https://civitai.com/user/kanto (Kanto's images have prompt. Feel free to use it for your image.) Trigger word : detailed city Best practice to upscale the image,fix broken crowds,silhouette,people shape,etc Trained on replicate steps : 1000 lora_rank : 32 optimizer : adamw8bit batch_size : 1 Recommended weight 0.8 - 1.0 1.0 is good to start with must include trigger word : detailed city It is best practice to upscale after generating the image since this lora requires very granular details and will use all pixel wisely. The image size is trained on 768 x 1344 pixels, so expect vertical orientation but feel free to use it on other ratio (I have not tried with other ratios yet. Let's explore them together) Not only city scape image, Detailed city lora can be applied to character scene as well. Try giving it a shot! Please post your image. I would love to see them and would love to get your feed back.