fal.ai: Reference-to-video with high controllability
fal.aiDemo summary
A demonstration of the Kling 03 reference-to-video model, combining a start image, a character element, and a text prompt to generate a specific scene.
Step-by-step
- Select the Kling Video 03 reference-to-video model
- Upload a start image to define the initial frame
- Upload a character image into the element field for reference
- Enter a text prompt describing the interaction between the start image and the element
- Click to generate the resulting video
Options
- Use multi-prompt functionality
- Upload an end image to define the final frame
- Add additional images for reference
- Use frontal images for full element functionality
- Upload a video of your character or subject as a reference
Highlights
“this model is probably the one with the highest degree of controllability”
All demos from “Introducing Kling 3.0 - A New Frontier for AI Video | fal Academy”
- 0:120:41Text-to-video generation with Kling 3.0The user demonstrates generating a multi-shot action movie scene from a single text prompt using the Kling 03 text-to-video playground on fal.fal.ai· Text to Video
- 1:310:28Complex motion generation in Kling 3.0A demonstration of Kling 3.0 handling complex physical interactions and character dialogue in a basketball scene using a detailed multi-step prompt.fal.ai· Text to Video
- 2:170:59Image-to-video with character consistencyThe user demonstrates Kling 03 image-to-video by providing a starting frame and a dialogue prompt to generate a cinematic scene with consistent character styling and camera cuts.fal.ai· Image to Video
- 4:210:23Cinematic camera and lighting controlA demonstration of precise camera control using a 'dolly zoom' prompt and a blue lighting shift to reflect character emotion in Kling 3.0.fal.ai· AI Animation Generator
- 5:280:52AI Video Editing with Image ElementsThe video shows how to edit an existing video by referencing an external image element (Popeye) to replace a character while maintaining the original video's structure.fal.ai· Video to Video
- 6:290:40Character replacement with reference imagesThe user demonstrates replacing two skydivers in a video with specific characters defined by frontal and reference images using Kling 03 video editing.fal.ai· Video to Video
- 7:090:49Reference-to-video with high controllabilityCurrentA demonstration of the Kling 03 reference-to-video model, combining a start image, a character element, and a text prompt to generate a specific scene.fal.ai· Image to Video
- Watch “Introducing Kling 3.0 - A New Frontier for AI Video | fal Academy” →
Image to Video
- 2:170:59Image-to-video with character consistencyThe user demonstrates Kling 03 image-to-video by providing a starting frame and a dialogue prompt to generate a cinematic scene with consistent character styling and camera cuts.fal
- 7:090:49Reference-to-video with high controllabilityCurrentA demonstration of the Kling 03 reference-to-video model, combining a start image, a character element, and a text prompt to generate a specific scene.fal