Higgsfield: Generate subtle human emotion with Kling 3.0
HiggsfieldDemo summary
Kling 3.0 is shown performing a text-to-video generation of a young woman shifting from joy to concern, demonstrating realistic facial micro-expressions.
Step-by-step
- Enter a complex prompt describing a specific emotional shift (e.g., joy to concern)
- Select the Kling 3.0 model
- Run the text-to-video generation without providing a starting reference frame
- Review the output for facial micro-expressions and transitions
Options
- Provide a start frame for guided generation
- Perform a pure text-to-video generation without image inputs
Watch out for
- Generating subtle human emotion is a common struggle for AI models
Tips
- Use highly detailed prompts to describe the specific transition of emotions you want to see
- Test the model's capabilities by withholding a start frame to see how it handles pure text-to-video
Highlights
“Out of the gate, this is just [music]”
Text to Video
2:50:294:47Generating emotional character performancesThe user demonstrates prompting Higgsfield AI for specific emotional physical cues, such as a trembling lip and sobbing, to create a dramatic character breakdown.Higgsfield AI
2:000:41Text-to-video with multi-shot in Kling 3.0The user demonstrates generating a dynamic military action scene using a text prompt and the new multi-shot feature in Kling 3.0 to create automatic scene cuts.Higgsfield AI- 5:050:49Complex camera movement controlThe user prompts a robotic arm scene with specific lateral passes, crash pushes, and pull-backs to demonstrate Kling 3.0's improved camera physics and motion smoothness.Higgsfield AI
2:160:42Text-to-Video with SeaDance 1.5 ProThe video shows how to configure settings like aspect ratio, clip length, and audio in SeaDance 1.5 Pro to generate a video from a text prompt.Malva AI
5:110:30Generate cinematic locations with HiggsfieldThe creator shows how to use the Location option in Higgsfield Cinema Studio to generate a cinematic background with specific lighting and camera angles based on a short text prompt.Youri van Hofwegen
3:360:49Generating AI video with tagged character referencesThe creator shows how to paste a multi-shot prompt into Higgsfield and use the '@' symbol to tag specific character images (detective and suspect) before generating the final video.Creative Pad Media- 1:260:24Generate subtle human emotion with Kling 3.0CurrentKling 3.0 is shown performing a text-to-video generation of a young woman shifting from joy to concern, demonstrating realistic facial micro-expressions.Joseph Martin
8:344:13Generate an atmospheric establishing shot with Seedance 2.0A live demonstration of generating a 15-second video of a 'Storm Manor' using a text prompt, showing the real-time wait and the final rendered output featuring lightning and darkness.JSFILMZ