TypeError: exceptions must derive from BaseException (help me with this)

[Lutom23]

Traceback (most recent call last): File "E:\SadTalker-main\venv\lib\site-packages\gradio\routes.py", line 401, in run_predict output = await app.get_blocks().process_api( File "E:\SadTalker-main\venv\lib\site-packages\gradio\blocks.py", line 1302, in process_api result = await self.call_function( File "E:\SadTalker-main\venv\lib\site-packages\gradio\blocks.py", line 1025, in call_function prediction = await anyio.to_thread.run_sync( File "E:\SadTalker-main\venv\lib\site-packages\anyio\to_thread.py", line 31, in run_sync return await get_asynclib().run_sync_in_worker_thread( File "E:\SadTalker-main\venv\lib\site-packages\anyio_backends_asyncio.py", line 937, in run_sync_in_worker_thread return await future File "E:\SadTalker-main\venv\lib\site-packages\anyio_backends_asyncio.py", line 867, in run result = context.run(func, *args) File "E:\SadTalker-main\src\gradio_demo.py", line 111, in test first_coeff_path, crop_pic_path, crop_info = self.preprocess_model.generate(pic_path, first_frame_dir, preprocess) File "E:\SadTalker-main\src\utils\preprocess.py", line 101, in generate x_full_frames, crop, quad = self.croper.crop(x_full_frames, still=True, xsize=512) File "E:\SadTalker-main\src\utils\croper.py", line 170, in crop raise 'can not detect the landmark from source image' TypeError: exceptions must derive from BaseException

[magejosh]

Same problem on my end. Weirdest part is it works sometimes, but others it throws this error right after the point where it renders the temp png, like step 4 based on console output. I'm on a lower end system so I'm of a suspicion it's related to a timeout of some sort but unsure how to test that or fix it.

[vinthony]

are you trying a different image? it seems the image does not contain a real face image.

[keetruda69]

is there any workaround how to apply model into animated character (not real human face)?

[ericmday]

anyone find an answer to this one?

[KiteA2022]

anyone find an answer to this one?

[BeyondYourself]

Are U OK?

[jelly-liu]

change another image, Realistic version, not too much cartoon.

[keetruda69]

I found out that image have to have enough facial landmarks in order to get pass through the face detection and trigger swap.

adding facial landmarks is more art or trial-and-error e.g. via SD1.5 inpainting.