Generative Media Foundations

Tracks

1. 01🎬Generative Media AI가 진짜로 뭘 하는가

   0/10 lessons

   검색이 아니라 합성 — 이게 출발점

   Generative media가 언어 AI랑 어떻게 다른지, prompt가 왜 마법처럼 느껴지는지, 모델이 실제로 *예측*하는 게 뭔지, 그리고 2026년 프로 스택이 어떻게 layered pipeline으로 돌아가는지 mental model을 세우는 트랙이야.

   Lesson list (10)

   • 01
     What "Generative AI" Means for Images and Video~12 min · foundations, mental-model, l1
   • 02
     Language Generation vs. Media Generation~15 min · foundations, mental-model, l2
   • 03
     Why It Feels Like Magic~18 min · foundations, mental-model, l3
   • 04
     Prediction, Not Understanding~14 min · foundations, mental-model, l4
   • 05
     Probability Distributions Over Possible Images~16 min · foundations, mental-model, l5
   • 06
     Why Prompting Is Not Magic~15 min · foundations, mental-model, l6
   • 07
     Generation vs. Editing vs. Transformation vs. Upscaling~17 min · foundations, mental-model, l7
   • 08
     The Modes: Text-to-Image, Inpainting, Video, and Everything In Between~14 min · foundations, mental-model, l8
   • 09
     Why Image Generation Matured Faster Than Video~16 min · foundations, mental-model, l9
   • 10
     The Modern Stack and Your Learning Roadmap~18 min · foundations, mental-model, l10

   ✓퀴즈 · 5 문제→

2. 02🌀Latent Space, Diffusion, 핵심 직관

   0/10 lessons

   Noise가 어떻게 이미지가 되는가

   엔진을 열어 보는 트랙. Latent space가 왜 raw pixel보다 똑똑한지, forward/reverse process의 비대칭, seed와 randomness, sampling steps tradeoff, guidance scale의 sweet spot, flow matching이 'diffusion 2.0'인 이유, 그리고 모델이 'taste 있는 것처럼 느껴지는' 진짜 이유까지.

   Lesson list (10)

   • 01
     Why Models Don't Generate Pixels Directly~12 min · diffusion, latent-space, l1
   • 02
     Diffusion Intuition: From Noise to Image~15 min · diffusion, latent-space, l2
   • 03
     Forward Process vs. Reverse Process~18 min · diffusion, latent-space, l3
   • 04
     Why Denoising Works as Generation — and Text Conditioning~14 min · diffusion, latent-space, l4
   • 05
     Noise, Randomness, Seeds, and Variation~16 min · diffusion, latent-space, l5
   • 06
     Sampling Steps: Speed vs. Quality Tradeoffs~15 min · diffusion, latent-space, l6
   • 07
     Guidance: Prompt Adherence vs. Natural Outputs~17 min · diffusion, latent-space, l7
   • 08
     Diffusion vs. Autoregressive vs. Flow-Based Approaches~14 min · diffusion, latent-space, l8
   • 09
     How Generators "Know" Style, Lighting, and Composition~16 min · diffusion, latent-space, l9
   • 10
     Why These Systems Feel Like They Have Taste~18 min · diffusion, latent-space, l10

   ✓퀴즈 · 5 문제→

3. 03✍️이미지 prompt 제대로 쓰기

   0/10 lessons

   단어가 실제로 하는 일 (그리고 못 하는 일)

   Prompt는 명령이 아니라 trajectory의 중력이야. 강한 prompt의 anatomy, 모델별 스타일 차이 (FLUX vs SDXL vs Midjourney vs DALL-E), negative prompt의 진실, prompt 실패 진단법, seed 전략, 그리고 'one perfect prompt' 신화를 깨고 practitioner mindset으로 가는 길.

   Lesson list (10)

   • 01
     What a Prompt Actually Does Inside the Model~12 min · prompting, control, l1
   • 02
     The Anatomy of an Image Prompt~15 min · prompting, control, l2
   • 03
     Prompt Order: When It Matters and When It Doesn't~18 min · prompting, control, l3
   • 04
     Descriptive Prompts vs. Keyword Stacks vs. Natural Language~14 min · prompting, control, l4
   • 05
     Quality Tags: When "Masterpiece" Helps and When It Hurts~16 min · prompting, control, l5
   • 06
     Negative Prompting: What It Is, When It Works~15 min · prompting, control, l6
   • 07
     Model-Specific Prompt Behavior~17 min · prompting, control, l7
   • 08
     Underspecification, Overspecification, and Prompt Conflict~14 min · prompting, control, l8
   • 09
     Reading Outputs Diagnostically~16 min · prompting, control, l9
   • 10
     Seeds, Rerolls, and the Practical Prompting Mindset~18 min · prompting, control, l10

   ✓퀴즈 · 5 문제→

4. 04🔍이미지 모델이 *예측 가능하게* 실패하는 이유

   0/10 lessons

   실패를 architecture로 읽기

   Text rendering·counting·hands·spatial layout·character consistency·style leakage·구조적 디자인 약점·표면적 그럴듯함 vs 실제 정확성·training prior bias — 그리고 4-카테고리 진단 프레임. 실패를 짜증이 아니라 architectural signal로 읽는 능력을 만드는 트랙.

   Lesson list (10)

   • 01
     Why Text Rendering Is Historically Hard~12 min · failures, diagnosis, l1
   • 02
     Why Counting Is Hard~15 min · failures, diagnosis, l2
   • 03
     Why Hands, Fingers, and Complex Objects Break~18 min · failures, diagnosis, l3
   • 04
     Why Spatial Relationships Are Fragile~14 min · failures, diagnosis, l4
   • 05
     Why Character Consistency Is Difficult~16 min · failures, diagnosis, l5
   • 06
     Style Leakage and Unintended Aesthetic Drift~15 min · failures, diagnosis, l6
   • 07
     Why Logos, UI, and Charts Need Different Strengths~17 min · failures, diagnosis, l7
   • 08
     Surface Plausibility vs. Actual Correctness~14 min · failures, diagnosis, l8
   • 09
     Training Data Priors and Composition Bias~16 min · failures, diagnosis, l9
   • 10
     Diagnosing Failures: A Systematic Framework~18 min · failures, diagnosis, l10

   ✓퀴즈 · 5 문제→

5. 05🎛️Control, Editing, Reference 기반 생성

   0/10 lessons

   Text-only 너머의 control 스펙트럼

   Reference image, image-to-image, inpainting, outpainting, masking, identity preservation, character consistency, ControlNet, iterative editing, 그리고 compositing mindset까지 — vending machine이 아니라 collaborator로 모델을 다루는 도구 박스.

   Lesson list (10)

   • 01
     Why Real Workflows Go Beyond Pure Text Prompting~12 min · control, editing, l1
   • 02
     Reference Images: Visual Anchors for Everything~15 min · control, editing, l2
   • 03
     Image-to-Image: Structure Preservation with Style Transformation~18 min · control, editing, l3
   • 04
     Inpainting: Editing Part of an Image~14 min · control, editing, l4
   • 05
     Outpainting: Extending the Canvas~16 min · control, editing, l5
   • 06
     Masking, Style Transfer, and Identity Preservation~15 min · control, editing, l6
   • 07
     Character Consistency Strategies~17 min · control, editing, l7
   • 08
     Pose and Composition Control: ControlNet Concepts~14 min · control, editing, l8
   • 09
     Iterative Editing Beats the One Perfect Prompt~16 min · control, editing, l9
   • 10
     The Compositing Mindset: Photoshop and Post-Processing~18 min · control, editing, l10

   ✓퀴즈 · 5 문제→

6. 06🎥비디오 생성 — 시간이 모든 걸 바꾼다

   0/10 lessons

   왜 'image × N'이 아닌가

   Temporal coherence가 뭘 multiplicative하게 만드는지, motion failure의 종류, camera motion vs subject motion 분리, 연속성의 숨은 난이도, T2V·I2V·V2V 비교, image-to-video anchoring이 왜 sweet spot인지, keyframe·shot design, 그리고 클립 기반 워크플로우의 실전 진실.

   Lesson list (10)

   • 01
     Video Is NOT Just Image Generation Repeated~12 min · video, temporal, l1
   • 02
     Temporal Consistency: The Core Challenge~15 min · video, temporal, l2
   • 03
     Motion Coherence: Why Things Drift, Melt, and Morph~18 min · video, temporal, l3
   • 04
     Camera Movement vs. Subject Movement~14 min · video, temporal, l4
   • 05
     The Hidden Difficulty of Continuity~16 min · video, temporal, l5
   • 06
     Text-to-Video, Image-to-Video, Video-to-Video~15 min · video, temporal, l6
   • 07
     Why Image-to-Video Anchoring Works~17 min · video, temporal, l7
   • 08
     Keyframe Intuition: Anchoring Moments~14 min · video, temporal, l8
   • 09
     Shot Design and Motion Prompting~16 min · video, temporal, l9
   • 10
     Clip Limits, Looping, and the Practical Truth~18 min · video, temporal, l10

   ✓퀴즈 · 5 문제→

7. 07🎵Audio, Voice, Multimodal Layer

   0/10 lessons

   사운드는 경험의 절반

   Voice 생성과 image 생성의 평행과 차이, TTS·voice cloning·emotional delivery, naturalness가 왜 uncanny valley로 빠지는지, lip sync의 100ms 허용 오차, native audio (Veo 3) vs post-generated workflow, 4-layer 사운드 디자인, multimodal vs specialized chain의 tradeoff.

   Lesson list (10)

   • 01
     Media Generation Now Includes Sound~12 min · audio, voice, l1
   • 02
     Voice Generation vs. Image Generation~15 min · audio, voice, l2
   • 03
     TTS, Voice Cloning, and Emotional Delivery~18 min · audio, voice, l3
   • 04
     Why "Say This Naturally" Is Harder Than It Sounds~14 min · audio, voice, l4
   • 05
     Lip Sync and Audio-Video Alignment~16 min · audio, voice, l5
   • 06
     Native Audio in Video Models vs. Post-Generated Audio~15 min · audio, voice, l6
   • 07
     Sound Effects, Ambience, Music, and Voice as Layers~17 min · audio, voice, l7
   • 08
     Why Multimodal Generation Matters~14 min · audio, voice, l8
   • 09
     Multimodal Models vs. Single-Modality Tools~16 min · audio, voice, l9
   • 10
     The Future: Unified Pipelines vs. Specialized Chains~18 min · audio, voice, l10

   ✓퀴즈 · 5 문제→

8. 08🧰모델 패밀리, Tradeoff, 도구 선택

   0/10 lessons

   Hype 말고 task 기준으로

   Image·video·voice 모델 landscape (FLUX·GPT-Image·Midjourney·SD·Imagen·Runway·Veo·Kling·ElevenLabs·Voxtral). Open vs closed, fast vs premium, task-to-model mapping, cost·speed·controllability tradeoff, 자기 자신을 속이지 않는 평가법, 그리고 실용적 decision framework.

   Lesson list (10)

   • 01
     The Image Model Landscape~12 min · models, tradeoffs, l1
   • 02
     The Video Model Landscape~15 min · models, tradeoffs, l2
   • 03
     Voice and Audio Models~18 min · models, tradeoffs, l3
   • 04
     Open vs Closed Models~14 min · models, tradeoffs, l4
   • 05
     Fast Models vs Premium Models~16 min · models, tradeoffs, l5
   • 06
     Matching Models to Creative Tasks~15 min · models, tradeoffs, l6
   • 07
     Cost, Speed, and Controllability~17 min · models, tradeoffs, l7
   • 08
     How to Evaluate a Model Without Fooling Yourself~14 min · models, tradeoffs, l8
   • 09
     One-Shot Hero vs Batch Exploration~16 min · models, tradeoffs, l9
   • 10
     Practical Decision Framework~18 min · models, tradeoffs, l10

   ✓퀴즈 · 5 문제→

9. 09🛠️실전 Creative Workflow

   0/10 lessons

   단일 prompt가 아니라 pipeline

   Ideation, production, character workflow, thumbnail, story shot 시퀀스, commercial, 비디오 pipeline, curation as superpower, human taste, 그리고 beginner → practitioner mindset shift. '하나의 마법 prompt' 신화를 버리고 layered pipeline으로 일하는 법.

   Lesson list (10)

   • 01
     The Ideation Workflow~12 min · workflows, production, l1
   • 02
     The Production Workflow~15 min · workflows, production, l2
   • 03
     The Character Workflow~18 min · workflows, production, l3
   • 04
     The Thumbnail Workflow~14 min · workflows, production, l4
   • 05
     The Story Workflow~16 min · workflows, production, l5
   • 06
     The Commercial Workflow~15 min · workflows, production, l6
   • 07
     The Video Workflow~17 min · workflows, production, l7
   • 08
     Why Curation Is a Superpower~14 min · workflows, production, l8
   • 09
     The Role of Human Taste~16 min · workflows, production, l9
   • 10
     What Beginners Should Stop Expecting and Start Doing~18 min · workflows, production, l10

   ✓퀴즈 · 5 문제→

10. 10🧭Drowning 없이 최신 유지하기

    0/10 lessons

    빗방울 말고 날씨를 추적

    분야가 왜 이렇게 빨리 바뀌는지 (구조적 이유), information diet 짜는 법, 7대 theme 추적, launch hype critical하게 읽기, demo cherry-picking과 repeatable value 구분, 자기만의 evaluation framework 만들기, 그리고 craftsperson의 mental model로 마무리.

    Lesson list (10)

    • 01
      Why Generative Media Changes Too Fast~12 min · evaluation, staying-current, l1
    • 02
      How to Track Progress~15 min · evaluation, staying-current, l2
    • 03
      Key Themes to Watch~18 min · evaluation, staying-current, l3
    • 04
      How to Compare Model Launches Critically~14 min · evaluation, staying-current, l4
    • 05
      Demo Cherry-Picking vs Repeatable Value~16 min · evaluation, staying-current, l5
    • 06
      What Matters More Than Hype~15 min · evaluation, staying-current, l6
    • 07
      How to Keep Learning~17 min · evaluation, staying-current, l7
    • 08
      Building Your Own Evaluation Framework~14 min · evaluation, staying-current, l8
    • 09
      The Evolving Ecosystem~16 min · evaluation, staying-current, l9
    • 10
      Final Synthesis~18 min · evaluation, staying-current, l10

    ✓퀴즈 · 5 문제→