Troubleshooting
Preserve outfits and backgrounds
Preserve outfits and backgrounds: understand what the uploaded image controls and what can still vary.
Written and maintained by Chinaza Filter · Updated
Use one consistent reference
Your image is the sole visual reference for the people, outfits and location. The generation request asks to preserve clothing colors, patterns and fit. Clear lighting, visible clothing and separated bodies provide better information than blur, occlusion or heavy image filters.
Separate the scene from the dance
The built-in video provides gesture order and rhythm. The model is told to ignore that video’s clothing, actors, scenery and camera movement. It is also told not to add people and to keep the uploaded background stationary.
Review before retrying
Inspect the entire video for appearance changes, hand artifacts and background motion. These instructions are goals, not guarantees. If you retry, improve a specific input problem and compare the results. A completed clip that differs from your expectations is not automatically a failed generation.
What else should I know?
Is the background locked pixel for pixel?
No. This is generative video, not a compositing editor with a protected background layer. The request explicitly asks for the original scene to stay still and the camera to remain fixed. Objects, shadows and textures can nevertheless change. Check the result over its full duration instead of comparing only the first frame.
Will clothes come from the motion video?
The request tells the model to use the uploaded image for clothing and ignore the reference performers’ outfits. Show the garment colors, shape and accessories clearly in your image. Changes can still occur when arms overlap clothing or the model invents hidden areas. A successful request does not guarantee exact garment preservation.
Is generation free, and do I need an account?
There are 0 free generation credits by default. No account is required for guest checkout: enter a delivery email and use the private result page. Signed-in users can buy account packs and return to My videos. Check the pricing page before paying because guest and account packs have different prices and allowances.
What file and output settings are supported?
Upload one JPG, PNG or WebP image up to 10 MB. Choose 480p for 20 credits or 720p for 45 credits, with 9:16 portrait or 16:9 landscape. The output is a 13-second MP4 with generated pop audio and no site-added watermark. A higher resolution does not guarantee more accurate movement.
Where is my image processed and saved?
Your image is sent to the AI service provider to generate the video; processing does not happen entirely in your browser. This browser may retain a local upload draft for 24 hours after the last save. Download the finished MP4 promptly because provider links may expire. Read the Privacy Policy for data handling details.
When should I retry a generation?
First check the existing task. Pending or processing means you should wait; a temporary status error is not a confirmed failure. Credits return when generation is confirmed failed. A new attempt consumes credits, including when the previous video completed but its appearance or movement did not meet your expectations. Contact support for unresolved billing issues.