add arxiv link

app.py

@@ -236,7 +236,7 @@ if __name__ == "__main__":
         outputs=gr.Model3D(label="Generated Scene"),
         allow_flagging="never",
         title="Invisible Stitch: Generating Smooth 3D Scenes with Depth Inpainting",
-        description="Hallucinate geometrically coherent 3D scenes from a single input image in less than 30 seconds.<br /> [Project Page](https://research.paulengstler.com/invisible-stitch) | [GitHub](https://github.com/paulengstler/invisible-stitch) | [Paper](#) <br /><br />To keep this demo snappy, we have limited its functionality. Scenes are generated at a low resolution without densification, supporting views are not inpainted, and we do not optimize the resulting point cloud. Imperfections are to be expected, in particular around object borders. Please allow a couple of seconds for the generated scene to be downloaded (about 40 megabytes).",
+        description="Hallucinate geometrically coherent 3D scenes from a single input image in less than 30 seconds.<br /> [Project Page](https://research.paulengstler.com/invisible-stitch) | [GitHub](https://github.com/paulengstler/invisible-stitch) | [Paper](https://arxiv.org/abs/2404.19758) <br /><br />To keep this demo snappy, we have limited its functionality. Scenes are generated at a low resolution without densification, supporting views are not inpainted, and we do not optimize the resulting point cloud. Imperfections are to be expected, in particular around object borders. Please allow a couple of seconds for the generated scene to be downloaded (about 40 megabytes).",
         article="Please consider running this demo locally to obtain high-quality results (see the GitHub repository).<br /><br />Here are some observations we made that might help you to get better results:<ul><li>Use generic prompts that match the surroundings of your input image.</li><li>Ensure that the borders of your input image are free from partially visible objects.</li><li>Keep your prompts simple and avoid adding specific details.</li></ul>",
         examples=[
             ["examples/photo-1667788000333-4e36f948de9a.jpeg", "a street with traditional buildings in Kyoto, Japan"],