Learning Timeline
Key Insights
Advantages of Picture-in-Picture (PiP)
The PiP technique helps the AI understand the 'end frame' or the look of the object being revealed. This is much more accurate than using text alone because the AI can directly reference the visual characteristics (like texture and shape) during the 'Outpainting' process.
The Challenge of 'Reverse' Movement
It is difficult to instruct AI to perform reverse movements (like tidying up a messy room). Therefore, the 'reveal' technique with a reference image is the best solution for showing transitions from one state to another.
Tips for Visual Consistency
If you want to 'pan' toward a specific object (e.g., the Titanic), include a reference image of that ship. This ensures the AI doesn't generate a random ship but instead follows the exact appearance provided in your reference.
Prompts
Prompt for Reveal/Uncover Technique
Target:
Google Veo
He uncovers this [reference image context]
Prompt for Camera Panning Technique
Target:
Google Veo
Remove this image and pan over to this ship [reference image of the ship]
Step by Step
'Advanced Motion Control' & 'Reveal' Techniques Using Reference Images
- Open the Google Veo interface via VideoFX.
- Upload the first image to serve as the 'Starting Image' (initial frame).
- Click the 'Upload Image' button again to add a second reference image (e.g., the object you want to reveal or the camera's final destination).
- Set the reference image as 'Picture-in-Picture' in the workspace to provide visual context for the AI.
- Write a prompt that connects the 'Starting Image' to the reference image, such as instructing the AI to 'reveal' or 'pan' toward the reference image.
- Click the 'Generate' button to start the 'Outpainting' process, where the AI will merge both images into one smooth camera movement.
