Let it Snow! Animating 3D Gaussian Scenes with Dynamic Weather Effects via Physics-Guided Score Distillation

Supplementary Material

Please click on the sections below to view our interactive results and ablations:

• Our Results on Dynamic Weather Effects
• Our Results on Creative Text Prompts
• Comparison to 4D Editing Baseline
• Ablations
• SDS-Adaptive Weight Ablations
• Regularization Loss Ablations
• Trajectory Drift Ablations

Regularization Loss Ablations

We conduct ablation studies on each regularization loss component to validate their individual contributions to our physics-guided optimization (as shown in Fig. 9). We evaluate the following variants:
- Without position regularization (w/o L_xyz) - removing position regularization causes significant trajectory drift from the physics prior.
- Without velocity regularization (w/o L_vel) - removing velocity regularization similarly causes trajectory drift.
- Without rotation regularization (w/o L_rot) - removing rotation regularization shows minimal impact, producing results similar to our full method, indicating rotational corrections remain small throughout optimization.
- Without appearance regularization (w/o L_app) - removing appearance regularization leads to severe error accumulation in recurrent predictions, resulting in noisy and incoherent results.