Generations may fail or take longer than expected for several reasons:
Queue and system load
Free generations are processed after paid jobs during peak hours. Subscribers and users with credits receive priority processing.
Connection problems
Unstable internet connections can interrupt file uploads or generation progress.
Input quality issues
Low-quality inputs produce poor results. Common problems include:
Unclear or contradictory prompts
Low-resolution images or video
Blurry, poorly-lit, or off-angle footage
Faces that are too small, obscured, or in profile
Non-human subjects (less reliable)
Account limits
Free accounts have restrictions on duration, resolution, and include watermarks. Generations will output an error when you click render if you've exhausted your available credits.
Content moderation
Generations that violate our content policies are automatically blocked. Learn more about content guidelines
Browser compatibility
Ad blockers, script blockers, outdated browsers, or aggressive privacy settings may interfere with uploads and previews.
Troubleshooting steps
Follow these steps in order for the best chance of resolving your issue:
1. Check your account status
View your available credits and plan limits in the top navigation bar.
2. Retry your generation
Server issues occasionally cause random failures. Retry once before troubleshooting further.
3. Switch models or modes
Different models handle different content better:
Lip Sync: Toggle between Standard (faster) and Pro (better quality)
Face Swap: Try Face Swap v1 for challenging scenes
Text-to-Video / Image-to-Video: Switch between available models if you see artifacts
4. Improve your inputs
For images:
Use high resolution (1080p or higher recommended)
Ensure good lighting and focus
Center your subject in the frame
For video:
Use frontal, well-lit faces when generating face content
Avoid motion blur and shaky footage
Keep subjects large in frame (30-70% for faces)
Use H.264 MP4 format at 24-30 fps
Trim unnecessary frames from start and end
For prompts:
Be specific and descriptive
Remove contradictory instructions
Structure prompts clearly: subject → action → setting → camera → style
Keep prompts concise but detailed
5. Generate shorter clips
AI performs best on 5-10 second clips. For longer content, generate multiple short segments and combine them in a video editor.
6. Try a different browser
Switch browsers or disable ad blockers and privacy extensions. Clear your browser cache if previews appear stuck.
Expected processing times
Processing times vary based on queue load, model complexity, and generation length:
Images: Under 1 minute (typically)
Short video (5-10s): 30 seconds to 2 minutes
Longer video: Increases with duration and resolution
These are typical ranges and not guaranteed. Peak hours may result in longer wait times for free users.
Tool-specific guidance
Face Swap
Best practices:
Use clear, frontal faces occupying 30-70% of the frame
Avoid sunglasses, heavy shadows, extreme angles, or obstructions
Ensure good, even lighting
If results look distorted:
Switch to Face Swap v1
Use footage where the subject faces the camera more directly
Crop tighter around the target face
For multi-person scenes, ensure your target face is dominant and well-lit
Note: Non-human faces and artistic styles are less reliable than human faces.
Lip Sync
Mode selection:
Standard: Faster processing, good for straightforward footage
Pro: Better quality, handles difficult footage
Audio requirements:
Clean speech with minimal background music
Remove long silences at start and end
Video requirements:
Front-facing subject with clearly visible lips
Stable footage (minimal head movement)
Trim clips to 5-10 seconds for best results
If timing drifts: Trim silence from the beginning of your audio or use footage with less subject movement.
Text-to-Video & Image-to-Video
Prompt structure: Use this format for best results:
Subject: What is in the scene ("close-up of a basketball player")
Action: What is happening ("drives left, executes step-back jumper")
Setting: Where and when ("packed arena at night")
Camera: How it's shot ("handheld, shallow depth of field, slow push-in")
Style: Visual aesthetic ("gritty, cinematic, high contrast")
Constraints: What to avoid ("sharp details, no text, stable lighting")
Tips:
Start with 5-8 second generations, then extend if needed
Avoid contradictory descriptions ("noisy yet ultra clean")
Don't mix incompatible genres or styles
Switch models if you see flickering, melting, or unwanted stylization
Video-to-Video (Stylize/Transform)
Source video tips:
Use steady camera footage
Ensure consistent exposure throughout
Avoid motion blur
Keep subjects centered and large in frame
If the output loses detail or identity:
Use brighter, steadier source footage
Generate shorter clips for action scenes
Image-to-Image
If your output doesn't match expectations:
Increase resemblance setting to stay closer to source
Lower creative strength for more subtle changes
Add negative prompts to avoid specific unwanted elements
Use the highest quality source image available
AI Image Upscaler
Upscaling magnifies existing details—including defects. Always start with the sharpest, highest-quality source image you have.
Best practices:
Try 2× upscaling first, then 4× if results are good
Available to all signed-in users with credits (no plan upgrade required)
AI Headshots
Photo requirements:
Upload 8-12 photos of the same person
Include varied angles and expressions
Use well-lit, high-quality images
Avoid heavy filters, sunglasses, and extreme side profiles
QR Code AI Art Generator
For scannable results:
Preserve quiet zones (blank borders) around the QR code
Maintain strong contrast between light and dark elements
Test with your phone camera during generation—highly decorative designs may not scan reliably
Note: Daily usage limits apply
Quick reference: Common issues
Problem | Solution |
Lip sync looks off on profile angles | Switch to Pro mode and use more frontal footage |
Face swap misses the subject | Switch to Face Swap v1 and crop tighter on target |
Text-to-video doesn't get my prompt | Try using a more detailed prompt that just mentions a single motion, shorten to 6-8s |
Creating longer content
"Can I generate a full movie or long video?"
AI video generators work best for short scenes (5-10 seconds). For longer content:
Break your concept into individual shots
Generate each shot separately (5-10s each)
Combine clips in a video editor
Maintain consistency across shots by using similar prompts (subject, color palette, camera style)
Use reference frames to maintain visual continuity between shots
Still need help?
If you've tried these steps and still experience issues, contact support using the in-app help icon.
Please include:
Job ID (found in your generation history and URL of your generation)
Screenshots or screen recordings showing the issue
What you've tried
This information helps us diagnose problems faster and provide accurate solutions.