Man, the problem of image quality in images made with artificial intelligence is not mine alone. Most image generators generate image with the quality of 1 megapixel.
Read this for more explanation:
"Creating small images with precision and then enlarging them not only results in amazing resolution, but also affects the amount in the image.
Let's say for example:
Image with a resolution of 5000 * 5000: It will contain specific details, including window building details such as glosses and patterns.
Image 1000 * 1000: It will contain real details, windows and will appear as simple lines only.
When a low resolution image (1000*1000) is enlarged to 5000*5000, the image area will increase, but new details that were not present in the original image will not be measured.
Simply put, the image will become larger, but it will shade out even the finer details you capture in the (5000*5000) image."
I also contacted support and they told me that my main problem is the lack of sharpness and poor quality of the images, so I am trying to improve that, hoping for your understanding.
I'm well aware of the problems that result when upscaling AI images. I submit 100% AI upscaled to 7200 px along the longest side. I spend anywhere from 30 minutes to 2 hours editing and retouching each image, and sometime 3 or more if I have a troublesome asset that I'm particularly fond of. Point is, the impression I get from your posts is that, while you understand how upscaling can degrade an image, you aren't following through with the necessary editng process itself. It appears you're going for good enough (which is a mistake) or tossing possibly fixable assets in the trash. Or perhaps I'm wrong and you're at least doing your best to edit your assets before submitting. If so, then it would appear that you just need to improve upon your editing and retouching skills.
AdChoices