Troubleshooting
OAuth fails during platform connection
Symptoms:
- callback returns with an error
- platform still shows disconnected
- scheduled publishing cannot start
What to check:
- finish the auth flow in the same browser session
- confirm the redirect path matches the settings page you started from
- reconnect the platform and retry after the callback completes
Publishing fails with token_expired
Symptoms:
- scheduled post moves to failed or stays pending
- content remains approved
Fix:
- Reconnect the platform in Settings.
- Confirm the connection health is good.
- Retry publish-now or wait for the next scheduled attempt.
Publishing fails with rate_limited
Most common causes:
- platform quota window exhausted
- too many retries in a short period
Fix:
- wait for the rate window to reset
- check YouTube quota or Pinterest daily limits specifically
- avoid bulk forcing publish-now on the same platform repeatedly
Generated face looks inconsistent
Most likely causes:
- weak reference images
- prompt drift
- too much variation in styling direction
Fix:
- replace filtered or low-quality references
- tighten the prompt
- review similarity failures before publishing
- train a LoRA only if the character already justifies it
LoRA training says face score is too low
Cause:
- one or more uploaded images failed the face-similarity threshold
Fix:
- remove weak images
- use cleaner front-facing photos
- keep age, lighting, and styling direction consistent
Voice clone sounds unstable
Cause:
- noisy source audio
- inconsistent speakers or accents
- trying to fix a bad sample set with tuning knobs
Fix:
- replace the samples first
- then adjust stability and similarity
Motion transfer fails before it starts
Most common causes:
- no reference image on the character
- insufficient credits
- LoRA requested but not ready
- invalid reference video upload
Fix:
- add at least one character reference image
- confirm credit balance
- disable
use_loraunless the LoRA is ready - make sure the uploaded file is actually a video
Credits disappear faster than expected
Common explanation:
- repeated regeneration on unstable characters
- heavy workflows like LoRA or motion transfer
Fix:
- reduce exploratory volume
- stabilize the character before scaling output
- buy a pack for bursts instead of blindly regenerating the same idea