Tencent/HunyuanDiT

TRT模型分辨率16:9问题

Opened this issue · 0 comments

https://github.com/Tencent/HunyuanDiT/blob/main/hydit/inference.py#L56
这里上面写的是16:9,但是下面的实际分辨率是5:3的,最终画出来是5:3的图

STANDARD_RATIO = np.array([
    1.0,        # 1:1
    4.0 / 3.0,  # 4:3
    3.0 / 4.0,  # 3:4
    16.0 / 9.0, # 16:9
    9.0 / 16.0, # 9:16
])
STANDARD_SHAPE = [
    [(1024, 1024), (1280, 1280)],   # 1:1
    [(1280, 960)],                # 4:3
    [(960, 1280)],                   # 3:4
    [(1280, 768)],                              # 16:9 这个实际是5:3的比例 如果要16:9应该是 1280*720
    [(768, 1280)],                              # 9:16
]

是哪个地方写错了?