Flux kontext issues with event emmiters

[Haervwe]

Hey!
So just a quick summary of the final issues.

1.Need to add random seed with each Ace step generation, as each song sound exactly the same as previous

2.Once Flux Kontext sends the edited image, the expandable enhanced prompt is gone, making it impossible to know exactly what the enhanced prompts was after the fact.

3.Flux Kontext final step to send back the image (processing..) can be avoided with normal save image node (like the normal open webui image creation using comfyui)
It makes the images not accessible through tailscale+caddy https for me, so I need to use http. Normal image generation works fine over https.
It also adds more time to provide the output image.

Originally posted by @iChristGit in #19

[iChristGit]

Would like a mention once the issues with both tools are fixed, il update from hub!
Thanks man

[Haervwe]

Hi ! all issues are fixed now ! please check the hub and tell me if its working on your end ! :)

[iChristGit]

Hey, Thanks for the heads-up

ace step:
First, when disabling save to /cache , it shows something that looks like an error, and should just be "Song Ready" and provide local link

second, when saving to /cache instead of doing an hyperlink, it shows the clickable link, it should be "Song Ready - press to listen"
Now it shows like that:

Should look like that:

The seed randomize works, so thats great!

As for Flux Kontext, I cannot get the script to work now,
And it still seems to use the old way of retrieving the image back from comfyui, with the "processing..." part, the optimal solution is using the same "save image" node and show the image in the ui like the usual comfyui integration in open-webui

if I use vision to enhance prompt I get this:

7ZGYXJkRkZwzVm3D22odxcGi4ZvqWekxjv24dpU5bDUBt5Py129pl9SI9d97nVbw7bv6WzH3jyxpVeRyjzrle32Keu1j9NENfU/jxPcjpTZcCCzPt8D12gZmyEvvGelAoJPbSy3Dj0ZzUBh3Y0ZjQ5eInyIL6zoLLbM1EdRdWaUAdS/yuEJQW0zw11YASCgQCGuF9G8LdHkzbSAr8QosKtQiflfgakULR4v4F9uGhqTLIPR/jnFukKagJ2uA5jiGUt+wYVFeUG+TS1oKiaUe4Teqg/9uAZfH5EY1z6FE7fgbKoPzs7Xk8UOzOf7ajvtCx1RASJjWu2l6ZQgyu8pZQj5DQhBAk+GmcLuBr4RUgI5UVsET21Yl7nw2Kqo9UEkOoCY6D4ggEljF6UdDox1xVrvAS3gZFMXwh3OUywsktAWZspOHfHAH4DFSOv9lXZlVPWp5gD/+qPPeqlSOlgwP8+Z0WlyOgWjuG1+/ZmGM9S2MdoJ3
2025-08-04 09:26:53.643 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "POST /api/chat/completions HTTP/1.1" 200 - {}
2025-08-04 09:26:53.657 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-04 09:26:53.664 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
### Task:
You are an autocompletion system. Continue the text in `<text>` based on the **completion type** in `<type>` and the given language.

### **Instructions**:
1. Analyze `<text>` for context
#atvepS
2025-08-04 09:26:55.392 | ERROR    | open_webui.functions:stream_content:282 - Error: 'NoneType' object is not callable - {}
2025-08-04 09:26:55.395 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
### Task:
Suggest 3-5 relevant follow-up questions or prompts that the user might naturally ask next in this conversation as a **user**, based on the chat history, to help continue or deepen the discu
#s p{
2025-08-04 09:26:55.433 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "POST /api/v1/tasks/auto/completions HTTP/1.1" 200 - {}
2025-08-04 09:26:55.497 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58283 - "POST /api/chat/completed HTTP/1.1" 200 - {}
2025-08-04 09:26:55.590 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "POST /api/v1/chats/562b74c3-a2c6-4a41-b813-8e7aa9ae5011 HTTP/1.1" 200 - {}
2025-08-04 09:26:55.596 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
### Task:
Generate a concise, 3-5 word title with an emoji summarizing the chat history.
### Guidelines:
- The title should clearly represent the main theme or subject of the conversation.
- Use emoji
#so aor
2025-08-04 09:26:55.666 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-04 09:26:55.672 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-04 09:26:55.757 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
### Task:
Generate 1-3 broad tags categorizing the main themes of the chat history, along with 1-3 more specific subtopic tags.

### Guidelines:
- Start with high-level domains (e.g. Science, Technolo
#g u
2025-08-04 09:26:55.842 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-04 09:26:55.844 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58286 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-04 09:26:55.855 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58286 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-04 09:26:55.857 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:58278 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-04 09:26:55.936 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}

Error in open-webui without vision after the image generation finishes (no output in webui)

2025-08-04 09:24:52.609 | ERROR    | function_comfyui_i2i_universal_pipe_v2:pipe:509 - An unexpected error occurred in pipe: 'NoneType' object is not callable - {}
Traceback (most recent call last):

  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code

  File "C:\Python\Scripts\open-webui.exe\__main__.py", line 7, in <module>
    sys.exit(app())
    │   │    └ <typer.main.Typer object at 0x0000019607ECA6D0>
    │   └ <built-in function exit>
    └ <module 'sys' (built-in)>

  File "C:\Python\Lib\site-packages\typer\main.py", line 324, in __call__
    return get_command(self)(*args, **kwargs)
           │           │      │       └ {}
           │           │      └ ()
           │           └ <typer.main.Typer object at 0x0000019607ECA6D0>
           └ <function get_command at 0x000001960A6C4900>
  File "C:\Python\Lib\site-packages\click\core.py", line 1442, in __call__
    return self.main(*args, **kwargs)
           │    │     │       └ {}
           │    │     └ ()
           │    └ <function TyperGroup.main at 0x000001960A6A5300>
           └ <TyperGroup >
  File "C:\Python\Lib\site-packages\typer\core.py", line 757, in main
    return _main(
           └ <function _main at 0x000001960A6A4400>
  File "C:\Python\Lib\site-packages\typer\core.py", line 195, in _main
    rv = self.invoke(ctx)
         │    │      └ <click.core.Context object at 0x000001960B2C95D0>
         │    └ <function Group.invoke at 0x000001960835C900>
         └ <TyperGroup >
  File "C:\Python\Lib\site-packages\click\core.py", line 1830, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
           │               │       │       │      └ <click.core.Context object at 0x0000019607A22750>
           │               │       │       └ <function Command.invoke at 0x000001960835B6A0>
           │               │       └ <TyperCommand serve>
           │               └ <click.core.Context object at 0x0000019607A22750>
           └ <function Group.invoke.<locals>._process_result at 0x000001960B2BD800>
  File "C:\Python\Lib\site-packages\click\core.py", line 1226, in invoke
    return ctx.invoke(self.callback, **ctx.params)
           │   │      │    │           │   └ {'host': '0.0.0.0', 'port': 8080}
           │   │      │    │           └ <click.core.Context object at 0x0000019607A22750>
           │   │      │    └ <function serve at 0x000001960B2BC7C0>
           │   │      └ <TyperCommand serve>
           │   └ <function Context.invoke at 0x000001960835A8E0>
           └ <click.core.Context object at 0x0000019607A22750>
  File "C:\Python\Lib\site-packages\click\core.py", line 794, in invoke
    return callback(*args, **kwargs)
           │         │       └ {'host': '0.0.0.0', 'port': 8080}
           │         └ ()
           └ <function serve at 0x000001960B2BC7C0>
  File "C:\Python\Lib\site-packages\typer\main.py", line 699, in wrapper
    return callback(**use_params)
           │          └ {'host': '0.0.0.0', 'port': 8080}
           └ <function serve at 0x000001960B1994E0>
  File "C:\Python\Lib\site-packages\open_webui\__init__.py", line 78, in serve
    uvicorn.run(
    │       └ <function run at 0x000001960A92F6A0>
    └ <module 'uvicorn' from 'C:\\Python\\Lib\\site-packages\\uvicorn\\__init__.py'>
  File "C:\Python\Lib\site-packages\uvicorn\main.py", line 580, in run
    server.run()
    │      └ <function Server.run at 0x000001960A92F920>
    └ <uvicorn.server.Server object at 0x00000196807CA310>
  File "C:\Python\Lib\site-packages\uvicorn\server.py", line 66, in run
    return asyncio.run(self.serve(sockets=sockets))
           │       │   │    │             └ None
           │       │   │    └ <function Server.serve at 0x000001960A92F9C0>
           │       │   └ <uvicorn.server.Server object at 0x00000196807CA310>
           │       └ <function run at 0x000001960A7E6660>
           └ <module 'asyncio' from 'C:\\Python\\Lib\\asyncio\\__init__.py'>
  File "C:\Python\Lib\asyncio\runners.py", line 190, in run
    return runner.run(main)
           │      │   └ <coroutine object Server.serve at 0x000001960A873880>
           │      └ <function Runner.run at 0x000001960A7E6A20>
           └ <asyncio.runners.Runner object at 0x0000019680610910>
  File "C:\Python\Lib\asyncio\runners.py", line 118, in run
    return self._loop.run_until_complete(task)
           │    │     │                  └ <Task pending name='Task-1' coro=<Server.serve() running at C:\Python\Lib\site-packages\uvicorn\server.py:70> wait_for=<Futur...
           │    │     └ <function BaseEventLoop.run_until_complete at 0x000001960A7E45E0>
           │    └ <ProactorEventLoop running=True closed=False debug=False>
           └ <asyncio.runners.Runner object at 0x0000019680610910>
  File "C:\Python\Lib\asyncio\base_events.py", line 637, in run_until_complete
    self.run_forever()
    │    └ <function ProactorEventLoop.run_forever at 0x000001960A8594E0>
    └ <ProactorEventLoop running=True closed=False debug=False>
  File "C:\Python\Lib\asyncio\windows_events.py", line 321, in run_forever
    super().run_forever()
  File "C:\Python\Lib\asyncio\base_events.py", line 604, in run_forever
    self._run_once()
    │    └ <function BaseEventLoop._run_once at 0x000001960A7E6340>
    └ <ProactorEventLoop running=True closed=False debug=False>
  File "C:\Python\Lib\asyncio\base_events.py", line 1909, in _run_once
    handle._run()
    │      └ <function Handle._run at 0x000001960A7831A0>
    └ <Handle <TaskStepMethWrapper object at 0x0000019681159870>()>
  File "C:\Python\Lib\asyncio\events.py", line 80, in _run
    self._context.run(self._callback, *self._args)
    │    │            │    │           │    └ <member '_args' of 'Handle' objects>
    │    │            │    │           └ <Handle <TaskStepMethWrapper object at 0x0000019681159870>()>
    │    │            │    └ <member '_callback' of 'Handle' objects>
    │    │            └ <Handle <TaskStepMethWrapper object at 0x0000019681159870>()>
    │    └ <member '_context' of 'Handle' objects>
    └ <Handle <TaskStepMethWrapper object at 0x0000019681159870>()>
  File "C:\Python\Lib\site-packages\open_webui\utils\middleware.py", line 2092, in response_handler
    await stream_body_handler(response, form_data)
          │                   │         └ {'stream': True, 'model': 'comfyui_i2i_universal_pipe_v2', 'messages': [{'role': 'user', 'content': [{'type': 'text', 'text':...
          │                   └ <starlette.responses.StreamingResponse object at 0x000001968232D990>
          └ <function process_chat_response.<locals>.response_handler.<locals>.stream_body_handler at 0x00000196823949A0>
  File "C:\Python\Lib\site-packages\open_webui\utils\middleware.py", line 1807, in stream_body_handler
    async for line in response.body_iterator:
                      │        └ <async_generator object generate_function_chat_completion.<locals>.stream_content at 0x0000019680D70430>
                      └ <starlette.responses.StreamingResponse object at 0x000001968232D990>
  File "C:\Python\Lib\site-packages\open_webui\functions.py", line 270, in stream_content
    res = await execute_pipe(pipe, params)
                │            │     └ {'body': {'stream': True, 'model': 'comfyui_i2i_universal_pipe_v2', 'messages': [{'role': 'user', 'content': [{'type': 'text'...
                │            └ <bound method Pipe.pipe of <function_comfyui_i2i_universal_pipe_v2.Pipe object at 0x0000019680E65750>>
                └ <function generate_function_chat_completion.<locals>.execute_pipe at 0x00000196823D45E0>
  File "C:\Python\Lib\site-packages\open_webui\functions.py", line 141, in execute_pipe
    return await pipe(**params)
                 │      └ {'body': {'stream': True, 'model': 'comfyui_i2i_universal_pipe_v2', 'messages': [{'role': 'user', 'content': [{'type': 'text'...
                 └ <bound method Pipe.pipe of <function_comfyui_i2i_universal_pipe_v2.Pipe object at 0x0000019680E65750>>
> File "<string>", line 490, in pipe

TypeError: 'NoneType' object is not callable

[iChristGit]

Also it seems like even in the older version of the script, the "Enhanced prompt" that actually get generated and shown in the Ui does not go to comfyui, it send the original user prompt as the image prompt.

[iChristGit]

So overall the Ace-step tool is almost ready, just formatting the link correctly, and a different message when user saved to local and not to /cahce
Could also be nice to system prompt it to always provide the link first, before the Lyrics/Tags used, so user can listen as fast as possible to the result

Flux Kontext tool needs some work.

[iChristGit]

Maybe this would help, this is my system prompt for Ace-step pipeline:

You are here to decide on tags and lyrics, pass it to ComfyUI and get back a URL of a ready to listen song to paste!
When user asks for instrumental song, the lyrics should just be “[Inst]”.
The tags should contain only words that can guide the music generation towards a great end result.
Be creative with the genre tags, lyrics, so the resulting song is always unique and of high quality.
Always provide a nicely formatted and named link to the song, and after the clickable link send the lyrics & tags of the song.

Examples:

Genres & Styles:
pop, rock, edm, house, techno, trance, dubstep, drum and bass, trap, hip hop, rap, lofi, ambient, chillwave, synthwave, vaporwave, funk, disco, soul, r&b, gospel, jazz, blues, reggae, ska, punk, metal, death metal, hard rock, acoustic, folk, country, bluegrass, latin, salsa, bachata, merengue, tango, flamenco, afrobeat, afro-cuban, world, k-pop, j-pop, c-pop, bollywood, middle eastern, celtic, classical, orchestral, soundtrack, industrial, experimental

Moods & Atmospheres:
upbeat, happy, joyful, melancholic, emotional, romantic, dark, moody, epic, powerful, aggressive, uplifting, relaxing, dreamy, mysterious, energetic, smooth, soulful, funky, dramatic, tense, inspiring, hopeful, nostalgic, retro, futuristic, cyberpunk, spacey, tropical, warm, cold, serene

Instruments & Elements:
piano, acoustic guitar, electric guitar, bass guitar, fretless bass, slap bass, violin, cello, viola, harp, saxophone, trumpet, trombone, flute, clarinet, harmonica, drums, percussion, congas, bongos, marimba, steel drums, tabla, sitar, oud, qanun, synths, modular synth, pads, arpeggiator, strings section, choir, male vocals, female vocals, rap vocals, vocoder, 808s, kick-heavy, crisp hi-hats, snare rolls, sub bass, plucked strings

Tempo & Production:
slow bpm, mid-tempo, fast bpm, 80 bpm, 90 bpm, 100 bpm, 120 bpm, 128 bpm, 140 bpm, acoustic mix, unplugged, live session, studio polished, raw lo-fi, analog warmth, vintage, retro 80s, retro 90s, modern mix, orchestral hybrid, cinematic mix, layered harmonies, deep stereo imaging


Example Tag Combinations:

Pop:
pop, bright vocals, layered harmonies, electric piano, acoustic guitar, punchy snare, warm bass, lush strings, 100 bpm, studio polished
Rap:
rap, dry punchy 808s, snappy clap, gritty sub bass, rap vocals, sharp hi-hats, vocal chops, 90 bpm, modern trap mix
Hip Hop:
hip hop, mellow electric piano, deep kick, crisp snare, soulful samples, warm bassline, rap vocals, vinyl texture, 85 bpm, analog warmth

Remember to be creative and use at least 1 of each (Genres, Styles, Instruments, Moods, Elements, Tempo, Production)
Try to combine different matching Tags to produce interesting and unique music!
You can create and imagine more creative tags to match any user prompt.

/no think

[iChristGit]

Another Issue I face when using from my phone (tailscale):

If I specify in the system prompt the url, it does work by just changing the file name to the current song.

mix"
}
"Generation completed (Job: 11792f5d-93dc-41f9-9eed-8b5a9777900d) but no audio files found. Job outputs: json\n{\n \"audio\": [\n {\n \"filename\": \"ComfyUI_00144_.mp3\",\n \"subfolder\": \"audio\",\n \"type\": \"output\"\n }\n ]\n}\n"

[Haervwe]

Hi! thanks for the feedback! for the song generation is all solved and on the hub!
""Generation completed (Job: 11792f5d-93dc-41f9-9eed-8b5a9777900d) but no audio files found. Job outputs: json\n{\n "audio": [\n {\n "filename": "ComfyUI_00144_.mp3",\n "subfolder": "audio",\n "type": "output"\n }\n ]\n}\n"! this one i dont get whats happening since its indiferent if you call it from the phone, pc , or whatever, it all hapens in the backend

and for the kontext error, its seems to be some incompatibilty with the image file format or size and ollama, can you share metadata on the image? (or the image) if you want we can setup a quick discrod call and check all the issues in a blink!

[iChristGit]

Thanks for the update!

I am not sure what is wrong with Flux, the old script did work without errors.
I used regular png images that works fine with the older script.

In your case it does enhance the prompt and provides the enhanced prompt to comfyui?

Also did you manage to use the regular "image save" node at the end instead of processing the BASE64 value?

Il go deeper if it all works in your end

[iChristGit]

As far as Ace-step, it all works great! thanks for the changes

As for Flux Kontext, I use any picture and It cannot process it..

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
2025-08-05 08:20:37.672 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51695 - "POST /api/chat/completions HTTP/1.1" 200 - {}
2025-08-05 08:20:37.683 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51695 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-05 08:20:37.692 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51695 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
### Task:
You are an autocompletion system. Continue the text in `<text>` based on the **completion type** in `<type>` and the given language.

### **Instructions**:
1. Analyze `<text>` for context
#atvepi
2025-08-05 08:20:41.741 | ERROR    | open_webui.functions:stream_content:282 - Error: 'NoneType' object is not callable - {}
2025-08-05 08:20:41.741 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
### Task:
Suggest 3-5 relevant follow-up questions or prompts that the user might naturally ask next in this conversation as a **user**, based on the chat history, to help continue or deepen the discu
#s p{
2025-08-05 08:20:41.791 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51695 - "POST /api/v1/tasks/auto/completions HTTP/1.1" 200 - {}
2025-08-05 08:20:41.944 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
### Task:
Generate a concise, 3-5 word title with an emoji summarizing the chat history.
### Guidelines:
- The title should clearly represent the main theme or subject of the conversation.
- Use emoji
#so aor
2025-08-05 08:20:41.995 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51695 - "POST /api/chat/completed HTTP/1.1" 200 - {}
2025-08-05 08:20:42.095 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
2025-08-05 08:20:42.143 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51695 - "POST /api/v1/chats/13ddc7c6-915c-4bed-907f-7af2e1c084f8 HTTP/1.1" 200 - {}
### Task:
Generate 1-3 broad tags categorizing the main themes of the chat history, along with 1-3 more specific subtopic tags.

### Guidelines:
- Start with high-level domains (e.g. Science, Technolo
#g u
2025-08-05 08:20:42.233 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51703 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-05 08:20:42.236 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51705 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-05 08:20:42.237 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51695 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-05 08:20:42.238 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51704 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-05 08:20:42.274 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51695 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-05 08:20:42.274 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51703 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-05 08:20:42.317 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:51705 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-05 08:20:42.320 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format

I use the default nodeID and workflow, just changing to localhost the ollama api url
a discord chat would be helpful

[iChristGit]

Initially I get this:

After another try, it does go through to comfyui, but the critical issue of not using the enhanced prompt is still present:

[{'type': 'text', 'text': ',make her smile'},

It does not use the enhanced prompt it crafted..

[iChristGit]

Settings and full comfy and open-webui logs for reference:

FU6lVwwRyBtYAR4MphoXWiLQBLiETWHQkgGvc0BRPETeKkUHZHxD7YaWenDCCKZxD3pL6LCXzGcB44Fw4nbetbo9WNqaYcNsxQ84/xjPyumlZr1/W16/Qg+8uoG925yW2ijHeoKUD62IIV2GOKDGSdD5lEnIw7W8MTAWHM/bgsztmgXtFcz3aXJEzH5h5y/HZrrud4jusJPS4ckQcVjIOtPbC5dQOMpRJxlLzT0Rwfr7G1XIYK/a+sUPR4weRYQ6+oy2A3ESAdijIP2R5hBXilc5/HTH7eI34/Azd0/Xi0zksl5hBH4KUAR3uCBkt/KhTowDg+e0OWjtwrD5m/PD0hpDh9HL++gsaqWvpaQH2xnk8eQed1OV8TedLSAvmUpUaXQSnP4acfIX+z9h6e+zbOYSP20e/w6htHz+e739ujxs6k/MKNWrmAlrZ9/1ZhwPHSCeB1MwUM9AwShaxl5GWM9iQGLaC8GHH7C0lIcXjR52nVsxAEDkTM1xYqGGyHxR/DIQPmJCg3ORwDcPFYFRQoJBVo3mSReGen1BYrB9qhErcADOYa4dqKgEuKrzcDcFV817KOyEfPiG5wS0zyQbxeNKYAME2m4wGWrvGepjQnppHqaqhXhyzs0asprYV0af5WmjMRtcwNEH8FlwvsRe2qQCEzI+crmy+IZOJVg0YmzOYNb3D1xsMNUNm90coLCk4A+cMcA3xkT59IHUlUWVU//Z/roB+z27Yw3PpMkzGgW6AnmamK2VYZNSOdTOKvFRHoKqhkEByBLTKNDW8MkvMCWeThwNSHr+UDwfaHAZRJ5Lm11ryiwwTQGhbVEYvlrwyFsFR7Bf6w+gBQ7haZPz7864mGmNoYgoYsztiOjqty8Trc+m4CJCQnWjelCL1z/Fm8y1cUEsk6a/xi37Fkq8HSLFLIovYFmgiV8J3uRUDDzZqgO13VE2H0oISkFwNJPzsZOgo2GoqVgRpgu0uVARnGDF1pbRPFGIbnDDpIGVs+AKhVu0L1XW0jeZGsrwinhgp6cz6apXImFDkyidLVVb0MSHOpbI7s4YGSSPaszAKkFQHLQaDHPtqW1FmCo875c6ZmW47bzalx54LeRxEGBYhqxIngRTdw6dekd5jR0WPV12lfD3jFZsOydYfHZa5Bewita/mQdR68XEKYc8Cy9/u5dh4ZODoIvRu2Fo2cM9r02fO29MKHAQGLjuFHfhVXFUL1HUYPj/G8ghCvQQ3UkK0z5xHSWeMF2xUDdDxFdQQ02gDIF/RsT4IEFLMaMgiBYjgwxZgVhm5M2foWUC8hNa2+QPBSSIZuSUXOfMTEj3sVh4D0MKSCJI/Q6uo4Sszs4BQVpnkr4QWAiWNyJHSdQj9WyaYnTiGNW1OAv2RNoGZB/7uOMDH/tM5MBBoFQP+YG4r+z7bD4pvzXqDtKqEBa7fKiaIkewFU9qODpSkR6Z0UbLmL2Y6pq3GS5gZQWSWWa6B5Lsffa8b02d2zLXHjdVrj4eXRLruZLIwi6FtYi/T0BJp4fC28F8wfv/Wdg+EPrMFoNfy92ox+bSvdkLoB2elTs+L8+kAaCvZbPAsdM2Kge0Uy8jbEcJvdIThniQDeJhlc7D25xmh6esNJeiQ8g+Ywpz+PxTHSxPEH47CAAAAAElFTkSuQmCC"
        },
        "class_type": "ETN_LoadImageBase64",
        "_meta": {
          "title": "Load Image (Base64)"
        }
      }
    },
    {
      "client_id": "997251f4-7709-4d08-bf26-5d23d025b88c"
    },
    [
      "136",
      "194"
    ]
  ],
  "outputs": {
    "194": {
      "images": []
    },
    "136": {
      "images": [
        {
          "filename": "owui_00011_.png",
          "subfolder": "owui",
          "type": "output"
        }
      ]
    }
  },
  "status": {
    "status_str": "success",
    "completed": true,
    "messages": [
      [
        "execution_start",
        {
          "prompt_id": "594b7c25-d5ba-4059-9665-0ec247b081e5",
          "timestamp": 1754371702784
        }
      ],
      [
        "execution_cached",
        {
          "nodes": [
            "37",
            "38",
            "39"
          ],
          "prompt_id": "594b7c25-d5ba-4059-9665-0ec247b081e5",
          "timestamp": 1754371702798
        }
      ],
      [
        "execution_success",
        {
          "prompt_id": "594b7c25-d5ba-4059-9665-0ec247b081e5",
          "timestamp": 1754371763083
        }
      ]
    ]
  },
  "meta": {
    "194": {
      "node_id": "194",
      "display_node": "194",
      "parent_node": null,
      "real_node_id": "194"
    },
    "136": {
      "node_id": "136",
      "display_node": "136",
      "parent_node": null,
      "real_node_id": "136"
    }
  }
} - {}
### Task:
Suggest 3-5 relevant follow-up questions or prompts that the user might naturally ask next in this conversation as a **user**, based on the chat history, to help continue or deepen the discu
#s p{
2025-08-05 08:29:26.568 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52028 - "GET /static/favicon.png HTTP/1.1" 200 - {}
2025-08-05 08:29:28.141 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
### Task:
Generate a concise, 3-5 word title with an emoji summarizing the chat history.
### Guidelines:
- The title should clearly represent the main theme or subject of the conversation.
- Use emoji
#so aor
2025-08-05 08:29:28.286 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
### Task:
Generate 1-3 broad tags categorizing the main themes of the chat history, along with 1-3 more specific subtopic tags.

### Guidelines:
- Start with high-level domains (e.g. Science, Technolo
#g u
2025-08-05 08:29:28.389 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52036 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-05 08:29:28.391 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52028 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-05 08:29:28.391 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52035 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
2025-08-05 08:29:28.456 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52028 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-05 08:29:28.463 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52028 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-05 08:29:28.467 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52028 - "GET /api/v1/folders/ HTTP/1.1" 200 - {}
2025-08-05 08:29:28.474 | ERROR    | open_webui.functions:generate_function_chat_completion:312 - Error: 500: Failed to create new sequence: failed to process inputs: image: unknown format
 - {}
2025-08-05 08:29:36.707 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52038 - "GET /_app/version.json HTTP/1.1" 304 - {}
2025-08-05 08:29:44.774 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 127.0.0.1:52039 - "GET /_app/version.json HTTP/1.1" 304 - {}

got prompt
Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
model weight dtype torch.float8_e4m3fn, manual cast: torch.float16
model_type FLUX
Requested to load AutoencodingEngine
loaded completely 18966.875 159.87335777282715 True
Using scaled fp8: fp8 matrix mult: False, scale input: False
Requested to load FluxClipModel_
loaded completely 9.5367431640625e+25 4903.231597900391 True
CLIP/text encoder model load device: cpu, offload device: cpu, current: cpu, dtype: torch.float16
clip missing: ['text_projection.weight']
Requested to load Flux
loaded completely 20505.99356211853 11350.067443847656 True
100%|███████████████████████████████████████████████████████████████████████████████████████████| 20/20 [00:49<00:00,  2.46s/it]
Prompt executed in 74.01 seconds
got prompt
100%|███████████████████████████████████████████████████████████████████████████████████████████| 20/20 [00:51<00:00,  2.57s/it]
Prompt executed in 60.31 seconds

[Haervwe]

I'm trying to replicate but couldn't, the emmiters are fixed in the latest version, but I don't understand what is that error you are having,.it seems it might be , the nontype error is weird, does this happen with Gemma 12b? That the vision model I use, maybe it's outputting the prompt in a weird way that's making the parsing fail, other thing that can be is that the image is too.large but is quite weird, does it happen every time ? Add me on discord I'm @Haervwe also there