Replicas serve as your digital talking head, and we offer a choice between utilizing our curated selection of stock replicas or crafting a unique, personalized version. Opting to create your own replica is highly recommended for a tailored experience. To do so, you’ll need to submit a training video. Our system leverages this video, applying extensive computational power to generate a personal replica tailored to you. But before you send us a training video it’s important to get a few things right. To uphold our commitment to ethical AI practices and ensure compliance with our terms of service, we require a consent video for each individual featured in the training videos provided to our AI generative video creation API. This consent video is a mandatory step in our submission process. Why Recording Consent Required? A consent video serves as a clear, recorded agreement from the individuals featured in the training videos, authorizing the use of their likeness for generating AI content. This step is crucial for:
  • Protecting the rights and privacy of individuals.
  • Ensuring that our AI models are trained ethically and responsibly.
  • Maintaining the integrity of our services and adhering to legal standards.
What Should the Consent Video Include? The consent video must feature the individual giving explicit permission for their likeness to be used in AI-generated content. It should include:
  • A clear verbal statement of consent.
  • Identification of the individual providing consent.
  • Acknowledgment of understanding what the consent is for.
Review Process Each consent video will be reviewed by our team to verify that consent has been adequately given and meets our requirements. Only upon approval will the associated training videos be used in our AI model training process. Please ensure that the consent video is included with your training video submissions to avoid any delays in processing. For detailed guidelines on creating a consent video, refer to our submission guidelines section.

​
Training Video Format, Parameters, Size

Replica training videos can be submitted in MP4 format only. Additional specifications:
  • Length: 1 to 2.5 minutes
  • File Size: Less than 750 MB
  • Resolution: We accept up to 4k, however more common webcam resolutions (such as 720p/1080p) are also known to produce excellent replicas.
  • Frames Per Second: 23.98 - 60 FPS
  • Codecs:
    • .MP4
      • H264

​
Guidelines for Training Video Submission

For the best replica creation, it’s essential to submit your training video in its most authentic form. We advise against using edited, cut, or otherwise processed footage. Our models perform best when trained with genuine, unaltered video content of you. To ensure high-quality replicas, please avoid:
  • Jump cuts
  • Audio enhancements or modifications
  • Adding titles or captions
  • Implementing transitions
  • Any form of video editing
Submitting raw, unedited footage will help in creating a more accurate and high-quality digital replica.

​
Video Generation Limitations

You can use your replicas (or ours) to generate videos. Those videos look like this: Our system is capable of generating videos of any duration. However, for optimal quality, we recommend keeping videos to around 5 minutes or less.

​
Video Generation Limits

You can generate videos using scripts, in multiple languages, of varying length and subject matter. There are a few limitations to keep in mind.
  • Script Length: