update teaser figure

README.md

@@ -17,6 +17,11 @@ Our FREUD design enables variable-length inputs, robustness to frame drops, and
 We enable **simple training** - no loss weight tuning, only a simple, stable rectified flow objective.
 A rectified-flow model in FREUD latent space achieves **state-of-the-art distributional and perceptual forecasting** quality.
 
-![Reconstruction distributions for different precipitation levels](https://compvis.github.io/weather-rf/static/images/teaser_figure_weather.svg)
+![Reconstruction distributions for different precipitation levels](./docs/teaser_figure_weather.svg)
+
 *Our generative decoder can quantify uncertainty about compression and covers the true precipitation in heavy-rain scenarios, while deterministic decoding collapses to incorrect modes.*
 
+<p align="center">
+  <img src="./docs/teaser_figure_weather.png" alt="Reconstruction distributions for different precipitation levels">
+</p>
+