This involved about 20 different prompts.

This is three shots made with Kling 01 and Kling 2.6. Reference poses are generated with Realistic Pony, converted to depth input for Pony Realism 2.2, then cleaned up and made uniform with Google Nano Banana Pro. The character is then converted to an asset using Kling Omni 1, which allows character consistency across shots. Google Veo 3.1 is then used to generate reference motion, then that motion is mapped onto the character asset in Kling 2.6 Motion Control and Kling Omni 1.

All said and done this gen ended up costing like $20. Was it worth it? No. But I have a consistent character asset to hopefully generate longer videos with in the future.