When it comes to generating high-quality images with Wan 2.2 t2v, the right settings can make all the difference. I’ve experimented with various configurations and want to share my findings with the community.
First and foremost, I’ve found that a cfg of 3.5 is crucial for achieving good results. Anything lower, and the prompt adherence suffers. I’ve also tried using lightx2v/causvid, but they seem to reduce the quality of the output.
On the other hand, using the clownsharksampler with res_X samplers has been a game-changer. The prompt adherence, variety, and detail are all significantly improved. I’ve also experimented with different schedulers, including bong_tangent and beta57, and found them to be effective.
In terms of total steps, I’ve found that 20-30 steps are ideal, with an even split between ksamplers. Using speed loras can also improve the quality of the output. Finally, I’ve found that sage attention and a shift value of 1-8 can also be beneficial.
One thing to keep in mind is that generating high-quality images with Wan 2.2 t2v can be expensive, especially when using cloud GPUs. However, the ability to generate uncensored content makes it a valuable option.
I’d love to hear from others in the community about their experiences with Wan 2.2 t2v and any tips they may have for achieving the best results.