I am trying to automate the interior design for architectural designers. And I wanted to generate 3D renders from an existing 2D plan. And I researched about this, and it said that AI lacks spatial awareness, and I should build it myself. It suggested that I build my own LLM to actually make the model learn how to step by step make the model identify the spatial arrangement and all before actually producing a 3D render. Is there any other way to do this, or are there any trained models? I have checked the Hugging Face, but I couldn't find a suitable model for this. Any kind of input or advice would be really helpful.