From YoWorld to My World

My journey to AI image creation stemmed from a forum post for a niche Facebook game I played called YoWorld. Someone created a post using AI to turn their bobble headed avatar into a more realistic person. The early results were not good. But with practice and the help of new models and tools on Budgetpixel.com images have improved. This post will take you on my journey from avatar to image. First, the two avatars that inspired the image you see above. In this scenario, I want to keep as much detail as possible, so I utilized AI to help me write a prompt describing the characters in detail, then I refine it. The following two images were generated using Z-image-turbo. The next step is to put both characters in one image and add a background. For this step I used Nano Banana i2i with the two images above as references. The prompt was simple to get the two ladies together with a background: Use the reference photos to create an image with these two woman standing at a wine barrel table overlooking a scenic vineyard at sunset. They are at a wine tasting on an outdoor patio. Golden hour light. Graceful and elegant. The resulting image is below. It did what I asked but not a very dynamic or interesting image. The last step was to refine the prompt adding style information and details based on the above image and trying it in different models. I finally settled on the creation from GPT-image-1.5 that is in the header. Below is the whole image. Thank you for taking the time to join me in my AI journey for this image!