This post, we’ll go over how different AI models are and how the prompts put in are generated differently. using Stablecog.
Kandinsky 2.2 https://b.stablecog.com/e2aed209-20ca-44ed-b264-b74f45cbf01e.jpeg
VS.
SSD-1B https://b.stablecog.com/14f4e0e9-d9a0-4a47-8d0d-3b06e15875cf.jpeg
As you can see here, for both prompts I put in doctor and the left side shows how Kandinsky 2.2 portrays a doctor and the fingers look messed up. The right side shows how SSD-1B portrays a doctor and one eye looks messed up along with some parts of his outfit.
Next up we have FLIX.1 and Stable Diffusion 3.
FLIX.1 https://b.stablecog.com/a21350c6-2597-4af2-a3db-46a0bb306bc1.jpeg
VS.
Stable Diffusion 3 https://b.stablecog.com/e32d2404-5da1-4adb-a6b4-c6714de2764a.jpeg
In this prompt, I chose for both AI engines to make Kratos from God of War. They both have their own depictions on how Kratos looks and both are incorrect. In both images, the red line across the face is not accurate at all. However, they were both able to grasp his muscular build.
This last test I’ll be choosing which AI generators have the best potential to show the next prompt. Those two will be FLIX.1 and SSD-1B.
FLIX.1 https://b.stablecog.com/0d5cbc5e-8796-4781-89bd-e5d7f7a49396.jpeg
VS.
SSD-1B https://b.stablecog.com/74964dde-1aa8-46d8-bcf5-00e415c2b8b8.jpeg
I gave both AI generations the prompt of Thanos from the avengers and it seems FLIX.1 decided to give him an extra arm for no apparent reason. But if I’m being honest SSD-1B looks far better therefor declaring it the better AI generation models out of the 4 introduced here.