Overview #
If the Short Video Maker workflow does not run correctly, the issue is usually caused by one of a few common areas:
- credentials or API access
- Google Sheet structure
- broken expressions
- AI prompt quality
- image generation moderation failures
- JSON2Video template mismatches
Because this workflow passes data through multiple connected services, a single wrong field, missing value, or disconnected credential can stop the whole process.
Recommended Troubleshooting Method #
The safest way to troubleshoot this workflow is to test it one section at a time, not all at once.
Recommended order #
- test Idea Generation
- test Video Creation
- test Publishing
Why this works #
This makes it much easier to identify exactly where the problem starts. If the first section fails, there is no need to debug later sections yet.
Quick Checklist #
Before going deeper, check these first:
- all credentials are connected
- API keys are valid
- the Google Sheet structure is correct
- required dropdown values are exact
- expressions still point to real data
- AI prompts still match the expected parser format
- image prompts are not being blocked by OpenAI safety checks
- the JSON2Video template is installed correctly
- the row selected for publishing actually has a valid video URL
Common Problems and Fixes #
1. Credentials or API Access Failure #
Symptoms #
- the workflow fails immediately
- a node stops with an authentication or permission error
- OpenAI, Google, Cloudinary, JSON2Video, or YouTube nodes do not run
Likely cause #
One or more credentials were not reconnected after importing the workflow, or the API key/account access is invalid.
What to check #
- OpenAI credential is connected and has active API access
- Google Sheets and Google Drive credentials are connected to the correct Google account
- Cloudinary credential is correct
- JSON2Video API key is valid
- YouTube credential is active and connected to the correct channel
- all credentials in n8n are mapped to your own environment, not the original one
Fix #
Reconnect every external service credential inside n8n and test the related node again. If needed, generate a new API key or reconnect OAuth access.
2. Google Sheet Structure Is Wrong #
Symptoms #
- no row is selected for production
- no row is selected for publishing
- updates fail when writing data back to the sheet
- wrong row is updated or no row is matched
Likely cause #
The Google Sheet does not match the required structure.
Required columns #
The sheet must contain these columns exactly:
- ID
- Idea
- Production Status
- Title
- Caption
- Cover Image
- Video URL
- Posting Status
- Published Date
Required dropdown values #
The following values must match exactly:
| Column | Required Values |
|---|---|
| Production Status | Not Produced, Produced |
| Posting Status | Not Published, Published |
Fix #
If the user created the sheet manually, compare it against the provided template and correct any differences. If possible, use the included spreadsheet template instead of rebuilding the structure from scratch.
3. Broken Expressions Inside n8n #
Symptoms #
- a node shows empty values where data should exist
- folder names are wrong or blank
- updates fail because a field is missing
- image or audio nodes do not receive the expected input
Likely cause #
An expression is pointing to the wrong field, or a previous output changed after customization.
Why this happens #
This workflow uses expressions to:
- pull values from previous nodes
- pass AI outputs into audio and image nodes
- select rows
- create folder names
- update sheet values
If the user changes:
- a prompt structure
- an output field name
- a parser schema
- a node name
- a mapped field
the related expressions may stop working.
Fix #
Open the failed node and inspect:
- the actual data coming from the previous node
- the expression currently being used
- whether the referenced field still exists
Then update the expression so it points to the real current output.
4. AI Prompt Is Too Weak or Incorrect #
Symptoms #
- parser errors
- incomplete AI output
- missing title, image prompts, sound prompt, or caption
- badly structured content that breaks later steps
Likely cause #
The AI prompts are not detailed or strict enough, or the AI output no longer matches the parser schema.
Why this matters #
In this workflow, prompt quality is not only a content issue. It is also a workflow reliability issue.
If the prompt becomes vague or inconsistent, the output may no longer match what the parser and later nodes expect.
Fix #
Review:
- the idea generator prompt
- the video creation prompt
- the output parser schemas
Make sure:
- the prompt still instructs the AI to return the exact structure required
- the parser still matches the returned structure
- all required fields are being produced every time
Best practice #
When customizing prompts, test the AI output by itself before connecting it back into the full workflow.
5. Image Generation Fails #
Symptoms #
- one or more image nodes fail
- only some images are created
- the workflow stops before rendering
Likely cause #
One of the image prompts violates OpenAI image generation safety rules.
Why this is common #
The workflow expects 6 image prompts, and if even one prompt is rejected by OpenAI, the image generation section may fail and prevent the final video from being rendered.
Fix #
Review the image prompt rules inside the main AI agent and make them:
- safer
- less sensitive
- more moderation-friendly
Recommended approach #
Avoid prompts that may trigger:
- violence
- graphic harm
- politically sensitive content
- religiously sensitive content
- unsafe or disallowed imagery
This is especially important when adapting the workflow to a new niche.
6. JSON2Video Render Fails #
Symptoms #
- the render step fails
- JSON2Video returns an error
- the final video is not produced
- the returned result is incomplete or unusable
Likely cause #
The JSON2Video template or payload mapping does not match what the workflow is sending.
What this workflow expects #
The workflow depends on a compatible JSON2Video template that:
- accepts 6 images
- uses the audio track
- creates subtitles/captions from the audio
What to check #
- the template JSON was uploaded correctly
- the correct template ID is being used
- the workflow is still sending the expected number of images
- field names in the workflow still match the template structure
- the JSON2Video account and API access are working
Fix #
Re-check the template installation steps and confirm that the workflow payload still matches the template inputs exactly. If the user changed the image count or the render structure, both the workflow and the template must be updated together.
7. Publishing Fails #
Symptoms #
- YouTube upload does not start
- the publish section runs but nothing is uploaded
- posting status is not updated
Likely cause #
The publish stage does not have a valid row ready to process, or YouTube credentials are inactive.
What to check first #
Make sure there is at least one row where:
- Posting Status = Not Published
- Video URL exists and is valid
- the content was already produced
Also confirm:
- YouTube credentials are active inside n8n
- the account has permission to upload to the intended channel
Fix #
If no valid row exists, produce a video first or correct the sheet data. If the row exists, test the YouTube node and reconnect credentials if needed.
Troubleshooting by Workflow Stage #
| Stage | First Things to Check |
|---|---|
| Idea Generation | OpenAI credential, AI prompt, parser structure, Google Sheet connection |
| Video Creation | row selection, Google Drive folder setup, AI output, image generation, audio generation, JSON2Video template |
| Publishing | Not Published row exists, Video URL exists, YouTube credential is connected |
Best Practice for Recovery #
If something breaks, do not try to fix everything at once.
Better approach #
- identify the first failed node
- inspect its input and output data
- confirm that the previous step returned what the node expected
- fix one issue at a time
- retest only that section first
This prevents unnecessary changes and makes debugging much faster.
Most Important Reliability Rule #
The most important thing to keep stable in this workflow is the relationship between:
- AI prompts
- output parsers
- expressions
- code nodes
- sheet structure
- JSON2Video template
If one of these changes, the others may need to change too. In most failure cases, the workflow breaks because one of these layers is no longer aligned with the rest.
Troubleshooting Summary #
If something goes wrong, start by checking these six areas:
- credentials and API access
- Google Sheet structure
- expressions inside n8n
- AI prompt quality and parser alignment
- moderation-safe image prompts
- JSON2Video template alignment
In most cases, the problem will come from one of these areas. Once that part is corrected, the workflow can usually continue working normally.
