使用万能热巴模型

gg31415 · 发表于 2021-8-8 20:38:29

星级打分

1
2
3
4
5

平均分:NAN 参与人数:0 我的评分:未评

data_dst切好脸，模型复制大model里。直接 7) merge SAEHD.bat
就成下面这种情况了。
具体流程是什么

deepfacelab版本是不是有问题
我用的是滚石给的
【汉化2021-0602版】
https://dfldata.xyz/forum.php?mod=viewthread&tid=2073
【汉化2021-0104版（修复了几个小问题）】
重新分开为30系显卡与非30系显卡版本
增加改进版GAN，可以训练出更多细节

正在启动合成程序.

                           _ _ _ _ _
                           / \ / \ / \ / \ / \
                        ( 猫 | の | 汉 | 化）
                           \_/ \_/ \_/ \_/ \_/

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] : DF-UD256 - latest
:
0
加载名为 DF-UD256_SAEHD 的模型...

可用设备列表：

[CPU] : CPU
  [0] : GeForce GTX 1660 Ti

[0] 选择哪一个设备？ :
0

初始化模型: 100%|########################################################################| 4/4 [00:04<00:00,  1.09s/it]
================ Model Summary =================
==                                           ==
==          Model name: DF-UD256_SAEHD    ==
==                                           ==
==    Current iteration: 1106461          ==
==                                           ==
==-------------- Model Options ---------------==
==                                           ==
==          resolution: 256                ==
==          face_type: f                ==
==    models_opt_on_gpu: True             ==
==                archi: df-ud             ==
==             ae_dims: 352                ==
==             e_dims: 88                ==
==             d_dims: 88                ==
==          d_mask_dims: 28                ==
==    masked_training: True             ==
==          uniform_yaw: False             ==
==          lr_dropout: n                ==
==          random_warp: True             ==
==          gan_power: 0.0                ==
==    true_face_power: 0.01             ==
==    face_style_power: 0.01             ==
==       bg_style_power: 0.0                ==
==             ct_mode: rct                ==
==             clipgrad: True             ==
==             pretrain: False             ==
==    autobackup_hour: 12                ==
== write_preview_history: True             ==
==          target_iter: 0                ==
==          random_flip: False             ==
==          batch_size: 4                ==
==    eyes_mouth_prio: False             ==
==          adabelief: True             ==
==                猫の汉化: http://t.hk.uy/4ks  ==
==                出售仙丹: QQ564646676       ==
==       gan_patch_size: 32                ==
==             gan_dims: 16                ==
==                                           ==
==---------------- Running On ----------------==
==                                           ==
==       Device index: 0                ==
==                Name: GeForce GTX 1660 Ti ==
==                VRAM: 4.66GB             ==
==                                           ==
================================================
[y] 是否启用交互式合成? ( y/n ) :
y
[12] CPU线程数量? ( 1-12 ?:help ) :
12
收集对齐头像: 100%|###############################################################| 1619/1619 [00:01<00:00, 865.81it/s]
计算运动矢量: 100%|###############################################################| 1619/1619 [00:01<00:00, 975.67it/s]
[y] 使用保存的会话? ( y/n ) :
y
运行在 CPU0.
运行在 CPU1.
运行在 CPU2.
运行在 CPU3.
运行在 CPU4.
运行在 CPU5.
运行在 CPU6.
运行在 CPU7.
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "multiprocessing\spawn.py", line 105, in spawn_main
  File "multiprocessing\spawn.py", line 115, in _main
  File "F:\DeepFaceLab_NVIDIA_up_to_RTX2080Ti_0602汉化版-猫之汉化\_internal\DeepFaceLab\merger\__init__.py", line 2, in <module>
from .MergerConfig import MergerConfig, MergerConfigMasked, MergerConfigFaceAvatar
  File "F:\DeepFaceLab_NVIDIA_up_to_RTX2080Ti_0602汉化版-猫之汉化\_internal\DeepFaceLab\merger\MergerConfig.py", line 1, in <module>
import numpy as np
  File "F:\DeepFaceLab_NVIDIA_up_to_RTX2080Ti_0602汉化版-猫之汉化\_internal\python-3.6.8\lib\site-packages\numpy\__init__.py", line 138, in <module>
from . import _distributor_init
  File "F:\DeepFaceLab_NVIDIA_up_to_RTX2080Ti_0602汉化版-猫之汉化\_internal\python-3.6.8\lib\site-packages\numpy\_distributor_init.py", line 26, in <module>
WinDLL(os.path.abspath(filename))
  File "ctypes\__init__.py", line 348, in __init__
OSError: [WinError 1455] 页面文件太小，无法完成操作。