We are thinking about the structure of a new puzzle. The general movement is set, but to make the puzzle work, we need footage of a bird flying and moving from branch to branch.
Initially, we considered using a 3D model, but we couldn’t create a model that could perform a series of actions such as taking off, turning, and perching on a branch.
The next thing we considered was preparing a series of images. However, we couldn’t find any free stock photos of the same type and in the desired pose.
So we tried out some image generation AI. First, we tried using online tools that are easy to use, such as Bing Image Creator, ChatGPT, and Gemini (NanoBanana).
The prompts were: “High-resolution photos, shot by a professional wildlife photographer, biologically and anatomically accurate depiction, avian behaviorally correct posture, small wild birds, green Japanese white-eye, attempting to perch on a branch, stretching out legs to grasp a branch, flapping wings with wings curled as if grasping the air, high-speed photography, blurred background.” Below are examples of the results.



We’ll try to see if we can generate it the same way in other poses.