For my first Wan 2.1 lora, I have chosen to train a version of my 'Western comic semirrealistic 2.5D style' lora that I trained for Flux and Pony some time ago. This style is a tribute to the adult fantasy and science fiction comic styles that we could find in the 80s and 90s in magazines such as the prestigious Heavy Metal. I did the training with Civitai's online training service and I'm very happy with the results.

I recommend using it with the LightX2V accelerator lora version 2, 6-8 steps, shift 7-8, CFG 1, with the sampler Euler and scheduler Simple. Interestingly, I have found that using lightx2v in its i2v version gives better results than in its t2v version, obtaining better movement and details in the textures. All the examples I show in the gallery are made with the mentioned configuration.