I tried experimenting removing people or object from the images. It requires prompting with much details to get appropriate image but the outputs are precise and really good. The only concern is that it compress the image on higher level which makes the images dull and blur. This can be an improvement scope.
Hi @Kaleen_Bhaiya, Welcome to the forum!!!
Thanks for your feedback. Since this model is still an experimental model, will try to improve by gathering feedbacks. I will raise this with internal team.
Thanks.
If image quality is your top priority then you can go with Imagen 3. Here is the doc you can follow.