Transform a single image into dynamic multi-scene AI video using specialized scene, pose, and action LoRAs. Choose a matching High Noise + Low Noise LoRA pair, select a preset prompt, and generate a short sequence while maintaining the subject and overall visual identity of the source image.
For best results, start with a close-up portrait or square image with the subject clearly visible and centered. Simpler compositions generally give the model more room to create convincing scene changes and motion.
Recommended Settings
Duration: 4-6 seconds
Resolution: 480p / 720p
Steps: 6-10
Shift: 6-8
Primary High / Low LoRA Strength: 0.7-1.0
Universal Low Branch Detail LoRA: Enabled
Input: Close-up portrait or square image
Always select the matching High Noise and Low Noise LoRA for the primary effect you want. Both selections should use the same numbered LoRA pair.
Preset prompts are designed specifically for their corresponding primary LoRAs and are recommended as the starting point. Advanced users can enable Custom Prompt to replace the preset prompt with their own instructions. When combining two LoRAs, the preset prompt is still based on the primary LoRA, so custom prompting may be useful when you want to explicitly incorporate characteristics of both.
The optional Universal Lora applies additional detail and refinement to both MiniMax and the low-noise WAN branch and is recommended for most generations.
Higher resolutions, longer durations, additional steps, different LoRA strengths, a second LoRA pair, and custom prompts can produce different results, but may reduce consistency depending on the selected LoRAs and source image.
USE AT YOUR OWN RISK
By using these AI Tools, you agree they are provided "as is" without warranty and assume full liability for all inputs and outputs. ObscuraLabs disclaim all responsibility for any claims, damages, or legal disputes arising from your use, including intellectual property, privacy, or defamation matters. Tools are restricted to fictional characters ONLY—any use involving real persons is prohibited and voids all creator liability. Any sample videos were made using fictional AI-generated characters.