# WAN2.2 Space Magic - Spatial Rift & Bending Effects
Unleash the power of spatial magic! This LoRA is designed to create surreal and magical effects, splitting landscapes and architecture to reveal something new and fantastical.
It excels at generating harmonious images even after the "space-splitting" effect, making it perfect for creating stunning, reality-bending scenes. While it performs exceptionally well with architecture, it can also be used effectively on natural landscapes and other subjects.
The primary goal of this LoRA is to help you create incredible transition effects and visually striking moments of transformation.
## 📖 Usage Guide
- Base Model: This LoRA is trained and tested on the WAN2.2 base model (specifically wan2.2_high).
- Trigger Word: The main trigger word is kjmf magic.
- Recommended Weight: 1.0 is recommended for the strongest effect. You can adjust it to your liking.
### ✨ Prompting Tips
This LoRA responds well to descriptive, narrative prompts that guide the AI through the transformation. You can use sequential phrases to build up the scene.
- Main Trigger: kjmf magic
- Action Phrases: To create a sense of action or sequence, you can include phrases like:
- start casting kjmf magic
- when the kjmf magic gesture is complete
Example of a Detailed Prompt:
> a beautiful outdoor landscape, a field of blooming purple flowers in the foreground, modern city skyline in the background. A pair of hands appears, slowly opening in front of the flowers, they start casting kjmf magic, as if interacting with the viewer. When the kjmf magic gesture is complete, a miraculous scene unfolds: the building in front of them precisely splits in half along its central axis, moving smoothly and parallel to both sides. In the empty space left between the two separated building halves, new green tracks are born, perfectly connecting the neat bushes on both sides. The tracks extend into the distance, flanked by colorful arches, with skyscrapers visible in the background.
## 💬 Additional Notes
This is the first version of the model. I'm excited to see what you create with it! Please share your results and provide feedback—it would be greatly appreciated.
☁️ No powerful GPU? No problem!
You can try this model on RunningHub. I've set up a basic workflow there to help you get started quickly. You can use it directly or modify it for your own creations.
https://www.runninghub.ai/post/1956032290172403714?inviteCode=rh-v1139
Free to generate! They give you 1,000 credits, which lets you make about 15 videos for free.
Hope you all have fun generating some awesome transitions!
Description
For V1: Characters boast greater flexibility, though dimensional magic excels on architecture but lacks versatility in other scenes.
For V2: Dimensional magic achieves superior generalization across scenarios, but character articulation adaptability diminishes.
*Select the version that best suits your creative priorities.
FAQ
Comments (17)
Heads-up: This model connects to wan2.2_itv_high_noise
For V1: Characters boast greater flexibility, though dimensional magic excels on architecture but lacks versatility in other scenes.
For V2: Dimensional magic achieves superior generalization across scenarios, but character articulation adaptability diminishes.
*Select the version that best suits your creative priorities.
it's loras like this that can create incredible potentials for highly imaginative ideas. Great job. I hope to see more inventive stuff like this.
Wow, thank you!! 😊 That's exactly the kind of feedback that makes all the hard work worthwhile! Thrilled to hear it's unleashing your imaginative ideas – that potential for creativity is why I love building these. And you bet, more 'inventive stuff' is cooking! Super grateful for your support!
Invisible_vortex thats what this is all about. Ive always had ideas in my mind about situations where this type of visual is an exact match for something i wanted to create, however there was no way for me to get that expression across. Thanks for bridging that gap.
whitespider9999 This warms my heart! 😊 Truly honored to help bridge your vision. Happy creating! ✨
I've been thinking about old vacation photos, I2V them with some 360 motion maybe something like this. A way to turn old memories into new ones, blow the minds of your older family members etc.
randomthug875 Yo that's a cool idea! Actually, there's this tool from Tencent called Hunyuan 3D (link: https://3d.hunyuan.tencent.com/sceneTo3D) that can turn your pics into 360 scenes. Perfect for making photos move
can you not have hands inside the videos? Just curious
Amazing idea! I couldn't believe I hadn't thought of that creative concept. I tested two model versions, and both can achieve the effect you described by simply adjusting the prompt words. However, Version 2 performs much better than Version 1. I've shared the video results in the feedback section.
Invisible_vortex I like the idea you have. I wanted to make where moses seperates the water. How ever I need to make it look like the oceans looks more less square and have oceans waves splashing upward. Because of this lora file...I feel you made my idea become a reality. :D
AIBeauty4Seen It's currently difficult to achieve a perfect ocean wave effect, but I'm exploring solutions.
Invisible_vortex try to add "chaotic storm of liquid ___ "
for example, chaotic storm of liquid smoke = work good from sd 1.5
chaotic storm of liquid kanagawa waves , seems to good here
Polymath_wtf Thank you very much for the tip! I have been experimenting with different prompts and workflows. While the results have shown some improvement, achieving satisfactory outcomes still requires a significant number of sampling attempts. Therefore, I believe the fundamental solution lies in improving the dataset itself. I am planning to create and collect a more diverse and richer dataset for retraining the model.
Polymath_wtf Will try that