Comments (9)
Can you show your cases? let me know what's the problem
from hallo.
Original image: 3466 ✖ 1942 . Output video: 512 ✖512
from hallo.
Can you show your cases? let me know what's the problem
The parameters are default, and the final output video resolution is 512 × 512.
from hallo.
You can modify data.source_image.width and data.source_image.height in the inference config to generate higher resolution videos. However, please be mindful of your VRAM usage.
from hallo.
Original image: 3466 ✖ 1942 . Output video: 512 ✖512
BTW, please use square images.
from hallo.
You can modify data.source_image.width and data.source_image.height in the inference config to generate higher resolution videos. However, please be mindful of your VRAM usage.
pipeline_output = pipeline(
ref_image=pixel_values_ref_img,
audio_tensor=audio_tensor,
face_emb=source_image_face_emb,
face_mask=source_image_face_region,
pixel_values_full_mask=source_image_full_mask,
pixel_values_face_mask=source_image_face_mask,
pixel_values_lip_mask=source_image_lip_mask,
width=1024,
height=1024,
video_length=clip_length,
num_inference_steps=config.inference_steps,
guidance_scale=config.cfg_scale,
generator=generator,
motion_scale=motion_scale,
)
change:
width=1024
height=1024
Traceback (most recent call last):
File "F:\workplace\hallo-webui\scripts\inference.py", line 424, in
inference_process(
File "F:\workplace\hallo-webui\scripts\inference.py", line 364, in inference_process
pipeline_output = pipeline(
File "F:\workplace\hallo-webui\venv\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "F:\workplace\hallo-webui\hallo\animate\face_animate.py", line 401, in call
noise_pred = self.denoising_unet(
File "F:\workplace\hallo-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\workplace\hallo-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "F:\workplace\hallo-webui\hallo\models\unet_3d.py", line 605, in forward
sample = sample + mask_cond_fea
RuntimeError: The size of tensor a (128) must match the size of tensor b (64) at non-singleton dimension 4
from hallo.
You can modify data.source_image.width and data.source_image.height in the inference config to generate higher resolution videos. However, please be mindful of your VRAM usage.
pipeline_output = pipeline( ref_image=pixel_values_ref_img, audio_tensor=audio_tensor, face_emb=source_image_face_emb, face_mask=source_image_face_region, pixel_values_full_mask=source_image_full_mask, pixel_values_face_mask=source_image_face_mask, pixel_values_lip_mask=source_image_lip_mask, width=1024, height=1024, video_length=clip_length, num_inference_steps=config.inference_steps, guidance_scale=config.cfg_scale, generator=generator, motion_scale=motion_scale, )
change: width=1024 height=1024
Traceback (most recent call last): File "F:\workplace\hallo-webui\scripts\inference.py", line 424, in inference_process( File "F:\workplace\hallo-webui\scripts\inference.py", line 364, in inference_process pipeline_output = pipeline( File "F:\workplace\hallo-webui\venv\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(*args, **kwargs) File "F:\workplace\hallo-webui\hallo\animate\face_animate.py", line 401, in call noise_pred = self.denoising_unet( File "F:\workplace\hallo-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1511, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "F:\workplace\hallo-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1520, in _call_impl return forward_call(*args, **kwargs) File "F:\workplace\hallo-webui\hallo\models\unet_3d.py", line 605, in forward sample = sample + mask_cond_fea RuntimeError: The size of tensor a (128) must match the size of tensor b (64) at non-singleton dimension 4
Do not modify the code. Just modify the data.source_image.width
and data.source_image.height
in configs/inference/default.yaml
.
from hallo.
您可以在推理配置中修改 data.source_image.width 和 data.source_image.height 以生成更高分辨率的视频。但是,请注意 VRAM 的使用情况。
pipeline_output = pipeline(ref_image=pixel_values_ref_img, audio_tensor=audio_tensor, face_emb=source_image_face_emb, face_mask=source_image_face_region, pixel_values_full_mask=source_image_full_mask, pixel_values_face_mask=source_image_face_mask, pixel_values_lip_mask=source_image_lip_mask, width=1024, height=1024, video_length=clip_length, num_inference_steps=config.inference_steps, guide_scale=config.cfg_scale, generator=generator, motion_scale=motion_scale, )
更改:宽度=1024 高度=1024
回溯(最近一次调用最后一次):文件“F:\workplace\hallo-webui\scripts\inference.py”,第 424 行,在 inference_process 中(文件“F:\workplace\hallo-webui\scripts\inference.py”,第 364 行,在 inference_process 中pipeline_output = pipeline(文件“F:\workplace\hallo-webui\venv\lib\site-packages\torch\utils_contextlib.py”,第 115 行,在 decorate_context return func(args,kwargs)文件“F:\workplace\hallo-webui\hallo\animate\face_animate.py”,第 401 行,在call* noise_pred = self.denoising_unet(文件“F:\workplace\hallo-webui\venv\lib\site-packages\torch\nn\modules\module.py”,第 1511 行,在 _wrapped_call_impl return self._call_impl(*args,**kwargs) 文件“F:\workplace\hallo-webui\venv\lib\site-packages\torch\nn\modules\module.py”,第 1520 行,在 _call_impl 中返回 forward_call(*args,**kwargs) 文件“F:\workplace\hallo-webui\hallo\models\unet_3d.py”,第 605 行,在正向样本 = 样本 + mask_cond_fea RuntimeError:张量 a (128) 的大小必须与非单例维度 4 处的张量 b (64) 的大小匹配不要修改代码。只需修改中的
data.source_image.width
和。data.source_image.height``configs/inference/default.yaml
I'll try
from hallo.
@Song367 Did it work?
Were you able to generate 1024 x 1024 frames?
Also, how was the quality? Is it just 512 x 512 resized or what?
from hallo.
Related Issues (20)
- Read the checkpoint of the last training and continue training. The value predicted by the model is nan HOT 5
- Question about hallo
- pip install -e ,
- Can it be used for task of Lip sync? HOT 1
- How to change the parameters in the py file to train gpu? HOT 3
- train stage1, not use audio feature, only learn the image generation? HOT 1
- Please upload a Google Colab Link
- The original image had a green background, why did the background change after it was generated? HOT 2
- how to keep the generated video the same size as the original image HOT 1
- some bugs when gradient_checkpointing is set True
- Some errors in unconditional audio forward
- Question about reduction=mean when computing loss in train stage 2 ?
- 显存恒定9G左右 HOT 1
- why this hell error?
- How can this be run on CPU only?
- Artifacts and Lip Sync Issues in Talking Face Model Training - Seeking Advice from Authors 谈话脸模型训练中的伪影和嘴形同步问题 - 寻求作者建议
- Artifacts and Lip Sync Issues in Model Training - Seeking Advice from Authors \n 模型训练中的伪影和嘴形同步问题 - 寻求作者建议
- Failed to execute the training process: HOT 1
- Which reference picture yields best result? HOT 1
- insightface can not pip success
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from hallo.