Creative AI Training Playbook

This guide is a practical curriculum for learning creative AI by making real work. The goal is not to collect tools. The goal is to become the kind of creator who can plan, generate, edit, package, and sell AI-made videos, ads, images, avatars, and campaigns.

Use it as a free training path for Ekly users. Each module ends in a concrete output, because AI skill only compounds when people ship finished work and review what happened.

Training principle

Teach the workflow, not just the tool. A strong creator knows how to define the brief, control references, keep characters and products consistent, direct motion, edit the result, and turn the output into a deliverable someone can use.

The five-day quickstart

The first week should remove confusion. Instead of sending new users into a library of lessons, give them a guided path that produces a first finished asset.

Pick one goal

Choose ads, short film, UGC, product visuals, consulting, or a reusable workflow.

e.g. “I want to make a 30-second spec ad for a fictional skincare brand.”

Collect references

Gather 3-5 examples for look, pacing, camera language, product style, and sound.

e.g. “One hero image, two shot references, one sound reference, one target ad.”

Make the first asset

Generate one usable scene, image set, voiceover, avatar clip, or edit.

e.g. “A product-on-counter hero shot with steam, morning light, and slow push-in.”

Review like a director

Score consistency, clarity, motion, emotion, and usefulness before making more.

e.g. “The product label drifts, but lighting and camera direction are strong.”

Package the output

Export, caption, title, localize, and write the handoff note.

e.g. “Deliver 9:16, 1:1, and 16:9 versions with caption files and prompt notes.”

Foundations every creator needs

Before specializing, train the basics that show up in every AI creative workflow: prompts, references, consistency, file organization, and critique.

Foundation	What to teach	Practice output
Prompting	Subject, action, environment, camera, style, constraints, and negative instructions.	Three prompt versions for the same scene: simple, controlled, cinematic.
Reference control	How to use images, video, audio, moodboards, and brand materials without muddying the brief.	A reference board with labels for look, motion, sound, character, and product.
Consistency	Character sheets, product sheets, location sheets, naming conventions, and shot memory.	A mini continuity bible for one character, product, or world.
Editing judgment	How to decide what to regenerate, what to edit, and what to accept.	A before-after breakdown explaining five specific creative decisions.

The simple scoring rubric

Rate every output from 1-5 on brief match, identity consistency, product consistency, motion quality, visual taste, sound fit, and delivery readiness. Regenerate only the lowest-scoring dimension.

The full training architecture

A complete creative-AI course should behave like a ladder: orient the creator, teach foundations, let them choose a path, then keep improving through critique, projects, and business practice.

Layer	Purpose	What to include
Start here	Remove confusion and get the user moving.	Orientation, account setup, tool setup, how to ask questions, first output, and where to post work.
Foundations	Teach the core skills that every path needs.	Image prompting, video prompting, reference control, project organization, character consistency, product consistency.
Choose your path	Help users specialize instead of wandering.	Ads, short films, UGC, product visuals, workflow systems, consulting, and portfolio building.
Tool library	Explain model categories by job-to-be-done.	Image, video, avatar, voice, music, editing, upscaling, sound, worldbuilding, and workflow automation.
Business layer	Turn creative outputs into offers.	Client acquisition, pricing, retainers, project scoping, audits, case studies, and handoff documents.
Community loop	Keep taste and execution improving.	Welcome calls, technical Q&A, feedback sessions, contests, showcases, events, and leaderboards.

How to structure this inside Ekly

Each lesson should end with an Ekly project template, a practice assignment, a review checklist, and an export target. That turns passive learning into repeatable creation.

Choose a specialization path

Once the foundations are clear, users should choose a path. The point of a path is focus: one audience, one deliverable, one repeatable workflow.

AI ads

Goal: Create brand-safe commercials and spec ads that feel intentional.

Deliverable: A 15-30 second product spot with captions, music, and export variants.

Ekly stack: Video generation, image generation, voiceover, captions, editor, brand kit.

AI short film

Goal: Build cinematic story pieces with controlled character, location, pacing, and sound.

Deliverable: A 45-90 second short with a shot list and director notes.

Ekly stack: Director Mode, video generation, music, sound design, editor.

UGC at volume

Goal: Produce batches of social clips for offers, products, and creators.

Deliverable: Ten short hooks or product clips from one reusable template.

Ekly stack: Avatars, voiceover, screen record, captions, templates.

Product visuals

Goal: Make commercial-grade images and motion assets for ecommerce and ads.

Deliverable: A product shoot pack: hero, detail, lifestyle, motion teaser.

Ekly stack: Image generation, video generation, upscaling, brand kit.

AI consulting

Goal: Help a business audit, design, and adopt practical AI workflows.

Deliverable: A paid audit deck plus one working prototype workflow.

Ekly stack: Workspace, library, exports, reusable prompt templates.

Scalable workflow

Goal: Design a repeatable process or tool that can be sold, batched, or delegated.

Deliverable: A documented pipeline with inputs, steps, checks, and outputs.

Ekly stack: Templates, agents, project library, editor, export presets.

What each path should teach

The course scan showed that the best path lessons are not abstract. They are built around a specific deliverable and the exact production problems that deliverable creates.

Path	Core lessons	Capstone project
AI ads	Offer research, concept angles, shot planning, product consistency, before-after comparisons, hooks, captions, and platform exports.	One complete spec ad with three cutdowns and a one-page production breakdown.
Short films	Story beats, character sheets, location sheets, visual continuity, camera language, music timing, edit rhythm, and emotional arc.	A 60-second cinematic short with storyboard, shot list, and director notes.
UGC and social	Hook writing, avatar/voice choices, batch templates, caption systems, volume production, and content calendars.	Ten clips from one reusable UGC template across three angles.
Image-heavy creative	Art direction, product sheets, commercial-grade lighting, background control, scene variants, and retouching strategy.	A product shoot pack with hero, lifestyle, detail, and motion assets.
Scalable systems	Reusable prompts, input forms, workflow diagrams, batch generation, QA steps, and repeatable delivery.	A documented pipeline someone else can run from brief to export.
AI consulting	Business audits, workflow mapping, stakeholder interviews, prototype design, training calls, and retainers.	A paid audit-style report plus one working prototype workflow.

The end-to-end production workflow

Train people to think like a small studio. The workflow below works for ads, films, UGC, product visuals, and client experiments.

Stage	Question	Output
Brief	Who is this for and what should they feel or do?	One-paragraph creative brief.
Pre-production	What must stay consistent?	Character, product, location, and style sheets.
Generation	Which model handles each shot or asset best?	First pass shots, images, voices, and music.
Direction	What is wrong, and what instruction fixes it?	Revision notes by scene, not vague opinions.
Edit	How does the piece move from idea to finished cut?	Timeline with pacing, captions, sound, and color choices.
Delivery	How will the audience or client use it?	Exports, thumbnails, caption files, and handoff notes.

✓ Reusable creative brief

Create a [format] for [audience] promoting [offer]. The emotional target is [feeling]. The visual world is [style references]. The piece must include [must-have details] and avoid [constraints]. Output a shot list, asset list, and first-pass prompts.

Use this to start a project before generating any media.

Foundation model map

Instead of teaching model names first, teach model categories. Users need to understand what each category is good at, what inputs it wants, and how to evaluate the output.

Model category	Best for	Inputs to collect	Quality checks
Text reasoning	Briefs, scripts, shot lists, prompt drafting, critique, scope, and documentation.	Audience, goal, references, constraints, deliverables, brand voice.	Specificity, structure, missing assumptions, actionability.
Image generation	Concept art, product visuals, storyboards, character sheets, moodboards, thumbnails.	Subject description, reference images, style direction, lighting, composition, constraints.	Identity consistency, product accuracy, hands/details, lighting, typography safety.
Video generation	Motion shots, camera moves, stylized scenes, transitions, animated product moments.	Subject, action, environment, camera move, duration, aspect ratio, reference frames.	Temporal stability, physics, camera intent, flicker, object drift, usable edit handles.
Video editing models	Changing or extending existing clips while preserving motion and context.	Original clip, target frame, mask or region, edit instruction, preserve constraints.	Preserved identity, clean boundaries, continuity, no unwanted global changes.
Voice and speech	Narration, UGC voice, character voice, localization, explainers.	Script, speaker profile, language, pacing, tone, pronunciation notes.	Natural cadence, emotion, intelligibility, mouth-fit for lip-sync.
Music and sound	Scores, ambient beds, transitions, hit points, commercial pacing.	Mood, genre, tempo, scene beats, reference adjectives, duration.	Loopability, mix room for voice, emotional fit, edit timing.
Avatars and lip-sync	Talking characters, spokesperson clips, multilingual variants, education content.	Character image/video, clean script, voice, expression target, framing.	Mouth alignment, facial stability, eye focus, gesture believability.
Enhancement and restoration	Upscaling, cleaning artifacts, sharpening, improving delivery quality.	Source asset, target resolution, preservation notes, crop/aspect requirements.	No waxy faces, no texture hallucination, readable details, natural edges.
Workflow automation	Batching prompts, repeatable pipelines, internal tools, creative operations.	Inputs, outputs, rules, QA checklist, export format, failure states.	Repeatability, inspectability, error handling, handoff clarity.

Avoid the model-name trap

Model names change constantly. Teach users to ask: what input does this model understand, what output does it control, and where does it fail? That knowledge survives every tool update.

Build a tool and workflow library

A useful training library is organized by jobs-to-be-done, not by hype. Each lesson should answer: when to use this tool, what inputs it needs, where it fails, and how to finish the output in Ekly.

Image creation: campaign visuals, character references, product mockups, storyboards, backgrounds.
Video generation: motion shots, scene extensions, controlled camera moves, stylized transitions.
Lip-sync and avatars: talking characters, explainers, founder clips, multilingual variants.
Enhancement and upscaling: recovering quality, polishing client exports, improving old assets.
Sound design: voice, music, ambient beds, hit points, and edit rhythm.
Worldbuilding: consistent spaces, recurring locations, maps, sets, and explorable visual systems.
Workflow building: prompt packs, reusable templates, batch systems, and AI-assisted pipelines.

Lesson format

Use case -> inputs -> workflow -> common failure -> fix -> Ekly finishing step -> practice assignment

Use live practice to improve faster

Recorded lessons are useful, but live loops are what sharpen taste. The training system should include regular critique and Q&A moments where users bring unfinished work.

Session	Purpose	User brings
Welcome call	Help new users choose a path and ship their first output.	Goal, skill level, one example they admire.
Technical Q&A	Debug tools, models, prompts, quality, and workflow problems.	Broken prompt, bad output, reference assets.
Feedback Friday	Review finished or nearly finished work with a director mindset.	Export, brief, target audience, question.
Pricing clinic	Turn work into an offer with scope, terms, and a confident price.	Client type, deliverable list, timeline.

Critique rule

Do not ask, "Is this good?" Ask, "What is the single biggest thing preventing this from working for its audience?"

Design the learning community

The course structure works because it combines lessons with social accountability. For Ekly, this can become a lightweight training system around user projects.

Community space	What happens there	Ekly version
General onboarding	New users introduce themselves, state a goal, and ask for help.	A guided onboarding prompt that creates the user's first project brief.
Creative work	Members post work-in-progress and finished pieces.	A share page with prompt notes, outputs, and a critique checklist.
Standouts	The best work is highlighted so others can study why it works.	A gallery of featured Ekly projects with teardown notes.
Job opportunities	People exchange paid work, requests, and client leads.	A creator offer board or template pack for packaging services.
New modules	Updates teach the newest workflows and model capabilities.	Release-based mini-guides tied to new Ekly features.
Q&A	Users debug prompts, tools, edits, and pricing problems.	A structured support prompt: goal, input, output, failure, desired fix.
Wins and pricing help	Members turn outcomes into confidence and business learning.	Case-study templates, pricing calculators, and proposal examples.
Contests	Challenges create urgency, repetition, and taste calibration.	Monthly Ekly challenges with constraints, judging criteria, and featured results.

Turn AI output into paid work

Training should make the business side concrete. People need examples of sellable deliverables, pricing logic, project structure, and client communication.

Offer	What client buys	Starter scope
Spec ad package	A proof-of-concept ad for a product or campaign.	One 15-30 second ad plus two cutdowns.
UGC batch	A month of short-form product clips.	Ten clips from three scripts and two avatar styles.
Product visual pack	Images and motion assets for listings and paid social.	Eight images, three loops, one hero video.
AI workflow audit	A map of where AI saves time or creates new output.	One process audit, one prototype, one training call.
Music video or brand film	A stylized story piece with a strong visual concept.	One concept deck, one 60-second film, project files.

✓ Pricing prompt

Act as a producer. For this deliverable [describe deliverable], estimate the scope, revision risk, asset requirements, timeline, and three price tiers: starter, standard, premium. Include what is excluded from each tier.

Use this to teach scope clarity, not to let AI decide the final price.

Path chooser quiz

The fastest way to help a learner is to stop them from trying every workflow at once. Give them a simple intake that turns their goal into a path and a first project.

If the learner says	Recommend	First project
I want brands or clients to hire me for campaigns.	AI ads	A 15-second spec ad for a real category with a fictional brand.
I care about cinema, storytelling, music videos, or trailers.	Short films	A 45-second story with one character, one location, and three shots.
I want fast content for social, offers, or creators.	UGC and social	Five hooks for the same product using one repeatable clip template.
I like visuals, art direction, products, and campaign imagery.	Product visuals	A product image pack with a matching motion teaser.
I want to help businesses adopt AI.	AI consulting	A one-process workflow audit and a prototype automation.
I want to build systems I can reuse, sell, or delegate.	Scalable workflow	A documented pipeline with inputs, prompts, QA checks, and exports.

✓ Learner intake prompt

Interview me as a creative AI coach. Ask 8 questions about my goals, taste, current skills, available time, tools, target audience, income goals, and examples I admire. Then recommend one learning path, one first project, and a 7-day plan.

This can become an onboarding form inside Ekly.

How to ask for useful help

Beginners often ask vague questions, then get vague answers. Teach them to bring enough context that another creator can diagnose the problem.

Include	Why it matters	Example
Goal	Reviewers need to know what the piece is supposed to do.	This is a 15-second skincare ad for paid social.
Input	The prompt, reference, script, or source media explains the cause.	Here is my prompt, product image, and target style board.
Output	The actual result shows the failure, not just the complaint.	The bottle label changes between shot 1 and shot 2.
What you tried	This prevents repeated advice and teaches troubleshooting.	I tried stronger product constraints and a closer crop.
Specific question	A narrow question gets a useful answer.	How would you preserve the label while keeping the camera move?

Help request format

Goal: ... / Inputs: ... / Output: ... / Problem: ... / Tried: ... / Question: ...

Critique rubrics by project type

A good free guide should teach people how to judge their own work before asking someone else. Rubrics make taste trainable.

Project	Score these dimensions	Pass condition
Spec ad	Hook clarity, product accuracy, benefit clarity, visual polish, pacing, caption readability.	A stranger understands the product and why it matters within 5 seconds.
Short film	Character clarity, story beat progression, emotional arc, shot continuity, music fit.	The viewer can retell what happened and what changed emotionally.
UGC clip	First line, authenticity, offer clarity, face/voice fit, retention, CTA.	The clip feels native to the platform and gets to the point quickly.
Product visual	Product fidelity, lighting, composition, brand fit, texture, commercial usefulness.	The asset could sit on a landing page, ad, or listing without explanation.
Workflow system	Repeatability, input clarity, QA checks, failure handling, export consistency.	Another person can run it and produce a predictable result.

Review cadence

First review the brief, then the output, then the edit. Most bad AI work fails because the brief is muddy, not because the model is weak.

Common AI failure modes and fixes

This is one of the highest-benefit learner sections because it turns frustration into a diagnosis process.

Failure	Likely cause	Fix
Character changes between shots	The model lacks a stable identity anchor.	Create a character sheet, reuse the same reference frame, reduce outfit changes, and describe only stable traits.
Product label or shape drifts	The product was treated as a style cue instead of a protected object.	Use a product sheet, close framing, preserve constraints, and regenerate only the shot that fails.
Motion looks floaty or random	The action and camera move are competing.	Separate subject action from camera movement and ask for one clear motion at a time.
Image looks generic	The prompt describes genre but not taste.	Add art direction: lens, lighting, surface, material, era, palette, composition, and negative constraints.
Voice sounds flat	The script has no performance direction.	Add audience, emotion, pacing, pauses, emphasis, and pronunciation notes.
Edit feels slow	Every shot has the same energy.	Cut to beats, alternate shot sizes, remove setup time, and let captions carry context.
Output is impressive but unsellable	It was made as a demo, not a deliverable.	Define target user, platform, format, CTA, export specs, and handoff package before generation.

Challenge system for faster growth

Challenges compress learning because they give everyone the same constraint, deadline, and review criteria. They also create examples future learners can study.

Challenge type	Constraint	What learners practice
30-second spec ad	One product, one audience, one benefit, three shots.	Concepting, product consistency, hooks, pacing, and export polish.
One-character short	One character, one location, clear beginning/middle/end.	Continuity, story structure, emotion, and shot planning.
Image-to-motion	Start from one still image and create a usable motion asset.	Reference control, camera movement, motion realism, and edit handles.
UGC hook lab	Same offer, five different opening lines.	Audience empathy, scripting, performance, and retention.
Workflow teardown	Publish the exact steps, failure points, and fixes.	Documentation, repeatability, and teaching ability.

Challenge brief template

Goal / Audience / Required assets / Constraints / Judging rubric / Deadline / Export specs / Reflection questions

Portfolio and case-study system

The learner gets more benefit when every assignment becomes a portfolio artifact. Teach them to document the process, not just upload the result.

Case-study block	What to write	Why it helps
Brief	The audience, goal, constraints, and deliverable.	Shows the work solved a problem instead of chasing novelty.
Creative direction	References, style choices, tone, pacing, and why they fit.	Shows taste and strategic thinking.
Workflow	Inputs, tools by category, prompts, edit steps, and QA checks.	Shows repeatability and makes the creator hireable.
Before and after	Failed outputs, fixes, final result, and decision notes.	Shows judgment, not just luck.
Deliverables	Formats, lengths, captions, thumbnails, and handoff assets.	Shows client readiness.

✓ Case-study writer

Turn this AI project into a concise portfolio case study. Include brief, creative direction, workflow, key failures, fixes, final deliverables, and what I would improve next. Keep it honest and client-friendly.

The best case studies show decision-making, not just a glossy final output.

Progress levels and learner milestones

Learners stay motivated when progress is visible. Use levels to define what they can do, not how many lessons they watched.

Level	Capability	Proof
1	Can create a first usable AI asset.	One exported image, clip, voiceover, or edit.
2	Can follow a guided workflow.	One completed quickstart project.
3	Can control consistency.	A character, product, or location stays stable across multiple outputs.
4	Can edit and package work.	A polished deliverable with captions, sound, and platform exports.
5	Can choose and execute a path.	One capstone project in ads, film, UGC, visuals, consulting, or systems.
6	Can critique and improve others' work.	Three useful teardown comments or review notes.
7	Can sell or deploy a workflow.	A portfolio case study, proposal, audit, or reusable system.

Practice assignments

These assignments can become free guide exercises, onboarding tasks, or challenges for users learning Ekly.

Make a 15-second product ad using one reference image, one generated scene, one voiceover, and captions.
Create a character sheet, then generate three scenes where the character stays recognizable.
Turn one product into a visual pack: ecommerce hero, lifestyle image, macro detail, and five-second motion loop.
Make three UGC-style hooks for the same product, each with a different emotional angle.
Take a weak generation and document the exact revision path that made it usable.

A 30-day training plan

For a public free guide, this plan is the strongest shape: it gives users momentum and creates obvious reasons to come back to Ekly.

Week	Focus	Outcome
1	Foundations and first finished asset.	One exported clip or visual pack.
2	Specialization path and consistency.	One complete mini-project in the chosen track.
3	Editing, sound, captions, and delivery.	A polished version with platform exports.
4	Offer, portfolio, and repeatable workflow.	A public case study and sellable package.

End state

A trained creator should leave with a portfolio piece, a repeatable workflow, a clear offer, and the confidence to make the next project faster.