The images uploaded rarely keep the same facial identity to allow for a continuation of the scenes. The viewer realises that the person in the image is not the same person in the following scenes even if the prompt states " Keep identical facial features or Keep identical face of person for the next scene".
This definately affects the realism of the video being generated negatively.
Thank you for sharing this feedback. We understand how important facial consistency is when building scene-to-scene continuity. At this time, prompts such as “keep identical facial features” don’t always guarantee the same identity across multiple generated images or frames. This could be a known limitation of current generative models. In the meantime, many users find success by working with reference images.
We truly value your input. It helps guide future development and enhancements. Please continue to share your experience, as it helps us prioritize features that matter most to you.
^KR
moved to discussions
Already have an account? Login
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.
AdChoices