Copy link to clipboard
Copied
I applied for an image created with adobe fire fly and it was rejected.
It was a quality issue.
I would like to know what is wrong.
The images I created with firefly produced 4 images after inputting the prompts. I enlarged those images, downloaded them, and submitted them.
Photoshop. By hand. There is no magic way to fix errors. I while I consider myself an above average retoucher/image editor, I can not image being able to fix these assets. You chose very complex subjects for any AI app to produce acceptable results "right out of the box."
Copy link to clipboard
Copied
Nothing in any of these images is rendered with the detail required for stock. Here are three examples. I haven't necessarily highlight all the errors. And keep in mind that it only takes one issue to result in a refusal.
Copy link to clipboard
Copied
Thank you for your advice!
Can details in photos like this be fixed with Adobe Firefly? Is it something to fix with Photoshop or something?
Copy link to clipboard
Copied
Photoshop. By hand. There is no magic way to fix errors. I while I consider myself an above average retoucher/image editor, I can not image being able to fix these assets. You chose very complex subjects for any AI app to produce acceptable results "right out of the box."
Copy link to clipboard
Copied
thank you. I thought more about the prompts, looked more closely at the images I was able to generate, and wanted to improve the quality of the images.
Copy link to clipboard
Copied
I wouldn't consider any of these to be "fixable". There are simply too many drawing errors including lack of details, fuzzy lines, etc. It would be better to regenerate and hope for better results next time.
Copy link to clipboard
Copied
thank you.
When I took a quick look at the generated image, I was impressed, but it was difficult to express the details. I would like to try it again and again.
Copy link to clipboard
Copied
I tried to make an image of a communications satellite in a number of AI applications, and the results were the same: none could produce the detail needed in the intricate pattern of the "wings" (or whatever they're called). I'd consider a different subject.
Copy link to clipboard
Copied
It is difficult to express detailed information with generative AI.
Solar panel images, similar to communications satellite images, were also difficult to generate detail.
I will also look for another theme.
Copy link to clipboard
Copied
AI doesn't particular like too many details. I always get a bit of a laugh with 100-200 word prompts, where more than half of the details are ignored. And yes...fine details in an AI images are particularly difficult to render properly. Even things like shirt buttons and shiny jewelry. Such details need to be edited or deleted.
Copy link to clipboard
Copied
Final checks and finishing touches need to be done by humans. I didn't understand the last check, so everyone in this thread taught me how to check images, which helped me a lot.
Copy link to clipboard
Copied
Try different niche ask chatgpt
Copy link to clipboard
Copied
thank you.
I got the idea for generated images in chat GTP, but I felt like I needed to experiment with it more.