Cannot download model "'inswapper_128.onnx'"

[Southmelon-Allen]

swapper = insightface.model_zoo.get_model('inswapper_128.onnx', download=True, download_zip=True)

cannot download the model from given url. RuntimeError: Failed downloading url https://github.com/deepinsight/insightface/releases/download/v0.7/inswapper_128.onnx.zip

[Southmelon-Allen]

cannot download zip file manually through this link, Github respond "Not Found"

[zhouzhengguang]

+1

[nttstar]

The swapping demo had been moved to discord bot.

[zhouzhengguang]

@nttstar That's awesome! But, I want the inswapper_128.onnx model file to test my photo locally, can you share the model on drive or somewhere? Many thanks!

[nttstar]

The 128 model does not perform as well as the Discord version. At present, there is no reason to upload it again.

[mike-hearn]

there is no reason to upload it again

The ability to run it locally seems like a good reason.

[nothingrealhappen]

But seems the discord app is down at the moment. In the meantime inswapper_128 is deleted. Seem like no easy way to try the face swap feature by now.

BTW, is the discord version swap model will be public in the future?

Thanks anyway.

[nttstar]

But seems the discord app is down at the moment

No, it's still in service.

BTW, is the discord version swap model will be public in the future?

Not very likely

[YuanBo66668888]

还有其他办法么？ @nttstar

[nttstar]

you can test your image by using the discord service.

[Tylersuard]

Ok but the discord service only allows 50 commands per day. When the face swapping only works well 1 in 50 times, that's not a lot. I want to run this locally so I can face swap more and get some quality results. How can I do that?

[mike-hearn]

Out of curiosity, is there a commercial goal by privatizing this model? I haven't used the Discord. Do you have to pay for it?

It feels like there must be some ulterior motive surrounding removing the 128x128 model, and refusing to release any larger models.

I'm not opposed to you making money off your work obviously, but I just want to be clear what the rationale is.

[dai-shuo]

The 128 model does not perform as well as the Discord version. At present, there is no reason to upload it again.

it's reasonable for you guys to close source. but it'd be nicer to remove the inswapper from the opensourced python package since it's not working anymore.

[CGMikeG]

awwww thats painful would be nice if we can get if we can use this locally

[Mayorc1978]

What is the discord bot we are talking about?

[saurabhaloneai]

Link : https://drive.google.com/file/d/1krOLgjW2tAPaqV-Bw4YALz0xT5zlb5HF/view { thanks to that reddit guy }

[Fibonacci134]

Out of curiosity, is there a commercial goal by privatizing this model? I haven't used the Discord. Do you have to pay for it?

It feels like there must be some ulterior motive surrounding removing the 128x128 model, and refusing to release any larger models.

I'm not opposed to you making money off your work obviously, but I just want to be clear what the rationale is.

They basically got greedy, released the model and then pulled it. Expected better from them. Too bad the code and model are out already so they cant really do anything about it now.

[SolessChong]

uld be nice if we can

People have their rights to pull the model. I appreciate what they left opensourced.

Link : https://drive.google.com/file/d/1krOLgjW2tAPaqV-Bw4YALz0xT5zlb5HF/view { thanks to that reddit guy }

And I appreciate this even more.

[wiesbaden6200]

How can I run this locally? or how can I forward it to a different address in colab.

"inswapper_128.onnx"

Thanks

[grahamquest]

uld be nice if we can

People have their rights to pull the model. I appreciate what they left opensourced.

Link : https://drive.google.com/file/d/1krOLgjW2tAPaqV-Bw4YALz0xT5zlb5HF/view { thanks to that reddit guy }

And I appreciate this even more.

Agreed, and VERY MUCH agreed for that 2nd part. Some people are different, some see chaos and misuse in their projects and its fine for them to do whatever they want with it. As for me? I learned to BACK UP files a lot more.

[shandowsoftware]

并且非常同意第二部分。有些人是不同的，有些人在他们的项目中看到混乱和滥用，他们可以用它做任何他们想做的事情。至于我呢？我学会了更多备份文件。

uld be nice if we can

People have their rights to pull the model. I appreciate what they left opensourced.

Link : https://drive.google.com/file/d/1krOLgjW2tAPaqV-Bw4YALz0xT5zlb5HF/view { thanks to that reddit guy }

And I appreciate this even more.

Agreed, and VERY MUCH agreed for that 2nd part. Some people are different, some see chaos and misuse in their projects and its fine for them to do whatever they want with it. As for me? I learned to BACK UP files a lot more.

Link : https://drive.google.com/file/d/1krOLgjW2tAPaqV-Bw4YALz0xT5zlb5HF/view { thanks to that reddit guy }

this is the real open source,some people are not open enough

[stasha17]

Link : https://drive.google.com/file/d/1krOLgjW2tAPaqV-Bw4YALz0xT5zlb5HF/view { thanks to that reddit guy }

Is there any reason why there are different inswapper_128.onnx files? This one at the google drive is 541mb and the other one I have is 182mb. Is there a difference in the files, and if so, what?

[Maxencehabar]

How can I run this locally? or how can I forward it to a different address in colab.

"inswapper_128.onnx"

Thanks

Did you get the answer ? I'm trying to do the same thing and I'm having hard times trying. Thanks!

[Mayorc1978]

Link : https://drive.google.com/file/d/1krOLgjW2tAPaqV-Bw4YALz0xT5zlb5HF/view { thanks to that reddit guy }

Is there any reason why there are different inswapper_128.onnx files? This one at the google drive is 541mb and the other one I have is 182mb. Is there a difference in the files, and if so, what?

I would like to know as well, I have the 5XX MB version too, where did you get the 182mb version? Can you share it? By the way, what is the official version size? If the 182mb version is like a pruned version and works precisely the same it could be better for memory consumption.

[Ubermeisters]

So there's also this, just FYI everyone.

https://github.com/facefusion/facefusion-assets/releases/download/models/inswapper_128.onnx

and

https://huggingface.co/ezioruan/inswapper_128.onnx/tree/main

[qym77]

Link : https://drive.google.com/file/d/1krOLgjW2tAPaqV-Bw4YALz0xT5zlb5HF/view { thanks to that reddit guy }

after download how to use it? i am watching Rob Mulla'video, but it can't work as the video,i think i put it into the right folder,but it from the code seems not find it? plz!

[qym77]

How can I run this locally? or how can I forward it to a different address in colab. "inswapper_128.onnx" Thanks

Did you get the answer ? I'm trying to do the same thing and I'm having hard times trying. Thanks!

AssertionError Traceback (most recent call last) Cell In[36], line 1 ----> 1 swapper = insightface.model_zoo.get_model('inswapper_128.onnx', 2 download=False, 3 download_zip=False)

File F:\anaconda3\envs\torch\lib\site-packages\insightface\model_zoo\model_zoo.py:91, in get_model(name, **kwargs) 89 if not osp.exists(model_file) and allow_download: 90 model_file = download_onnx('models', model_file, root=root, download_zip=download_zip) ---> 91 assert osp.exists(model_file), 'model_file %s should exist'%model_file 92 assert osp.isfile(model_file), 'model_file %s should be a file'%model_file 93 router = ModelRouter(model_file)

AssertionError: model_file inswapper_128.onnx should exist

i am using jupyter notebook,after i download and put it into C:\Users\usrname.insightface\models\inswapper_128,it still can't work,can anybody helps!plz!