Graduation Practice: Review This Film, Plan the Next One
💡 One-sentence summary: Turn this 30–60-second AI short drama practice film into the starting point for the next one — five-dimension acceptance, five review questions, one minimal plan page.
What You'll Be Able to Do After This Lesson
- A five-dimension acceptance checklist: duration, visuals, audio, subtitles, and licensing, each with a conclusion of "Pass / Fail / Not Checked."
- A review document: answers the five questions required by the course acceptance, explicitly marking any places where records are missing.
- An updated cost and trial-and-error log: how much you spent this time and which shot was retried more than 3 times—just open the table to find out.
- A one-page minimal plan for your next practice film: shot count cap, scene count, retry cap per shot, budget cap—all checkable numbers.
Why Review with an Acceptance Checklist Instead of by Gut Feeling?
| Capability | Reviewing by Gut Feeling | Reviewing with an Acceptance Checklist |
|---|---|---|
| Judging whether the film is finished | ❌ "Feels about done" | ✅ A conclusion for each of the five dimensions, item by item |
| Explaining where the money went | ❌ Only remembering "it was pretty expensive" | ✅ Checking against the cost and trial-and-error log |
| Locating failed shots | ❌ "A few segments weren't great" | ✅ Reviewing retry counts and failure reasons shot by shot |
| Screening for licensing risks | ❌ It never occurs to you to check | ✅ A dedicated item that must have a conclusion |
| Planning the next film | ❌ Adding difficulty on a whim | ✅ Deriving hard constraints from review conclusions |
Your Current Struggles
- The film is exported and the pre-publish checks are done, but you're not sure whether it "counts as finished"—no one has drawn a passing line for you.
- About the money spent and quite a few late nights, all you can say is "probably a lot," and you can't name which shot cost the most.
- You want to make a second film but don't know whether to change the workflow or switch tools, so you'll most likely end up reshooting the first one exactly as it was.
The problem isn't your instincts, it's your records: without acceptance criteria and a ledger, a review can only rely on impressions.
When to Use This Technique
- After your first practice film is exported and you've completed Export and Pre-Publish Checks: Against YouTube's Official Guidelines—the conclusions from the previous page are the input for this one.
- Before you start shooting the second film: review conclusions become the hard constraints for the next one directly.
- This page teaches no new tools; it does only two things: review this film using the course template, then write a smaller plan for the next one.
- Let's clarify the boundaries first: 30–60 seconds, a single scene, and few characters are the design goals of this course; they don't mean any model can generate the entire film in one go. This page reviews your production workflow, not model capability.
🎒 Preparation Before You Start
- The finished film file, plus the conclusions left from the pre-export checks on the previous page.
- The cost and trial-and-error log: how much you spent this time and how many times each shot was retried. Step 2 will use it as raw material.
- A blank review document: the Prompt below will help you generate the skeleton.
- Optional: an AI chat assistant you normally use. It organizes records; it doesn't watch the film for you.
Follow Along
Step 1: Score this AI short drama practice film against the five-dimension acceptance checklist
Why: Acceptance requires checkable conclusions, not impressions. AI can turn the course acceptance checklist into a table, but you must watch the film yourself to fill in the conclusions.
Send this message to your AI assistant:
I'm reviewing a 30–60-second AI short drama practice film. Please generate a five-dimension acceptance checklist,
presented as a Markdown table: rows are the five dimensions—duration, visuals, audio, subtitles, and licensing—
and columns are "Check Points / Conclusion / Notes."
Leave the conclusion column blank; I'll fill it in myself after watching the film.You should see: an empty table with five rows and three columns, with the five dimensions matching the course acceptance checklist one to one.
Then watch the film and score it item by item yourself, writing only "Pass / Fail / Not Checked" for each. The licensing row is the easiest to miss:
- A model license is not the same as asset licensing. Wan2.2 uses the Apache 2.0 license with user-responsibility clauses, which only covers rights claims about the model itself.
- Using a real person's likeness requires their authorization; all exercises in this course use AI-generated characters.
- edge-tts's commercial terms are not stated in the project documentation; verify them yourself before an official release.
Step 2: Hand your records to AI and answer the five review questions
Why: Every page of this course requires acceptance to specify inputs, output files, checkable completion criteria, a failure rollback method, and sources. The graduation review asks the same questions back about your own film—questions you can only answer with records, and places without records get exposed.
Please organize my materials into a review document based on the five questions below.
Answer each question in one or two sentences, citing only the records I provide; mark any you can't answer as "record missing."
My materials:
[Paste the completed five-dimension acceptance checklist]
[Paste the cost and trial-and-error log]
The five questions:
1. Were the pre-shoot inputs (script, shot list, character reference images) complete?
2. After the output files are placed back in the asset directory, can they be found by the naming rules?
3. Which shot was retried more than 3 times? Was the failure reason recorded in the log?
4. Is the completion criterion checkable, or is it just a matter of "feels okay"?
5. When something went wrong, was there a rollback method, or did you have to start over from scratch?You should see: one or two sentences for each of the five questions, all citing the records you pasted; whichever question is marked "record missing" points to the habit you need to build first in your next film.
Step 3: Pick hard constraints for the next film from your review conclusions
Why: The output of a review must be able to constrain the next film; otherwise you'll fall into the same pitfalls again.
Based on this review document: [Paste the review document]
Please pick 3–5 hard constraints for the next practice film; each must:
- Correspond to a specific conclusion in the review;
- Be checkable—use numbers, not adjectives.
For example, a shot count cap, a retry cap per shot, scene count, a budget cap.You should see: 3–5 constraints with numbers, each noting which review conclusion it comes from.
💡 This is the unique value of this course
Ever since the page Cost and Trial-and-Error Log: Try the First Pass with the Least Money, you've been keeping records. Now these ledgers are not just account books but the raw data for your review—no need to rely on memory, just open the table.
Step 4: Write a minimal plan for the next film
Why: The next film is still a practice film; the goal is to run the workflow faster, not to add difficulty.
Turn the hard constraints below into a one-page minimal plan for the next practice film:
[Paste the hard constraints]
The plan includes: one-sentence story, shot count cap, scene count, retry cap per shot, budget cap.
The next film is still a practice film, no larger in scale than the last one.You should see: a one-page plan where every item is a number or one sentence, ready to copy straight into a new budget table.
Once the plan is written, take it back to Quick Start: From Zero to Your First AI Short Drama to start your second film.
📋 Magic Prompt
The three prompts below are teaching examples and can be reused for every practice film. Placeholders marked with 【】 should be replaced, and optional items marked with ? can be deleted as whole lines.
✅ Five-Dimension Acceptance Checklist Generator
Expected result: an empty table with the same structure as the course acceptance checklist, with the conclusion column left blank for you to fill in while watching the film.
Role: short-video acceptance assistant.
Task: generate a five-dimension acceptance checklist for an AI short-drama practice film.
Input information:
### Required items
- 【Target film duration】: e.g. 30–60 seconds
- 【Delivery context】: e.g. "Not publishing yet, just for practice"
### Optional items
- 【Known issues】?: e.g. "Shot 3 has blurry motion"
Output specification: a Markdown table with rows for the five dimensions of duration, visuals, sound, subtitles, and licensing,
and columns of "Check points / Conclusion / Notes", with the conclusion column left entirely blank.
Constraints:
✅ In the licensing row, include the reminder that "a model license does not equal material authorization"
✅ Provide check points only; do not draw conclusions for the user
❌ Do not fabricate specific licensing terms for any tools or materials
❌ Do not add scoring items beyond the five dimensions📝 Five-Question Retrospective Organizer
Expected result: turn the acceptance conclusions and logs into a five-question retrospective document, with missing records clearly flagged instead of glossed over.
Role: project retrospective assistant.
Task: organize the acceptance conclusions, costs, and retry records provided by the user into a retrospective document.
Input information:
### Required items
- 【Five-dimension acceptance checklist】: conclusions for the five dimensions
- 【Cost and trial-and-error log】: total spend, retries per shot
### Optional items
- 【Failure reason notes】?: reasons recorded at the time of retries
Output specification: write one or two sentences for each of the five questions — whether inputs were complete before shooting,
whether output files are findable via the naming rules, which shot was retried more than 3 times and why,
whether the completion criteria are checkable, and whether there is a rollback method; mark "record missing" where the records hold no answer.
Constraints:
✅ Cite only the records the user provides
✅ Say plainly when records are missing; do not guess for the user
❌ Do not judge whether the film looks good
❌ Do not give action-free conclusions like "try harder next time"🚀 Next Film Minimal Plan Generator
Expected result: turn the retrospective conclusions into a one-page minimal plan where every item is checkable, to serve directly as the basis for starting the second film.
Role: short-film planning assistant.
Task: turn the retrospective conclusions into a minimal plan for the next AI short-drama practice film.
Input information:
### Required items
- 【Retrospective document】: answers to the five questions
- 【Previous film's shot count and retry limit】: e.g. "6 shots, at most 3 retries per shot"
### Optional items
- 【New approaches to try in the next film】?: e.g. "first-and-last-frame workflow"
Output specification: a one-page plan covering a one-sentence story, a shot count cap, a scene count, a per-shot retry cap,
and a budget cap, with each item noting which retrospective conclusion it comes from.
Constraints:
✅ Every item checkable: write numbers, not adjectives
✅ The next film's scale is no larger than the previous one, unless the retrospective conclusions support it
❌ Do not recommend paid services or specific products
❌ Do not promise view counts, revenue, or that the work is fit for commercial useCheckpoint ✅
- Every item on the five-dimension acceptance checklist has a conclusion, with no "unchecked" items left over.
- The retrospective document answers the five questions, and you have already figured out how to fill the gaps in the next film where "record missing" was flagged.
- The next film's plan specifies the shot count cap and the per-shot retry cap.
- The cost and trial-and-error log has been updated up to this finished film.
Pitfall Reminders
- Don't let the AI watch the film for you. It hasn't seen your finished film; the conclusions must be filled in by yourself, with the AI only doing the organizing.
- "Feels okay" is not a completion criterion. A completion criterion must be checkable by another person following it.
- Don't rush to add duration or characters to the next film. For practice films, first make the workflow fast; add difficulty later.
⚠️ The acceptance checklist is not legal advice
The acceptance checklist and retrospective question list on this page are teaching examples, not legal guarantees; the course promises no revenue and does not promise your film can be used commercially. The Wan2.2 model uses the Apache 2.0 license and includes user-responsibility clauses; it only covers the model's own rights claims and does not imply that voiceovers, music, or materials are fit for commercial use.
Advanced Tips
- Don't start the retrospective document from scratch: store it in the materials directory, use the same naming rules as the shot files, and flip through it directly when the second film starts shooting.
- No need to wait until graduation to do a retrospective: at the end of each stage, use a simplified version of the five questions, asking only "where are the output files, and what are the completion criteria".
Lesson Summary
- The graduation standard for this AI short-drama practice film is not "looks good", but that all five dimensions have conclusions and all five questions have answers.
- The raw material of a retrospective is the log you kept along the way. Without records, retrospection is just guesswork; with records, you only need to look at the tables.
- The next film grows out of the retrospective conclusions: fewer shots, a simpler, more unified scene, a clear retry cap — turning one production into a sustainable iteration.
Next Lesson Preview
This is the last page of the course; there is no next lesson. Your next step is your second film: take the minimal plan from this page back to Quick Start: From Zero to Your First AI Short Drama; first copy Cost and Trial-and-Error Log: Try a First Pass with the Least Money into the budget table for the second film; whichever shot fails repeatedly, go back to Shot Retry and Replacement Strategy: At Most 3 Retries per Shot to check against the retry cap.