Higgsfield: Generating character reference images with Nano Banana
Higgsfield
Demo summary
The user demonstrates using the Nano Banana model on Higgsfield to generate a character reference image from a video frame to ensure pose and composition consistency.
Step-by-step
- Select either the Nano Banana Pro or Nano Banana 2 model
- Upload the first frame of your video as a reference image
- Enter a text prompt describing the character
- Include the phrase 'keep the same pose and composition' in the prompt
Options
- Use any of the available image models on Higgsfield
- Use Nano Banana Pro
- Use Nano Banana 2
Tips
- Explicitly mention pose and composition in the prompt to help Kling Motion Control match the original video movement
Highlights
“easy enough to do using any of the image models on Higsfield”
All demos from “I Pushed Kling 3.0 Motion Control Further Than Anyone Else”
- 0:210:16Accessing Kling Motion Control on HiggsfieldThe creator shows the Higgsfield interface, navigating to the video tab and explaining how to upload a source motion video and a target character image.Higgsfield· Video to Video
- 2:000:19Generating character reference images with Nano BananaCurrentThe user demonstrates using the Nano Banana model on Higgsfield to generate a character reference image from a video frame to ensure pose and composition consistency.Higgsfield· Image to Image
- 4:240:30Configuring Scene Control and Advanced Settings in KlingThe video shows how to toggle Scene Control mode and use advanced text prompts to influence the AI's generation, such as adding a 'driving fast' effect.Higgsfield· AI Animation Generator
- 12:410:40Adding foreground elements with Kling EditThe user demonstrates Kling 3.0 Omniedit by uploading an existing video and using a text prompt to seamlessly add foliage into the foreground.Higgsfield· AI Video Editor
- Watch “I Pushed Kling 3.0 Motion Control Further Than Anyone Else” →
Image to Image
2:540:55Generate a 4K thumbnail with Higgsfield GPT Image 2The user uploads a wireframe and reference photos to Higgsfield, selecting the GPT Image 2 model to generate a cinematic 16:9 YouTube thumbnail.Creating with Conor
10:540:57Generate consistent AI environments with HiggsfieldThe creator demonstrates how to use a reference image in Higgsfield to maintain visual consistency across multiple AI-generated images of a dog salon.Jack Craig
1:14:101:15Draw-to-Image generationThe user sketches a rough landscape of mountains and a sun on a blank canvas, which Higgsfield AI then renders into a realistic image.Artturi Jalli- 2:000:19Generating character reference images with Nano BananaCurrentThe user demonstrates using the Nano Banana model on Higgsfield to generate a character reference image from a video frame to ensure pose and composition consistency.Atomic Gains