Spark is a next-gen video understanding large language model that understands every aspect of your video, from podcasts to sports events, game streams, and vlogs. It identifies key moments, actions, emotions, and sound cues in your video, making it possible to clip videos with little to no dialogue.
Here's how it works:
Understand your video
Visual Analysis: Identifies objects, scenes, characters, and visual cues within each frame.
Audio Analysis: Analyzes audio content, including speech, music, and sound effects.
Sentiment Analysis: Detects emotions and sentiments expressed in the video.
Clip with your prompts
You can find specific moments or highlight the best moments from your video, just type your prompts. Automatically extract and order the most relevant moments based on your prompts.
To learn more about how to use prompts to find specific moments, please go to How to prompt with Spark 1.0.
Spark 1.0 FAQ
Is it possible to prompt again and regenerate clips from the same video?
Absolutely! If you're not satisfied with the initial results, feel free to resubmit the AI clipping request with a different prompt.
Each project allows for up to 10 prompts. Experimenting with various wordings and descriptions can help in discovering the ideal clips for your needs.
What is the pricing for Spark?
During the Beta testing phase, Spark is available to all users with access to the model. Re-prompting does not require extra credits.
Please note that while re-prompting is free, upload minutes will still be consumed when creating new projects.
Why I'm not seeing 'Spark 1.0'
Some users may not currently have access to the 'Spark 1.0' model. Please contact our support team to apply for access.