search

Interpause / auto-sd-paint-ext

Extension for AUTOMATIC1111 to add custom backend API for Krita Plugin & more
MIT License
477 stars 41 forks source link

This plugin is not working in https://github.com/vladmandic/automatic #144

Open spreck opened 1 year ago

spreck commented 1 year ago

Describe the bug The extension worked in automatic 1111, but that project hasn't seen any updates in a month and so it was recommended to switch to vladmandic's fork which is not abandoned.

However that fork has --api always on and it is not included as part of the webui.bat file or when calling the command as a command line argument. There have been issues with using openOutpaint due to this, https://github.com/vladmandic/automatic/issues/282, and I suspect something similar is going on with this.

This is a great extension and it would be awesome to be able to use it without needing to have to use a stale project to run it. I'm sure that vlad would like to see it work as well.

To Reproduce Steps to reproduce the behavior:

  1. Go to 'https://github.com/vladmandic/automatic'
  2. Install automatic from that fork
  3. Add, https://github.com/Interpause/auto-sd-paint-ext/ to extensions, extension installs, restart server and UI
  4. Tab is not displayed in automatic web ui. Krita plugin shows "Network error: Not Found", and then when I try to run a module it says "Cannot reach backend!". However

Screenshots This is the error in Krita, I started automatic and then waited for it to image

image

Then when I quit Krita I see this Script error window with a 'Cannot reach backed!' message that I have to close about 5 or 6 times: image

RuntimeError
Python 3.8.1: C:\Program Files\Krita (x64)\bin\krita.exe
Fri Apr 28 10:24:34 2023

A problem occurred in a Python script.  Here is the sequence of
function calls leading up to the error, in the order they occurred.

 C:\Users\guy\AppData\Roaming\krita\pykrita\krita_diff\pages\upscale.py in <lambda>(s='Cannot reach backend!')
   56     def cfg_connect(self):
   57         self.upscaler_layout.cfg_connect()
   58         self.downscale_first.cfg_connect()
   59         self.btn.released.connect(lambda: script.action_simple_upscale())
   60         script.status_changed.connect(lambda s: self.status_bar.set_status(s))
global script = <krita_diff.script.Script object>
script.status_changed = <bound PYQT_SIGNAL status_changed of Script object>
script.status_changed.connect = <built-in method connect of PyQt5.QtCore.pyqtBoundSignal object>
s = 'Cannot reach backend!'
self = <krita_diff.pages.upscale.UpscalePage object>
self.status_bar = <krita_diff.widgets.status_bar.StatusBar object>
self.status_bar.set_status = <bound method StatusBar.set_status of <krita_diff.widgets.status_bar.StatusBar object>>

 C:\Users\guy\AppData\Roaming\krita\pykrita\krita_diff\widgets\status_bar.py in set_status(self=<krita_diff.widgets.status_bar.StatusBar object>, s='Cannot reach backend!')
   18             return
   19         if s == STATE_DONE and STATE_INTERRUPT == self.text():
   20             return
   21 
   22         self.setText(f"<b>Status:</b> {s}")
self = <krita_diff.widgets.status_bar.StatusBar object>
self.setText = <built-in method setText of StatusBar object>
RuntimeError: wrapped C/C++ object of type StatusBar has been deleted
    __cause__ = None
    __class__ = <class 'RuntimeError'>
    __context__ = None
    __delattr__ = <method-wrapper '__delattr__' of RuntimeError object>
    __dict__ = {}
    __dir__ = <built-in method __dir__ of RuntimeError object>
    __doc__ = 'Unspecified run-time error.'
    __eq__ = <method-wrapper '__eq__' of RuntimeError object>
    __format__ = <built-in method __format__ of RuntimeError object>
    __ge__ = <method-wrapper '__ge__' of RuntimeError object>
    __getattribute__ = <method-wrapper '__getattribute__' of RuntimeError object>
    __gt__ = <method-wrapper '__gt__' of RuntimeError object>
    __hash__ = <method-wrapper '__hash__' of RuntimeError object>
    __init__ = <method-wrapper '__init__' of RuntimeError object>
    __init_subclass__ = <built-in method __init_subclass__ of type object>
    __le__ = <method-wrapper '__le__' of RuntimeError object>
    __lt__ = <method-wrapper '__lt__' of RuntimeError object>
    __ne__ = <method-wrapper '__ne__' of RuntimeError object>
    __new__ = <built-in method __new__ of type object>
    __reduce__ = <built-in method __reduce__ of RuntimeError object>
    __reduce_ex__ = <built-in method __reduce_ex__ of RuntimeError object>
    __repr__ = <method-wrapper '__repr__' of RuntimeError object>
    __setattr__ = <method-wrapper '__setattr__' of RuntimeError object>
    __setstate__ = <built-in method __setstate__ of RuntimeError object>
    __sizeof__ = <built-in method __sizeof__ of RuntimeError object>
    __str__ = <method-wrapper '__str__' of RuntimeError object>
    __subclasshook__ = <built-in method __subclasshook__ of type object>
    __suppress_context__ = False
    __traceback__ = <traceback object>
    args = ('wrapped C/C++ object of type StatusBar has been deleted',)
    with_traceback = <built-in method with_traceback of RuntimeError object>

The above is a description of an error in a Python program.  Here is
the original traceback:

Traceback (most recent call last):
  File "C:\Users\guy\AppData\Roaming\krita\pykrita\krita_diff\pages\upscale.py", line 60, in <lambda>
    script.status_changed.connect(lambda s: self.status_bar.set_status(s))
  File "C:\Users\guy\AppData\Roaming\krita\pykrita\krita_diff\widgets\status_bar.py", line 22, in set_status
    self.setText(f"<b>Status:</b> {s}")
RuntimeError: wrapped C/C++ object of type StatusBar has been deleted

Desktop (please complete the following information):

Interpause commented 1 year ago

Above error is unrelated to the connection issue. Seems to be a result of the code trying to set the status bar text while Krita is closing.

Interpause commented 1 year ago

Can you send me a screenshot of the terminal running the webUI? It should give a warning like COMMANDLINE_ARGS does not contain --api, API won't be mounted.. https://github.com/vladmandic/automatic/issues/282 does mention he added some sort of compatibility, so it might be a different issue.

spreck commented 1 year ago

Okay, it's connecting now, but there is still an error: 13:50:44-112566 ERROR Exception: 'int' object is not iterable 13:50:44-130539 ERROR INFO: 127.0.0.1:52632 - "GET /sdapi/interpause/config HTTP/1.1" 200 OK Arguments: args=('', 0INFO: 127.0.0.1:52633 - "GET /sdapi/v1/progress HTTP/1.1" 200 OK , 'test'INFO: 127.0.0.1:52631 - "GET /sdapi/v1/progress HTTP/1.1" 200 OK , '', 'None', <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=666x222 at 0x2241F8A6DA0>, None, None, None, None, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=666x222 at 0x2241F8A6DA0>, None, 40, 19, 0, None, 1, False, False, 1, 1, 12.0, 0, 0.8, -1, -1, 0.0, 0, 0, False, 1, 224, 672, 1.0, 1, False, 0, False, '', '', '', [], 0) kwargs={} 13:50:44-178543 ERROR gradio call: TypeError ┌───────────────────────────────────────── Traceback (most recent call last) ─────────────────────────────────────────┐ │ C:\Users\fwss\github\automatic\modules\call_queue.py:34 in f │ │ │ │ 33 │ │ │ try: │ │ > 34 │ │ │ │ res = func(*args, **kwargs) INFO: 127.0.0.1:52634 - "GET /sdapi/v1/progress HTTP/1.1" 200 OK │ │ 35 │ │ │ │ progress.record_results(id_task, res) │ │ │ │ C:\Users\fwss\github\automatic\modules\img2img.py:81 in img2img │ │ │ │ 80 │ │ sampler_index = 0 │ │ > 81 │ override_settings = create_override_settings_dict(override_settings_texts) │ │ 82 │ │ │ │ C:\Users\fwss\github\automatic\modules\generation_parameters_copypaste.py:341 in create_override_settings_dict │ │ │ │ 340 │ params = {} │ │ > 341 │ for pair in text_pairs: │ │ 342 │ │ k, v = pair.split(":", maxsplit=1) │ └─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘ TypeError: 'int' object is not iterable

JaeHeyK commented 1 year ago

Okay, it's connecting now, but there is still an error: 13:50:44-112566 ERROR Exception: 'int' object is not iterable 13:50:44-130539 ERROR INFO: 127.0.0.1:52632 - "GET /sdapi/interpause/config HTTP/1.1" 200 OK Arguments: args=('', 0INFO: 127.0.0.1:52633 - "GET /sdapi/v1/progress HTTP/1.1" 200 OK , 'test'INFO: 127.0.0.1:52631 - "GET /sdapi/v1/progress HTTP/1.1" 200 OK , '', 'None', <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=666x222 at 0x2241F8A6DA0>, None, None, None, None, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=666x222 at 0x2241F8A6DA0>, None, 40, 19, 0, None, 1, False, False, 1, 1, 12.0, 0, 0.8, -1, -1, 0.0, 0, 0, False, 1, 224, 672, 1.0, 1, False, 0, False, '', '', '', [], 0) kwargs={} 13:50:44-178543 ERROR gradio call: TypeError ┌───────────────────────────────────────── Traceback (most recent call last) ─────────────────────────────────────────┐ │ C:\Users\fwss\github\automatic\modules\call_queue.py:34 in f │ │ │ │ 33 │ │ │ try: │ │ > 34 │ │ │ │ res = func(*args, **kwargs) INFO: 127.0.0.1:52634 - "GET /sdapi/v1/progress HTTP/1.1" 200 OK │ │ 35 │ │ │ │ progress.record_results(id_task, res) │ │ │ │ C:\Users\fwss\github\automatic\modules\img2img.py:81 in img2img │ │ │ │ 80 │ │ sampler_index = 0 │ │ > 81 │ override_settings = create_override_settings_dict(override_settings_texts) │ │ 82 │ │ │ │ C:\Users\fwss\github\automatic\modules\generation_parameters_copypaste.py:341 in create_override_settings_dict │ │ │ │ 340 │ params = {} │ │ > 341 │ for pair in text_pairs: │ │ 342 │ │ k, v = pair.split(":", maxsplit=1) │ └─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘ TypeError: 'int' object is not iterable

when I ran vlad's webui with debug option(.\webui --debug), I found somehow 'clip_skip' parameter is missing(-1 in log below, but should be 2 according to my setting). I think that's why all prameters after 'clip_skip' took wrong position and returned type error. 

04:40:14-267444 DEBUG    txt2img: id_task=|prompt=(masterpiece, best quality:1.2), 1girl raise her hands, correct
                         anatomy|negative_prompt=(worst quality, low quality:1.4), (verybadimagenegative_v1.3:0.8),
                         text|prompt_styles=None|steps=30|sampler_index=3|restore_faces=False|tiling=False|n_iter=1|batch_size=1|cfg_scal
                         e=7.0|clip_skip=-1|seed=-1|subseed=0.0|subseed_strength=0|seed_resize_from_h=0|seed_resize_from_w=False|seed_ena
                         ble_extras=640|height=512|width=True|enable_hr=0.6|denoising_strength=0|hr_scale=SwinIR_4x|hr_upscaler=0|hr_seco
                         nd_pass_steps=1024|hr_resize_x=1280|hr_resize_y=[]|override_settings_texts=0args=()

So as quick fix, I add req.clip_skip parameters in app.py file and at least txt2img seems to work now. Go to extensions/auto-sd-paint-ext/backend/app.py file, and add req.clip_skip between req.cfg_scale and req.seed.

 output = wrap_gradio_gpu_call(modules.txt2img.txt2img)(
        "",  # id_task (used by wrap_gradio_gpu_call for some sort of job id system)
        parse_prompt(req.prompt),  # prompt
        parse_prompt(req.negative_prompt),  # negative_prompt
        "None",  # prompt_styles: saved prompt styles (unsupported)
        req.steps,  # steps
        get_sampler_index(req.sampler_name),  # sampler_index
        req.restore_faces,  # restore_faces
        req.tiling,  # tiling
        req.batch_count,  # n_iter
        req.batch_size,  # batch_size
        req.cfg_scale,  # cfg_scale
        req.clip_skip, # clip_skip <<<<< here
        req.seed,  # seed
        req.subseed,  # subseed
        req.subseed_strength,  # subseed_strength
        req.seed_resize_from_h,  # seed_resize_from_h
        req.seed_resize_from_w,  # seed_resize_from_w
        req.seed_enable_extras,  # seed_enable_extras
        height,  # height
        width,  # width
        req.highres_fix,  # enable_hr: high res fix
        req.denoising_strength,  # denoising_strength: only applicable if high res fix in use
        0,  # hr_scale (overrided by hr_resize_x/y)
        req.upscaler_name,  # hr_upscaler: upscaler to use for highres fix
        0,  # hr_second_pass_steps: 0 uses same num of steps as generation to refine details
        req.orig_width,  # hr_resize_x
        req.orig_height,  # hr_resize_y
        [],  # override_settings_texts (unsupported)
        *args,
)

edit: okay, I just checked there's two 'output' sections in app.py (line 129 for txt2img and line 243 for img2img). By adding clip_skip param on both side, all of features(t2i, i2i, inpaint, upscale) seems to work. image

Daryl-Horton commented 1 year ago

JaeHeyK, thanks for the solution. This seems to work for me, but vladmandic responds with a number of TypeErrors for missing required positional arguments.

JaeHeyK commented 1 year ago

@Daryl-Horton yes, those errors appeared on my side too. And I suspect that's because vladmandic need default script parameters even for unused scripts. Maybe it can be covered in new issue.