hunyuan_V1This time I used more than 150 pictures as training set, and did not filter them. I was curious about what results hunyuanvideo would give me. Although the results disappointed me a little, they were okay.Flux_V1This is still a difficult face to train and need to reduce img for training.SD1.5I just try to cut out the crap I make and publish a better version.

Files with SHA256: b743ef20d873ce440e7956846683b89538cb9a7c8a225d2acc5e725feaf20af4