ruling on AI-generated visual art copyright. The Office cancelled the copyright on individual images generated by Midjourney within the work while preserving copyright on the text and on the selection and arrangement of the images as a human creative act.

The ruling established a principle the Office would reinforce throughout 2023: copyright requires human authorship, and AI-generated images — absent sufficient human creative control over their specific expression — do not qualify. The debate it opened has not closed.

The Copyright Office's Evolving Position

Throughout 2023, the U.S. Copyright Office received more than 10,000 AI-related submissions during a public comment period and issued guidance documents clarifying its developing framework. The core principle: human authorship is required for copyright, and simply providing a text prompt to an AI system does not constitute sufficient creative control over the specific expression in the resulting image.

However, the Office also acknowledged a spectrum. Cases where a human extensively modifies an AI-generated image — using Photoshop, compositing, manual retouching, addition of hand-drawn elements — may qualify for protection on those human-authored portions. The line is fact-specific and will take years of case law to clarify.

In March 2023, the Office separately denied copyright protection to a series of images generated by the autonomous system "DABUS" — reinforcing that the authorship question centers on human creative agency, not just the presence of a human somewhere in the process.

The Training Data Lawsuits

Three class action lawsuits filed between November 2022 and January 2023 framed the other half of the legal debate: whether training AI models on copyrighted artworks constitutes infringement.

Andersen v. Stability AI et al. (filed January 2023) brought by artists Sarah Andersen, Kelly McKernan, and Karla Ortiz against Stability AI, Midjourney, and DeviantArt, alleged that scraping billions of web images for training without consent or compensation violated copyright. As of 2024, the case survived a motion to dismiss on direct infringement claims and was proceeding through discovery.

Getty Images v. Stability AI (filed January 2023 in the UK, February 2023 in Delaware) alleged that Stability AI copied approximately 12 million Getty images without license. Getty presented evidence that Stable Diffusion outputs sometimes contained visible watermark artifacts resembling Getty's watermark — a striking concrete demonstration of memorization in training data.

These cases have not yet produced final verdicts, but they have already changed behavior: Adobe built Firefly on licensed data precisely because these lawsuits made "we scraped the web" an untenable position for enterprise customers.

Practical Implication

Until the law settles, professional practice means choosing tools with commercially licensed training data (Firefly, or tools using explicitly licensed datasets) for any work where IP indemnification matters. For personal, experimental, or editorial work, the legal risk is lower but the ethical question of compensating source artists remains unresolved.

The Style Prompt Question

Legally, style is not protected by copyright in the United States — you cannot copyright "painting in a loose impressionist style." But the AI case introduces a nuance: prompting a model with a living artist's name may invoke not just a general style but specific compositional and color choices that appear in that artist's actual works, potentially crossing from style influence into reproduction.

In response to community pressure, DeviantArt launched an opt-out system called "NoAI" in late 2022 that lets artists tag their work to signal they don't want it used in AI training. Spawning.ai launched "Have I Been Trained?" — a tool that lets artists check if their work appeared in LAION-5B, the dataset used to train Stable Diffusion — and request removal. These opt-out systems are widely criticized as placing the burden on artists rather than on AI companies, but they represent the first practical tools addressing the consent gap.

Ethical Framework for Practitioners

Three practical questions for any AI-generated visual work: (1) Does the tool I'm using have a defensible training data provenance? (2) Am I generating in a specific living artist's style — and if so, does my use harm their market? (3) Would I be comfortable disclosing that this work was AI-assisted to the audience, client, or publication? Affirmative answers to all three represent the current professional standard.

Key Terms
Human authorship requirementThe U.S. Copyright Office's standard that copyright protection requires sufficient human creative control over the specific expression being protected — not met by text-prompting alone under current guidance.
LAION-5BA dataset of approximately 5 billion image-text pairs scraped from the web, used to train Stable Diffusion and other models. Subject to ongoing legal scrutiny regarding the copyright status of the images it contains.
Opt-out systemA mechanism by which artists can flag their work as not available for AI training (e.g., DeviantArt's NoAI tag, Spawning.ai's removal request tool). Currently non-binding and reliant on voluntary compliance.
Module 4 · Quiz 3

Style, Ownership, and Ethics

5 questions — select the best answer for each
1. In its February 2023 ruling on Zarya of the Dawn, what did the U.S. Copyright Office protect?
Correct. The Office cancelled copyright on the individual Midjourney-generated images while preserving protection for the human-authored text and the creative selection and arrangement of the images.
Incorrect. The Office protected the human-authored text and the selection/arrangement of images, while cancelling copyright on the individual AI-generated images themselves.
2. What did visible watermark artifacts in Stable Diffusion outputs in the Getty Images lawsuit demonstrate?
Correct. The appearance of Getty-style watermark artifacts was striking evidence that the model had memorized specific features of training images — a concrete demonstration of data memorization in generative models.
Incorrect. The watermark artifacts demonstrated memorization — that specific visual features of training data images were being reproduced in model outputs.
3. Under current U.S. Copyright Office guidance, what would most likely qualify an AI-assisted image for copyright protection?
Correct. The Office has indicated that human authorship applied on top of AI outputs — such as extensive manual modification, addition of hand-drawn elements, or significant retouching — may qualify the human-authored portions for protection.
Incorrect. Current guidance indicates that extensive human modification of AI outputs — adding hand-drawn elements, significant manual retouching — is more likely to qualify for protection than prompt writing alone.
4. What is the primary criticism of opt-out systems like DeviantArt's NoAI tag?
Correct. Opt-out systems are widely criticized for inverting the burden — requiring artists to actively protect their work rather than requiring AI companies to obtain consent before training on it.
Incorrect. The primary criticism is that opt-out systems place the protection burden on artists, who must actively flag work rather than having companies obtain consent proactively.
5. Which of the three ethical framework questions from this lesson asks about audience relationship?
Correct. The disclosure question addresses your relationship with audiences and clients — transparency about AI assistance is increasingly part of professional and editorial ethics standards.
Incorrect. The audience-relationship question is specifically: "Would I be comfortable disclosing AI assistance to the audience, client, or publication?" — addressing transparency and professional ethics.
Module 4 · Lab 3

Ethics Case Analysis

Work through real-world AI art ethics scenarios with guided reasoning

Lab Goal

Practice applying the three-question ethical framework to real AI visual art scenarios. The assistant will present cases drawn from documented 2022–2024 situations and help you reason through the ownership, attribution, and disclosure dimensions.

Work through at least three different scenarios — push back on the assistant's reasoning, ask about edge cases, and explore where the ethical framework breaks down.

Starter prompt: "A magazine wants to use a Midjourney image for an editorial illustration. The art director wrote a detailed prompt but didn't tell the writer or readers. Walk me through the ethical considerations using the three-question framework."
Ethics Analyst — Lab 3
AI Art Ethics
Welcome to Lab 3. I'll help you work through AI visual art ethics using the framework from this lesson: training data provenance, style and market harm, and disclosure transparency. Bring me a scenario — real or hypothetical — and we'll reason through all three dimensions together. I'll also point out where the framework has genuine ambiguities.
Module 4 · Lesson 4

Integrating AI Art Into Your Practice

Real workflows, real studios, and the creative roles that AI tools are changing — and not changing
What does it actually look like when a working creative professional absorbs AI image tools into their daily practice without losing their voice?

In a 2023 GDC (Game Developers Conference) session, a concept artist from a mid-size independent studio described their team's workflow shift after adopting AI image tools for early pre-production. Previously, producing a "visual target" — a single polished image showing the intended look of a game world — took three to five days of iteration with one artist. With AI-assisted concepting, the same team could produce twenty directional images in a morning, then spend two days refining the two or three that resonated with the creative director. The ratio of exploration to refinement had inverted.

The same artist noted that their ability to evaluate AI outputs quickly — to recognize which images were compositionally strong, which had interesting lighting, which matched the game's tone — was the scarce skill. Taste and visual judgment had become more valuable, not less.

Where AI Image Tools Add Value in Real Workflows

Based on documented studio reports and practitioner case studies from 2022–2024, AI image tools provide the most consistent value in four specific workflow phases:

1. Early concepting and mood boarding. Generating directional images to communicate a visual language before committing human art time. Used by film studios for set design direction, game studios for world-building, and brand teams for campaign direction.

2. Reference image generation. Creating custom reference images that don't exist as stock photography — unusual lighting setups, specific architectural styles, rare natural conditions — for human artists to work from. This eliminates hours of reference hunting while giving artists something precisely matched to their need.

3. Rapid variation generation. Once a strong design exists, generating color, texture, and compositional variations for client review. Traditionally a time-consuming manual process; AI tools reduce it to minutes.

4. Texture and pattern work. Generating seamless textures, surface patterns, and background elements — particularly useful in game development and product design where these assets are needed in volume.

What AI Image Tools Still Don't Do Well

Honest assessment of limitations is as important as understanding capabilities. As of 2024, AI image tools consistently struggle with:

Consistent character design. Generating a specific character that looks the same across multiple images — same face, same proportions, same costume details — remains unreliable without LoRA fine-tuning or heavy post-processing. This is why film and game production still relies on human character artists for hero assets.

Accurate hands and complex anatomy. A well-documented limitation that has improved across model versions but not been solved. Professional use typically involves inpainting or manual correction of anatomical errors.

Legible text within images. DALL·E 3 has made progress here, but text generation remains unreliable enough that any design requiring specific legible text in an image needs human execution.

Brand-specific visual consistency. Producing images that match a specific brand's established visual language without fine-tuning requires extensive prompt engineering and still produces inconsistent results.

The Studios That Got It Right

In 2023, Publicis Groupe (one of the world's largest advertising holding companies) released an internal AI policy establishing that AI-generated images could be used in client work only when: (a) the client was informed and gave explicit consent; (b) the tool used had a defensible training data provenance; and (c) a human creative director had reviewed and approved all outputs. This policy became a template for agency AI governance documents across the industry.

Independent illustrator Sam Yang, who has 1.2 million Instagram followers for his character illustration work, published a detailed 2023 blog post documenting how he incorporated Stable Diffusion into his workflow — specifically using it for background environment concepting while all character work remained fully hand-drawn. He noted that his followers responded positively once he explained the hybrid process, suggesting that transparency about AI's role resolves most audience trust concerns.

At the other end of the scale, The New York Times issued a policy in 2023 prohibiting AI-generated images in its journalism while permitting editorial illustrations that clearly identify AI tools as one element of a human-directed creative process — a distinction that separates journalistic documentation from artistic commentary.

Integration Principle

The professionals adapting most successfully to AI image tools are not the ones using AI to replace their creative process — they're the ones using it to extend the exploration phase of that process. More iterations, more directions, more "what if" images early on — then their own judgment and craft to select and refine. AI expands the front end of the creative funnel; it hasn't replaced what happens at the back.

Building Your Own Integration Protocol

A practical integration protocol answers five questions before you start any project involving AI-generated visuals: (1) What phase of the work is AI appropriate for? (2) Which tool fits this use case's IP requirements? (3) What will I do with the outputs that requires human skill? (4) How will I disclose AI involvement to clients or audiences? (5) What would I do if the AI tool became unavailable — can the work stand without it?

That fifth question is more than a contingency plan. It forces clarity about whether AI is genuinely serving the creative work or whether dependency on a specific tool's aesthetic has started substituting for artistic development.

Key Terms
Visual targetA single polished "north star" image used in pre-production to align a team around the intended look of a project. AI tools have significantly accelerated visual target iteration in game and film production.
InpaintingA technique where a mask is drawn over a portion of an AI-generated image and the model is asked to regenerate only that region — used for correcting anatomical errors, replacing backgrounds, or adding details.
Hybrid workflowA creative process that deliberately assigns different tasks to AI tools and human execution based on each one's strengths — for example, AI for environment concepting and human execution for character design.
Module 4 · Quiz 4

Integrating AI Art Into Your Practice

5 questions — select the best answer for each
1. According to the GDC 2023 session case study, what happened to the ratio of exploration to refinement after AI tools were adopted?
Correct. The concept artist described an inverted ratio — where previously a visual target took 3–5 days of single-artist iteration, AI enabled 20 directional images in a morning, followed by two days refining the best ones.
Incorrect. The GDC case study described the ratio inverting — far more exploration images generated quickly, then focused human refinement on the few that resonated.
2. Which of the following is a documented ongoing limitation of AI image tools as of 2024?
Correct. Consistent character design — same face, proportions, and costume details across multiple images — remains a documented limitation requiring LoRA fine-tuning or manual correction to address.
Incorrect. A well-documented limitation is character consistency — generating the same specific character reliably across multiple images remains unreliable without fine-tuning.
3. Publicis Groupe's 2023 internal AI policy required which three conditions for AI-generated images in client work?
Correct. Publicis required: (a) explicit client consent, (b) defensible training data provenance, and (c) human creative director review of all outputs — a three-part governance framework adopted widely as a template.
Incorrect. Publicis required: client informed consent, defensible training data provenance, and human creative director review — a specific three-part standard that became an industry template.
4. Illustrator Sam Yang's documented approach to hybrid AI workflow was:
Correct. Yang used Stable Diffusion specifically for background environment concepting while maintaining fully hand-drawn character work — and reported his audience responded positively when he explained the hybrid process transparently.
Incorrect. Yang's hybrid approach used Stable Diffusion for background environments while all character illustration remained hand-drawn — and he disclosed this process to his audience.
5. What is the purpose of the fifth integration protocol question — "What would you do if the AI tool became unavailable?"
Correct. This question is a diagnostic — it tests whether your work stands on its own creative foundations or whether dependency on a specific tool's aesthetic has started replacing your own artistic development.
Incorrect. The fifth question is a creative health check — it forces you to examine whether AI dependency has started substituting for genuine artistic development, not just a contingency planning step.
Module 4 · Lab 4

Build Your Integration Protocol

Design a personalized AI visual art workflow for your specific creative practice

Lab Goal

Work with the assistant to design your own AI image tool integration protocol tailored to your actual creative work — your medium, your client or audience relationships, your IP context, and your goals. Apply all four lessons: tool selection, prompt strategy, ethics, and workflow integration.

Describe your creative practice and the assistant will help you build a custom five-question integration protocol and tool stack recommendation.

Starter prompt: "I'm a freelance illustrator who does children's book work and occasional brand licensing. Help me build an integration protocol for AI image tools that fits my specific context."
Workflow Architect — Lab 4
Integration Design
Welcome to Lab 4 — the synthesis lab. Tell me about your creative practice: your medium, your typical clients or audiences, whether IP protection matters for your work, and what phase of your process you most want to accelerate or improve. I'll help you build a customized integration protocol using everything from this module: the right tool for your context, a prompt strategy, an ethics framework, and a workflow design that keeps your creative voice central.
Module 4 · Final Assessment

Module Test: Visual Art and AI

15 questions — 80% required to pass · covers all four lessons
1. Which AI image tool launched via Discord in April 2022 and was the most-cited platform in professional design contexts by 2023?
Correct. Midjourney launched in April 2022 via Discord and became the most-cited AI image platform in professional design contexts, with roughly one million daily active users by mid-2023.
Incorrect. Midjourney launched via Discord in April 2022 and was the leading tool in professional design circles by 2023.
2. What training data policy most directly distinguishes Adobe Firefly from Midjourney and Stable Diffusion?
Correct. Adobe's "commercially safe" training data claim — licensed Adobe Stock, openly licensed works, and public domain content — is Firefly's core differentiator for enterprise and commercial use cases.
Incorrect. Firefly's distinction is its training on licensed Adobe Stock, openly licensed content, and public domain works — the basis for its "commercially safe" positioning.
3. A LoRA (Low-Rank Adaptation) in the Stable Diffusion ecosystem is:
Correct. LoRA is a fine-tuning technique enabling users to train a small supplementary model on their own images and attach it to a base Stable Diffusion model for style-consistent or subject-consistent outputs.
Incorrect. LoRA is a fine-tuning technique — a small supplementary model trained on custom images and attached to a base Stable Diffusion model.
4. In professional image prompt structure, what should typically come immediately after specifying the subject?
Correct. The recommended sequence is Subject → Medium/Style → Lighting → Composition → Quality modifiers → Negatives. Medium/style comes second because it has the largest single effect on overall visual language.
Incorrect. The recommended sequence places medium/style immediately after the subject — before lighting, composition, and technical modifiers.
5. Stable Diffusion's CFG scale at very high values (e.g., 15–20) typically produces:
Correct. Very high CFG values force rigid prompt adherence, which typically produces over-saturated, over-sharpened, or artifacts-heavy images. The sweet spot for professional work is generally 7–9.
Incorrect. High CFG values produce rigid adherence to the prompt — typically over-saturated or artifact-heavy. Lower values (7–9) are the professional sweet spot.
6. The U.S. Copyright Office's February 2023 Zarya of the Dawn ruling established that:
Correct. The Office cancelled copyright on individual Midjourney-generated images while protecting the human-authored text and creative selection/arrangement — establishing that human authorship over specific expression is required.
Incorrect. The ruling cancelled copyright on individual AI-generated images while preserving protection for human-authored text and the creative selection and arrangement of images.
7. In Getty Images v. Stability AI, what was the significance of watermark artifacts appearing in model outputs?
Correct. Watermark artifacts were striking evidence of training data memorization — specific visual features from Getty images being reproduced in model outputs, central to Getty's infringement claim.
Incorrect. The watermark artifacts evidenced training data memorization — specific features of Getty images being reproduced in outputs.
8. LAION-5B is:
Correct. LAION-5B is a web-scraped dataset of approximately 5 billion image-text pairs used to train Stable Diffusion — and is at the center of ongoing legal and ethical scrutiny.
Incorrect. LAION-5B is the ~5 billion image-text pair web-scraped dataset used to train Stable Diffusion, and subject to ongoing legal challenges.
9. The Andersen v. Stability AI et al. lawsuit was filed by artists alleging:
Correct. Andersen, McKernan, and Ortiz sued Stability AI, Midjourney, and DeviantArt alleging that scraping billions of web images without consent or compensation for training violated their copyrights.
Incorrect. The lawsuit alleged copyright infringement through scraping artist work for training without consent or compensation.
10. Which of the following is NOT one of the four workflow phases where AI image tools provide the most documented value?
Correct. Final hero character design remains a documented AI limitation — consistent character appearance across images requires extensive fine-tuning. The four documented value phases are concepting, reference generation, variation generation, and texture/pattern work.
Incorrect. Final hero character design is a documented AI weakness, not a strength. The tool still struggles with consistent character appearance across multiple images.
11. Inpainting in AI image workflows is used for:
Correct. Inpainting involves masking a region of an existing generated image and asking the model to regenerate only that area — used primarily for anatomical corrections and detail additions.
Incorrect. Inpainting is the process of masking a specific region in an existing image and regenerating only that area — key for correcting anatomical errors.
12. The New York Times' 2023 AI image policy distinguished between:
Correct. The Times prohibited AI-generated images in journalism while permitting editorial illustrations where AI's role was clearly disclosed as one element of a human-directed creative process.
Incorrect. The Times' policy prohibited AI images in journalism while allowing editorial illustrations that identified AI as one disclosed element in a human-directed creative process.
13. The DeviantArt "NoAI" tag and the Spawning.ai "Have I Been Trained?" tool are both examples of:
Correct. Both are opt-out systems — they require artists to actively flag or request removal of their work rather than requiring AI companies to obtain consent before training on it. This burden-placement is their primary criticism.
Incorrect. Both are opt-out systems that require artists to take action — the primary criticism is that this places the protection burden on artists rather than on the companies using their work.
14. DALL·E 3's policy of not generating images "in the style of living artists by name" was a direct response to:
Correct. DALL·E 3's living-artist policy was a direct response to the Rutkowski controversy — his name being used thousands of times daily in Midjourney prompts to invoke his oil-painting style without consent.
Incorrect. The policy was a direct response to the Rutkowski situation — his name becoming a top Midjourney prompt, invoking his style without consent or compensation.
15. According to the module's integration principle, how are the most successful creative professionals using AI image tools?
Correct. The integration principle states: AI expands the front end of the creative funnel — more iterations, more "what if" directions — while human judgment and craft remain essential at the selection and refinement stage.
Incorrect. The integration principle is: AI expands exploration at the front of the creative funnel; human judgment and craft are what select and refine at the back. AI serves the process, not replaces it.