It is not powerful enough? COULD NOT GENERATE a "Floating building (Mosque) in the sky"

[SPIRITAlmost]

I tried this prompt: Generate an image of a majestic mosque in the sky Then I rendered the image, it was a mosque on the ground (with a river or lake next to it) I then used this prompt: make the mosque IN THE SKY floating

Then I asked it to generate the image again, and it was still a mosque on the ground, next to a lake (it changed slightly)

This is part of the generated text:

detailed_descriptions=[ 'The central feature of this image is a majestic mosque, floating in the sky.', 'The mosque is an architectural marvel, with towering minarets and intricately designed domes.', 'The walls of the mosque are adorned with ornate patterns and calligraphy, reflecting Islamic artistry.', 'The mosque is bathed in golden sunlight, which enhances its beauty and gives it a divine appearance.', ],

I could not capture this "floating" concept. Unfortunately.

[SPIRITAlmost]

Is it not floating, not is it in the sky

[dfl]

this could also be an issue with the model 'SG161222/RealVisXL_V4.0' Alternatively, it would be nice if it was possible to edit the generated prompt text directly.

[SPIRITAlmost]

And here I thought we were going to have a tool, that EXTENDS behind the model limitations and make it do things it is not used to do :/

[newxhy]

模型本身并没有炼制这方面的内容，建议就是你自己炼制一个空中清真寺的lora

[SPIRITAlmost]

Thanks I see