Error when sending to img2img

[4lt3r3go]

[SOLVED] this is happening when i press "send to img2img" or "send to inpaint" from the Image Browser tab. ..but same buttons pressed in txt2img tab works fine. :(

[Alexey1327]

Same problem, but I receive bit different error after update webUI:

File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\gradio\routes.py", line 321, in run_predict output = await app.blocks.process_api( File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\gradio\blocks.py", line 1015, in process_api result = await self.call_function(fn_index, inputs, iterator, request) File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\gradio\blocks.py", line 856, in call_function prediction = await anyio.to_thread.run_sync( File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\anyio\to_thread.py", line 31, in run_sync return await get_asynclib().run_sync_in_worker_thread( File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\anyio_backends_asyncio.py", line 937, in run_sync_in_worker_thread return await future File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\anyio_backends_asyncio.py", line 867, in run result = context.run(func, *args) File "C:\ai\stable-diffusion-webui\modules\generation_parameters_copypaste.py", line 113, in send_image_and_dimensions img = image_from_url_text(x) File "C:\ai\stable-diffusion-webui\modules\generation_parameters_copypaste.py", line 59, in image_from_url_text filedata = base64.decodebytes(filedata.encode('utf-8')) File "C:\Users\Aleksey\AppData\Local\Programs\Python\Python310\lib\base64.py", line 562, in decodebytes return binascii.a2b_base64(s) binascii.Error: Incorrect padding

[4lt3r3go]

Same problem, but I receive bit different error after update webUI:

File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\gradio\routes.py", line 321, in run_predict output = await app.blocks.process_api( File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\gradio\blocks.py", line 1015, in process_api result = await self.call_function(fn_index, inputs, iterator, request) File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\gradio\blocks.py", line 856, in call_function prediction = await anyio.to_thread.run_sync( File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\anyio\to_thread.py", line 31, in run_sync return await get_asynclib().run_sync_in_worker_thread( File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\anyio_backends_asyncio.py", line 937, in run_sync_in_worker_thread return await future File "C:\ai\stable-diffusion-webui\venv\lib\site-packages\anyio_backends_asyncio.py", line 867, in run result = context.run(func, *args) File "C:\ai\stable-diffusion-webui\modules\generation_parameters_copypaste.py", line 113, in send_image_and_dimensions img = image_from_url_text(x) File "C:\ai\stable-diffusion-webui\modules\generation_parameters_copypaste.py", line 59, in image_from_url_text filedata = base64.decodebytes(filedata.encode('utf-8')) File "C:\Users\Aleksey\AppData\Local\Programs\Python\Python310\lib\base64.py", line 562, in decodebytes return binascii.a2b_base64(s) binascii.Error: Incorrect padding

you get that red error text too in img2img tab? your error looks like the same to mine somehow, i'm not expert btw

[Alexey1327]

This is error from console, in the tab I see just the same Error message in red color. But, yeah, looks like same problem.

[itsagoodbrain]

Just an FYI this happened due to some major changes to the webui's gradio. Borked lots of extensions. Options are to roll back webui repo, wait for image-browser author to update, submit a PR with changes that fix.

[DrMacabre]

Yep same here, you can always drag and drop your original image from browser until there is a fix, settings are still sent to img2img

[4lt3r3go]

so i guess we just have to wait for an update of this extention. Ty guys for reply

[crazyilian]

Looks like pull request #51 fixes the issue.

[edgardmello]

Did not fix here.

[4lt3r3go]

same. ERROR is still there

[EriIaz]

Looks like pull request #51 fixes the issue.

fixed for me

[Kerikhan]

Looks like pull request #51 fixes the issue.

Seems to have worked for me as well. Don't know if it's relevant, but I overwrote the main files with the ones from #51, so I still have the .DS_Store file present in my install.

[Kevman323]

Making the change in #51 fixed the issue for me as well.

[Googolplexed0]

Did not fix here.

same. ERROR is still there

Pull Request #51 fixed the issue for me, but I had actually solved it myself prior to seeing this change. If #51 does not work for you, I would suggest trying my fix, which actually changes stable-diffusion-webui\modules\generation_parameters_copypaste.py. The function I edited begins at line 39 of generation_parameters_copypaste.py. I barely understand GitHub so I don't know how to link to various branches or suggest my changes.

def image_from_url_text(filedata):
    if filedata == None: return None #CUSTOM CODE
    if type(filedata) == list and len(filedata) > 0 and type(filedata[0]) == dict and filedata[0].get("is_file", False):
        filedata = filedata[0]

    if type(filedata) == dict and filedata.get("is_file", False):
        filename = filedata["name"]
        is_in_right_dir = ui_tempdir.check_tmp_file(shared.demo, filename)
        assert is_in_right_dir, 'trying to open image file outside of allowed directories'

        return Image.open(filename)

    if type(filedata) == list:
        if len(filedata) == 0:
            return None

        filedata = filedata[0]

    # if filedata.startswith("data:image/png;base64,"): #ORIGINAL CODE
    #     filedata = filedata[len("data:image/png;base64,"):]

    # filedata = base64.decodebytes(filedata.encode('utf-8'))
    # image = Image.open(io.BytesIO(filedata))
    # return image

    if filedata.startswith("data:image/png;base64,"): #CUSTOM CODE
        filedata = filedata[len("data:image/png;base64,"):]
        # filedata = filedata+"=="
        filedata = base64.decodebytes(filedata.encode('utf-8'))
    else:
        filedata = filedata.encode('utf-8')
    image = Image.open(filedata)
    return image

Just replace the current function with this version and it should work. The original code is trying to decode base64-encoded text, but it tries this whether or not the filedata has the "data:image/png;base64," string in front, which I expect to be an identifier that the decoding is necessary. My fix performs the decoding if the base64 identifier is present, or skips it if unnecessary. While I was digging around, I also realized that clicking any "send to ____" button while no image was present in the above viewer would cause an error, since the if filedata.startswith("data:image/png;base64,") check would be performed on a None type, so I added the one line if filedata == None: return None #CUSTOM CODE to circumvent the error.

[ligerye]

@Googolplexed06 It does work well with this fix, you are the real hero!

[jipsita]

Thank you Googolplexed06

[4lt3r3go]

The function I edited begins at line 39 of generation_parameters_copypaste.py

SOLVED!

i tryed remove that file and let it renew with git pull. nothing. then i made that modification at line 39 manually and voilà.. solved. thank you everyone!

[DrMacabre]

seems to be broken again despite having edited the file. the fix worked until today.

[ice051128]

[SOLVED] this is happening when i press "send to img2img" or "send to inpaint" from the Image Browser tab. ..but same buttons pressed in txt2img tab works fine. :(

Same shit, but why u made the UI look ugly?

[ice051128]

Did not fix here.

same. ERROR is still there

Pull Request #51 fixed the issue for me, but I had actually solved it myself prior to seeing this change. If #51 does not work for you, I would suggest trying my fix, which actually changes stable-diffusion-webui\modules\generation_parameters_copypaste.py. The function I edited begins at line 39 of generation_parameters_copypaste.py. I barely understand GitHub so I don't know how to link to various branches or suggest my changes.

def image_from_url_text(filedata):
    if filedata == None: return None #CUSTOM CODE
    if type(filedata) == list and len(filedata) > 0 and type(filedata[0]) == dict and filedata[0].get("is_file", False):
        filedata = filedata[0]

    if type(filedata) == dict and filedata.get("is_file", False):
        filename = filedata["name"]
        is_in_right_dir = ui_tempdir.check_tmp_file(shared.demo, filename)
        assert is_in_right_dir, 'trying to open image file outside of allowed directories'

        return Image.open(filename)

    if type(filedata) == list:
        if len(filedata) == 0:
            return None

        filedata = filedata[0]

    # if filedata.startswith("data:image/png;base64,"): #ORIGINAL CODE
    #     filedata = filedata[len("data:image/png;base64,"):]

    # filedata = base64.decodebytes(filedata.encode('utf-8'))
    # image = Image.open(io.BytesIO(filedata))
    # return image

    if filedata.startswith("data:image/png;base64,"): #CUSTOM CODE
        filedata = filedata[len("data:image/png;base64,"):]
        # filedata = filedata+"=="
        filedata = base64.decodebytes(filedata.encode('utf-8'))
    else:
        filedata = filedata.encode('utf-8')
    image = Image.open(filedata)
    return image

Just replace the current function with this version and it should work. The original code is trying to decode base64-encoded text, but it tries this whether or not the filedata has the "data:image/png;base64," string in front, which I expect to be an identifier that the decoding is necessary. My fix performs the decoding if the base64 identifier is present, or skips it if unnecessary. While I was digging around, I also realized that clicking any "send to ____" button while no image was present in the above viewer would cause an error, since the if filedata.startswith("data:image/png;base64,") check would be performed on a None type, so I added the one line if filedata == None: return None #CUSTOM CODE to circumvent the error.

A simple parameter change from #51 works flawlessly. What's the point of changing the whole function?

[DrMacabre]

A simple parameter change from #39 works flawlessly. What's the point of changing the whole function?

doens't work here, i just checked the line in #39 and it's already changed on my script

[AlUlkesh]

This whole thing is somewhat confusing. Who has tried out fix #51 and it still doesn't work? Can you please post your whole console log?

[crazyilian]

This whole thing is somewhat confusing. Who has tried out fix #51 and it still doesn't work? Can you please post your whole console log?

@4lt3r3go posted screenshot in the issue description. #51 didn't work for him.

[DrMacabre]

#39 + #51 fixed it for me

[ice051128]

A simple parameter change from #39 works flawlessly. What's the point of changing the whole function?

doens't work here, i just checked the line in #39 and it's already changed on my script

Mentioned the wrong PR, I meant #51

[AlUlkesh]

Well, I just had something interesting happen. Even with fix #51 I suddenly again ran into

    if filedata.startswith("data:image/png;base64,"):
AttributeError: 'NoneType' object has no attribute 'startswith'

Some looking into it later, I can see this in Chrome's DevTools log: index.0b923826.js:76 Uncaught (in promise) TypeError: Cannot read properties of undefined (reading 'forEach') at lt (index.0b923826.js:76:2728) lt @ index.0b923826.js:76 Promise.then (async) (anonymous) @ index.0b923826.js:76 (anonymous) @ index.0b923826.js:4 ae @ index.0b923826.js:4 f @ Number.svelte:16 (anonymous) @ index.0b923826.js:4 (anonymous) @ index.0b923826.js:4 b @ Number.svelte:19 Z.a.$$.update @ Number.svelte:32 Rl @ index.0b923826.js:4 mt @ index.0b923826.js:4 Promise.then (async) so @ index.0b923826.js:4 zl @ index.0b923826.js:4 (anonymous) @ index.0b923826.js:4 (anonymous) @ index.0b923826.js:76 lt @ index.0b923826.js:76 Promise.then (async) (anonymous) @ index.0b923826.js:76 (anonymous) @ index.0b923826.js:4 ae @ index.0b923826.js:4 _ @ Button.svelte:10 (anonymous) @ index.0b923826.js:4 ae @ index.0b923826.js:4 B @ Button.svelte:9 (anonymous) @ (index):1947 (index):1799 in images_history_click_image ! (index):1800 DOMTokenList(3) ['gallery-item', 'group', 'svelte-1g9btlg', value: 'gallery-item group svelte-1g9btlg'] (index):1875 tabName img2img (index):1876 img_index -1 (index):1877 img2img_images_history_set_index (index):1799 in images_history_click_image ! (index):1800 DOMTokenList(3) ['gallery-item', 'group', 'svelte-1g9btlg', value: 'gallery-item group svelte-1g9btlg'] (index):1875 tabName txt2img (index):1876 img_index -1 (index):1877 txt2img_images_history_set_index (index):1799 in images_history_click_image ! (index):1800 DOMTokenList(9) ['gallery-item', '!flex-none', '!h-9', '!w-9', 'transition-all', 'duration-75', 'scale-90', 'transform', 'svelte-1g9btlg', value: 'gallery-item !flex-none !h-9 !w-9 transition-all duration-75 scale-90 transform svelte-1g9btlg'] (index):1875 tabName txt2img (index):1876 img_index 2 (index):1877 txt2img_images_history_set_index (index):1799 in images_history_click_image !

So, the first time I clicked on an image I got img_index -1, which would lead to the error message about base64. Second click on the same image, same thing. Third click and now I get the correct img_index 2

So what is going on here? Directly before this is an error message connected to asynchronous operations. I don't know if that has anything to do with the problem, but if it does that might hint at a problem with the threading involved

[crazyilian]

Well, I just had something interesting happen. Even with fix #51 I suddenly again ran into

    if filedata.startswith("data:image/png;base64,"):
AttributeError: 'NoneType' object has no attribute 'startswith'

This problem was mentioned and fixed in pull request AUTOMATIC1111/stable-diffusion-webui#6541. Because of the other problems, it is still not merged.

[AlUlkesh]

This problem was mentioned and fixed in pull request AUTOMATIC1111/stable-diffusion-webui#6541.

Where do you see this there? I can only find a mention for if no image was present in the viewer. In my case the picture was visible.

[AlUlkesh]

So my current understanding is:

• #51 solves the issue for some people, some of the time
• @Googolplexed06 's solution works for most people, but Automatic makes a good point, that a function that decodes base64 should not handle something else

I hope I might now have found a new solution, that does not require a change in webui's code.

This is how we access generation_parameters_copypaste: modules.generation_parameters_copypaste.bind_buttons(send_to_buttons, img_file_name, img_file_info)

The second parameter is currently img_file_name, which is actually a gradio textbox object with should return the filename.

#51 changes this to history_gallery, the gradio gallery object, which should return a list of images.

generation_parameters_copypaste.py's "entry" code for us is:

def send_image_and_dimensions(x):
    if isinstance(x, Image.Image):
        img = x
    else:
        img = image_from_url_text(x)

So both versions, filename and list of images, will be handled by the else, which is not really that function's purpose.

I now found that the images-browser has yet another object called hidden, which is a (hidden) gradio.image and contains the selected image and so would not have to go into the else.

So I changed the line into: modules.generation_parameters_copypaste.bind_buttons(send_to_buttons, hidden, img_file_info)

and so far it seems to work all the time.

Could someone else please confirm this?

https://github.com/AlUlkesh/stable-diffusion-webui-images-browser/commit/d223344938b9580da20f163b4025d26cc16ddf5c

[ibushong]

I found this thread/issue after looking into this error. I haven't dug into the history of previous fixes nor the details of what the change above is doing, but it seems to work!

[crazyilian]

Could someone else please confirm this?

AlUlkesh@d223344

#51 worked for me, this fix works too. Will you create a pull request? @AlUlkesh