What prompts can I use for layout-aware t2i generation

[carvychen]

What prompts can I use for layout-aware t2i generation to reproduce the same result in your paper?

[zyxElsa]

Layouts are generated in the early stages of generation, corresponding to lower-numbered prompt.
Please try to add the token embedding of the input image in the 1st, 2nd and 3rd prompt position
e.g.

prospect_words = ['strawberry cup cakes',  # 10 generation ends\ 
                       'strawberry cup cakes',  # 9 \
                       'strawberry cup cakes',  # 8 \
                       'strawberry cup cakes',  # 7 \ 
                       'strawberry cup cakes',  # 6 \ 
                       'strawberry cup cakes',  # 5 \
                       'strawberry cup cakes',  # 4 \
                       'strawberry cup cakes',  # 3 \
                       '*',  # 2 \
                       '*',  # 1 generation starts\ 
                      ], \

[zyxElsa]

and thanks for your question, I have also improved the README:)

[AnkitSinha123]

Can you let me know the approach for , when I have two different inversion and want to make it layout aware , material aware by mixing two prompts ?