Hi-res Fix does massive help for those small appearances. You already know that AI is terrible at making things small.
Hello again. Another composition layout. This time is a window layout.
It will try to create something like this and the environment will be blended in.
It will generate a large window this time. You will need to intentionally prompt it to include some kind of window. A “big window” does not have to be stated directly in the prompt, but there needs to be something that triggers the AI to draw one. It could be phrases like “huge billboard,” “inside and outside view,” “massive frame,” “large display panel,” or “oversized window.”
You may need to think beyond just “Dome (Layout1).” It does work if you describe the layout clearly enough. If you don’t, the AI will generate random results.
Trigger: layout2
Helpful Tip:
Describe a wide display or a large framed opening.
Add “close-up” if you want the character to appear closer to the camera.
Think outside the box.
Talk to Chatgpt to make a prompt for you. Show the picture to it.
Lower weight might be helpful. It is pretty strong if prompted correctly.
Extra Thought:
Close-ups can create a very interesting effect and might even cause a placebo result.
AI is always terrible at drawing small things. It is not this LoRA's fault. It is an AI issue. Use face restoration if you can. You can also try to close up somehow. This will be hard.
I hope to see people get creative with this. I understand it might be difficult, but I promise it works. I usually use a strength of 0.8 most of the time.
You can use this LoRA for any creative idea you have in mind. I’ve tried to post as many images as I can to give you an idea of how to use it. I hope it helps.