|
|
发表于 2021-12-3 12:12:52
|
显示全部楼层
本帖最后由 seahearing 于 2021-12-3 17:12 编辑
这个模型我下载后，把dst视频分解为图片，然后提取完DST头像，结果在合成时发现没有RCT的人脸。合成可以运行，但有报错如下： 启动合成程序.
选择一个模型, 或者输入一个名称去新建模型。
[r] : 重命名
[d] : 删除
[0] : new - latest
:
0
加载名为 new_SAEHD 的模型...
可用设备列表：
[CPU] : CPU
[0] : NVIDIA GeForce GTX 1070 Ti
[1] : Quadro P600
[0] 选择哪一个设备？ :
0
缓存GPU内核...
初始化模型: 100%|########################################################################| 4/4 [00:48<00:00, 12.01s/it]
==================== Model Summary ====================
== ==
== Model name: new_SAEHD ==
== ==
== Current iteration: 4714856 ==
== ==
==------------------ Model Options ------------------==
== ==
== resolution: 224 ==
== face_type: wf ==
== models_opt_on_gpu: True ==
== archi: liae-udt ==
== ae_dims: 512 ==
== e_dims: 64 ==
== d_dims: 64 ==
== d_mask_dims: 32 ==
== masked_training: True ==
== eyes_mouth_prio: True ==
== uniform_yaw: True ==
== blur_out_mask: True ==
== adabelief: True ==
== lr_dropout: y ==
== random_warp: False ==
== random_hsv_power: 0.1 ==
== true_face_power: 0.0 ==
== face_style_power: 0.0 ==
== bg_style_power: 0.0 ==
== ct_mode: none ==
== clipgrad: False ==
== pretrain: False ==
== autobackup_hour: 0 ==
== write_preview_history: False ==
== target_iter: 0 ==
== random_src_flip: False ==
== random_dst_flip: True ==
== batch_size: 8 ==
== gan_power: 0.1 ==
== gan_patch_size: 28 ==
== gan_dims: 32 ==
== ==
==------------------- Running On --------------------==
== ==
== Device index: 0 ==
== Name: NVIDIA GeForce GTX 1070 Ti ==
== VRAM: 8.00GB ==
== ==
=======================================================
[y] 是否启用交互式合成? ( y/n ) :
y
[16] 线程数量? ( 1-16 ?:help ) :
16
收集对齐头像: 100%|###############################################################| 9488/9488 [01:24<00:00, 112.38it/s]
计算运动矢量: 100%|############################################################| 12412/12412 [00:05<00:00, 2127.42it/s]
运行在 CPU2.
运行在 CPU1.
运行在 CPU3.
运行在 CPU0.
运行在 CPU5.
运行在 CPU4.
运行在 CPU8.
运行在 CPU7.
运行在 CPU9.
运行在 CPU6.
运行在 CPU11.
运行在 CPU10.
运行在 CPU12.
运行在 CPU13.
运行在 CPU15.
运行在 CPU14.
合成进度: 0%|
| 0/12412 [00:00<?, ?it/s]Y:\DeepFaceLab_NVIDIA20210104\_internal\DeepFaceLab\core\imagelib\color_transfer.py:196: RuntimeWarning: divide by zero encountered in float_scalars
l = (lStdSrc / lStdTar) * l
Y:\DeepFaceLab_NVIDIA20210104\_internal\DeepFaceLab\core\imagelib\color_transfer.py:196: RuntimeWarning: invalid value encountered in multiply
l = (lStdSrc / lStdTar) * l
Y:\DeepFaceLab_NVIDIA20210104\_internal\DeepFaceLab\core\imagelib\color_transfer.py:197: RuntimeWarning: divide by zero encountered in float_scalars
a = (aStdSrc / aStdTar) * a
Y:\DeepFaceLab_NVIDIA20210104\_internal\DeepFaceLab\core\imagelib\color_transfer.py:197: RuntimeWarning: invalid value encountered in multiply
a = (aStdSrc / aStdTar) * a
Y:\DeepFaceLab_NVIDIA20210104\_internal\DeepFaceLab\core\imagelib\color_transfer.py:198: RuntimeWarning: divide by zero encountered in float_scalars
b = (bStdSrc / bStdTar) * b
Y:\DeepFaceLab_NVIDIA20210104\_internal\DeepFaceLab\core\imagelib\color_transfer.py:198: RuntimeWarning: invalid value encountered in multiply
b = (bStdSrc / bStdTar) * b
合成图片 MergerConfig: 00001.png:
合成模式 Mode: overlay
遮罩模式 mask_mode: learned-prd*learned-dst
遮罩大小 erode_mask_modifier: 0
遮罩羽化 blur_mask_modifier: 0
动态模糊 motion_blur_power: 0
头像缩放 output_face_scale: 0
颜色模式 color_transfer_mode: rct
锐化模式 sharpen_mode: None
锐化力度 blursharpen_amount: 0
超分力度 super resolution power: 0
图片降噪力度 image_denoise_power: 0
双三次插值降低力度 bicubic_degrade_power: 0
颜色降低力度 color_degrade_power: 0
================
结果就是目标人物DST的人脸没有被替换，无论哪种设置我都尝试过了没效果。我的显卡配置是1070ti，DFL版本是DeepFaceLab_NVIDIA_up_to_RTX2080Ti，请问各位大神，你们有没有出现过这个现象？是版本问题还是什么其他问题导致的？
|
|
[复制链接]
|Archiver|手机版|deepfacelab中文网
|网站地图
