Reproducing 8-bit Game Console Screen with GPT-Image2 and Pixverse

Recently, image generation AI has been evolving rapidly, making many previously difficult expressions possible. At the same time, different models have their strengths and weaknesses, and knowing this can affect the handling of output images. So, this time I used Pixverse's image generation function to reproduce the "screen of an 8-bit game console." Based on the original image, I used the same prompt to perform image-to-image with both NanobananaPro and GPT-Image2. What were the results? NanobananaPro had a pixel coarseness close to the Super Famicom, while GPT-Image2 had a pixel coarseness close to the Famicom. I think GPT-Image2 is closer to an 8-bit game screen. Since that alone is not interesting, I used PixverseV6 to turn the GPT-Image2 image into a video. The original image has little information and is somewhat problematic, but the motion itself is not bad. By the way, PixverseV6 is an easy-to-use model that can output from 360p quality and also supports audio, so it is recommended when you want to try some motion without needing large actions like Seedacne2.0.