Evolution/Comparison of my Necromancer
By Sealine
This is another journey of an image culminating with a comparison of the three levels of GPT and a video generation. This image, like many of my female creations, evolved from my little bobble headed YoWorld avatar who in this case was dressed as a Necromancer and flanked by her undead protectors. This image also got its start as an AI image on my first platform, PicLumen. At the time, I liked it but not enough to keep running the prompt in different models. With my move to BudgetPixel, I started pulling old prompts and running them here. Flux 1.1 pro was giving me the best results but still not the feel I wanted for the image. Flux is just too pretty for this subject. This was shelved again. Then today on X a challenge for a Dark Art image popped up in the Art Odyssey community I'm part of and I immediately thought of my Necromancer. I knew I had to try this prompt in GPT 1.5, my favorite. I decided to run this image in all three options: low, medium and high. The results are in order below from left to right. Note: I did make the choice to add the low angle view to my prompt for a more dynamic composition. Low does a nice job and I could have been perfectly happy with this image. Medium adds some detail, but I don't like the changes to the main subject. The brighter green gives it an eye-catching pop, but it's not the feel I wanted for this image. High delivered everything I was looking for from the beginning. It has the right feel, a fantastic subject and the intricate details in the outfit. Next up was animation. My go to lately is Grok on BudgetPixel. It delivers great animation with little prompt guidance. Note: If you do not specify no talking, it will often give your characters lines, for better or worse. Thank you for taking the time to read this blog. Please let me know what you think in the comments!