Thanks for your work and released code. The idea of direct inversion is pretty amazing but I can not see if there are any difference between your direct inversion and DDIM inversion approach? Also, you did claim in your paper that: "Given these inputs, our method is able to elabo- rately modify the original image using only the target edit text prompt – adding objects, as well as modulating scene, background, pose, style, color, and even race/ethnicity", but there are no relevant results demonstrated in your paper. It would be great if you could show successful editing scenarios as u did claim in your work. Thanks.
Hi,
Thanks for your work and released code. The idea of direct inversion is pretty amazing but I can not see if there are any difference between your direct inversion and DDIM inversion approach? Also, you did claim in your paper that: "Given these inputs, our method is able to elabo- rately modify the original image using only the target edit text prompt – adding objects, as well as modulating scene, background, pose, style, color, and even race/ethnicity", but there are no relevant results demonstrated in your paper. It would be great if you could show successful editing scenarios as u did claim in your work. Thanks.
Best.