AI Test Generative Models Showcase
Bilel Aroua - Bils
End Note (Test)
Testing Summary: I ran 20+ examples across the latest models—Kling, PixVerse, and WAN 2.5. For anime, WAN 2.5 consistently underdelivered, while Veo produced the strongest results, especially with complex prompts.
Workflow: For the project, the workflow will be to generate multiple outputs per prompt for each scene, then curate and select the top result for final delivery.
Important: Videos are not upscaled - All outputs are shown in their native resolution.
AI Model