Is the v2 better or difference ? (and thx for this lora)

How did you get the training data for this? Is it from other AI video generators? Or Wan itself?

Very good LORA. Coolly working

can you pls pls do handheld camera moving/walking POV lora please!!!!

A very simple yet absolutely essential LoRA. It has literally saved us a huge chunk of work in 3D. Massive respect to the author!

Previously, when we didn’t have access to the real location where the deepfake was supposed to take place, we had to build an entire realistic 3D scene from photos, just like in CGI. Now we train the neural net on specific objects, generate images, run them through a 3D neural network for high-poly and retopology, then assemble photorealistic scenes based on that. Once again, huge thanks to the author! 😊

v2 flat out doesn't work. Tested many prompts, models and different pictures and it never worked once. Took a chance and used v1 after seeing all the examples were listing that version and it worked right away.

Has anyone figured out a configuration that allows us to provide context of both front and back to the model (to avoid it hallucinating the flipside)?