Course info

The following reference materials were provided as context. Use them as guidance only -- always prefer the latest, most current information over any potentially outdated content in these files. --- Reference: AI_Image_Generation_Course_Lesson_Scripts.pdf --- ## AI Image Generation: Introductory Course Full Lesson Scripts -- All 10 Modules Each module below is written as a ready-to-teach script: an opening hook to earn attention, clear learning objectives, a detailed content breakdown, a hands-on activity, common mistakes to watch for, and a bridge line into the next module. Module 1: How Generative Image AI Works Opening Hook Ever watch a Polaroid develop in your hand? This is that, in reverse and turned up to eleven: start with pure TV static, then have a machine slowly “un-static” it until a photograph appears out of nowhere. That’s diffusion -- and it’s the beating heart of every AI image tool your students will touch in this course. Learning Objectives • Explain in plain language what a diffusion model does. • Understand why the same prompt produces different results every time. • Name three or four major tools and recognize they share the same underlying process. Lesson Breakdown • The “sculpture from noise” analogy: Michelangelo said he just removed everything that wasn’t David. A diffusion model does the same thing to noise, guided by the prompt as its target shape. • Walk through the loop in plain terms: random noise, the model predicts what to remove, repeat over many steps, a coherent image emerges. • Why results vary: randomness (a “seed”) is baked in from the very first step, like a fingerprint on that specific run. • Quick tool landscape tour: Midjourney (artistic community-driven), DALL-E (built into ChatGPT), Adobe Firefly (commercial-safe, Creative Cloud-integrated), Stable Diffusion (open-source, highly customizable) -- all diffusion-based under the hood. Hands-On Activity Live demo: generate the exact same prompt three times back to back. Have learners spot the differences between the three results and guess why before you explain seeds. Common Mistakes to Flag • Assuming the AI “understands” an image the way a person does. • Expecting identical output every time from an identical prompt. Wrap-Up & Bridge Now that you know what’s happening under the hood, let’s go open one of these tools and actually drive it. ## Module 2: Navigating a Generation Tool Opening Hook Every generative tool looks like a spaceship cockpit the first time you open it. Good news: they’re all flying the same basic ship. Learning Objectives • Identify the five core interface elements shared across nearly every tool. • Generate a first image start to finish. • Understand what “variations,” “upscale,” and “aspect ratio” controls actually do. Lesson Breakdown • Anatomy of a generation tool: the prompt box, the generate button, the output gallery, variation/re-roll controls, and the settings panel (aspect ratio, resolution, style strength). • Live screen-share tour of one tool, naming each element and its equivalent in two other tools so the pattern transfers. • The “first generation” ritual: type a prompt, hit generate, wait, review. • Why version numbers matter: a newer model version doesn’t behave identically to the last one, so note which version you’re using. Hands-On Activity Learners open a tool of their choice and produce their first image within five minutes, then drop a screenshot in the shared chat or board. Common Mistakes to Flag • Over-adjusting settings before understanding what the defaults already do well. • Ignoring a queued or processing state and spamming the generate button. Wrap-Up & Bridge You can now drive the car. Time to learn what to actually say to it. ## Module 3: Prompt Writing Fundamentals Opening Hook A prompt is an order ticket, not a wish. The clearer the ticket, the closer the kitchen gets it right on the first pass. Learning Objectives • Build a prompt using a clear, repeatable structure. • Write a negative prompt to exclude unwanted elements. • Iterate a prompt from vague to specific and see the results shift. Lesson Breakdown • The six-part anatomy: subject, action or pose, setting, style or medium, lighting, and mood -- with a quick example for each. • Camera and lens language as a shortcut: “85mm portrait lens, shallow depth of field” communicates a lot to the model very efficiently. • Positive vs. negative prompting: what to state directly vs. what to explicitly exclude. • The “onion” method: start with one clear sentence, then layer in modifiers one at a time, checking the shift after each layer. Hands-On Activity Group exercise: everyone starts from the same three-word prompt (“a coffee shop”) and layers in three modifiers per round. Compare results after each round as a group. Common Mistakes to Flag • Cramming twenty adjectives into one run-on sentence. • Contradicting yourself -- for example, asking for “minimalist” and “highly detailed baroque” in the same prompt. Wrap-Up & Bridge Now you can talk to the machine. Next: what to do when it almost gets it right. ## Module 4: Iteration and Control Opening Hook Nobody’s first draft is the final draft -- not novelists, not painters, and definitely not you and your AI co-pilot. Learning Objectives • Use variations and re-rolls with intention rather than at random. • Use seeds to hold a look steady while changing details. • Apply basic inpainting and outpainting. • Use a reference image to steer style without a wordier prompt. Lesson Breakdown • The single most important iteration habit: change one variable at a time. • Seeds explained simply: a seed is like a save file for the randomness -- reuse it to keep a look consistent while you adjust other details. • Inpainting fixes a small area; outpainting expands the canvas past the original borders. • Reference images: uploading a mood-board photo to steer color and style faster than describing it in words. Hands-On Activity Take one “pretty good” generated image and run three targeted iterations -- change only the lighting, change only the background, fix only one flawed detail -- then compare all four versions side by side. Common Mistakes to Flag • Regenerating from scratch instead of iterating on something that was already close. • Changing five things at once and losing track of what actually worked. Wrap-Up & Bridge You’ve got a great image. Now, do you actually know if it’s good? ## Module 5: Visual Design Basics for Evaluating Output Opening Hook You don’t need an art degree to catch a bad photo -- but you do need to know where to look. Learning Objectives • Apply the rule of thirds to judge composition. • Identify a clear focal point. • Recognize basic lighting and color quality. • Describe visual hierarchy in plain terms. Lesson Breakdown • Rule of thirds: the classic tic-tac-toe grid overlay and why the intersections matter more than dead center. • Focal point: what draws the eye first, and why an image without one feels flat no matter how detailed it is. • Lighting basics: direction, hardness vs. softness, and color temperature -- what “good” actually looks like. • Visual hierarchy: what the viewer should notice first, second, and third. Hands-On Activity Show four generated images -- two strong, two weak. In pairs, learners rank them by composition quality and must explain why using today’s vocabulary, not “I just like it better.” Common Mistakes to Flag • Judging an image only on subject matter and ignoring composition entirely. • Assuming perfect symmetry always means “good.” Wrap-Up & Bridge You can now judge good bones. Next, let’s train your eye to catch what’s actually broken. ## Module 6: Spotting AI Artifacts and Quality Issues Opening Hook AI is a brilliant art forger with one tell: count the fingers. Learning Objectives • Identify the most common AI failure modes. • Decide when to regenerate versus manually fix. • Build a personal artifact-checking checklist. Lesson Breakdown • The greatest hits of AI mistakes: hands and fingers, garbled text, warped small objects, inconsistent shadows or reflections, asymmetric earrings or eyes, and melting backgrounds. • Why these happen, in one plain sentence: the model is a pattern-matcher, not an anatomist. • The ten-second scan technique, in order every time: hands, face, background edges, text, shadows. • A simple decision tree: a minor flaw in an easy zone gets an inpaint fix; a major structural flaw gets a full regenerate. Hands-On Activity “Spot the Flaw” speed round: flash six images for five seconds each and have learners call out what’s wrong before the timer runs out. Common Mistakes to Flag • Not checking backgrounds and edges because attention stays locked on the main subject. • Over-trusting an image that looks “good enough” on a small screen when it’s destined for a large print, where flaws show up bigger. Wrap-Up & Bridge You’ve got the eye of a quality inspector. Now let’s get the file into the right shape for wherever it’s actually going. ## Module 7: Output Specs for Real-World Use Opening Hook The image that looks perfect on your phone can look like garbage on a banner -- and that’s not the AI’s fault, that’s a spec problem. Learning Objectives • Match resolution and DPI to the destination: web, print, or presentation. • Choose the right file format for the job. • Plan aspect ratio before generating, not after. Lesson Breakdown • Web: pixel dimensions matter more than DPI online; low DPI is fine as long as the pixel size is right. • Print: 300 DPI minimum at final size -- the quick math of “will this look sharp on paper,” and sizing up front instead of after the fact. • Presentation: match the deck’s native slide ratio, typically 16:9, to avoid an awkward crop at the last minute. • File formats decoded: JPEG (smaller, lossy, universal), PNG (supports transparency, larger), TIFF (print-shop preferred, largest). • The golden rule: decide the destination before writing the prompt, so the image generates at the right aspect ratio the first time. Hands-On Activity Hand out a “final destination” card -- Instagram post, trade-show banner, PowerPoint slide -- and have learners state the correct resolution, format, and aspect ratio for it on the spot. Common Mistakes to Flag • Generating square and cropping into a banner shape later, which damages composition and quality. • Exporting print materials as a low-resolution JPEG pulled straight from a web preview. Wrap-Up & Bridge Your file is the right shape and size. But can you legally put it on a billboard? ## Module 8: Rights, Licensing, and Commercial Use Opening Hook Just because the AI made it doesn’t mean you own it -- or that you’re allowed to sell it. Let’s not find that out the hard way. Learning Objectives • Explain what “commercial use” typically covers. • Identify red-flag content categories before publishing. • Know where to actually check a tool’s terms, rather than assume. Lesson Breakdown • Every platform’s terms are different -- there is no universal rule, so “I read this somewhere online” is not due diligence. • Commercial use usually means the output can be used in for-profit materials, but ownership language and license language are not the same thing -- check both. • Risk categories to flag before publishing: recognizable real people, trademarked logos or characters, and an art style clearly mimicking one specific living artist. • Practical habit: save or screenshot the commercial-use section of a platform’s terms on the date you generate, since terms can change later. Hands-On Activity Present four short scenarios -- for example, using a generated mascot that looks suspiciously like a famous cartoon character in an ad -- and have learners flag each as green, yellow, or red risk. Common Mistakes to Flag • Assuming a paid subscription automatically means full commercial ownership. • Not rechecking terms after a platform updates its model or policy. Wrap-Up & Bridge Your image is legally sound. Now let’s make it actually look finished. ## Module 9: Light Post-Processing Opening Hook The AI gets you ninety percent of the way there. The last ten percent is where amateurs stop and professionals don’t. Learning Objectives • Perform a basic crop to spec. • Make simple exposure and color corrections. • Clean up a distracting background element. • Apply a light upscale when appropriate. Lesson Breakdown • Crop last, not first -- crop to the destination spec decided back in Module 7. • The three-slider fix: exposure, contrast, saturation -- small nudges, not dramatic swings. • Background cleanup basics: simple object removal and quick color-match patching. • Upscaling, honestly explained: it adds plausible detail, it doesn’t recover information that was never there, so know when it’s actually worth it. Hands-On Activity Give everyone the same flawed-but-fixable generated image. Five-minute timer to crop, correct, and clean it using only basic tools, then compare finished versions as a group. Common Mistakes to Flag • Over-editing until the image looks unnatural or over-processed. • Skipping this step entirely because “the AI already did the work.” Wrap-Up & Bridge Your image is polished and ready. One job left: actually deliver it to someone who asked for it. ## Module 10: Briefing and Delivery Workflow Opening Hook This is the module where you stop being a hobbyist playing with a toy and start being the person a team actually depends on. Learning Objectives • Translate a vague stakeholder request into a workable prompt strategy. • Present generated options professionally. • Take revision feedback and route it to the right fix. • Deliver a final asset in the correct spec. Lesson Breakdown • Translating the brief: “we need a hero image for the homepage” becomes subject, mood, brand colors, aspect ratio, and deadline -- a mini-interview checklist. • The options rule: present three solid variations, not one favorite and not ten overwhelming choices. • Feedback triage: is this a prompt change, an iteration (Module 4), or a post-processing fix (Module 9)? • Delivery checklist: correct file format, correct resolution, correct aspect ratio, and a licensing note attached (Module 8). Hands-On Activity Full role-play: one learner plays “stakeholder” with a vague request, their partner runs the entire pipeline live -- brief, prompt, three options presented, one round of feedback taken, final asset delivered in spec. Common Mistakes to Flag • Skipping the clarifying-questions step and guessing at brand intent. • Delivering a single option instead of a curated, presentable set. Wrap-Up & Bridge That’s the whole loop, start to finish. You’re now ready to walk directly into Single-Frame Image Foundations and produce a real, standalone, production-ready still.