Based on StableDiffusion "CompVis/stable-diffusion-v1-4" pre-trained model, Docker Image : runpod/stable-diffusion.
Novel view synthesis based on Diffusion text embedding space.
Validation set Best metric by textual inversion Average LPIPS : 0.067
Metric of Synthesised novel view by our method Average LPIPS : 0.16
Result Example
Embeddig space example image
