this is pretty cool. i've generated two characters in t-pose on greenscreen, then a panoramic environment which i can then crop as i wish to ensure the shot i need are made within the part of the environment i want. then i am ingesting those pictures as references in nanobanana pro with the prompt describing the scene. and then i use qwen edit 2509 with the multi-angle camera lora to change the camera as i wish while keeping entire consistency this is awesome. i'll do some tests using wan2.2 FLFV & Wan Animate (or Vace?) to generate some animations out of it.