Image cache error after update — generation stops when Queue > 4 (Noobl AI XL / Reference mode)

[usr7072]

Description:
After the latest update of Krita AI Diffusion (and the local ComfyUI server), image generation stops with a cache error when the queue size exceeds 4 tasks.

Environment:

• Model: Noobl AI XL
• Mode: Reference mode
• Connection: Local Server mode
• Krita AI Diffusion & LocalServer both updated to latest version
• OS: Windows 11

Steps to reproduce:

1. Launch Krita AI Diffusion in LocalServer mode.
2. Select Noobl AI XL and enable Reference mode.
3. Create a queue of 8 tasks (my usual setting).
4. Start the generation process.

Expected behavior:
All queued tasks should complete successfully.

Actual behavior:
When the queue count exceeds 4, generation stops with an error like the following:

C:\Users\Username\AppData\Roaming\krita\ai_diffusion\server\ComfyUI\comfy_extras\nodes_easycache.py:267: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq].

!!! Exception during processing !!!
Image with ID b64e24cc-f045-4c27-bcca-87bc34598270 not found in cache.
Traceback (most recent call last):
  File "C:\Users\Username\AppData\Roaming\krita\ai_diffusion\server\ComfyUI\custom_nodes\comfyui-tooling-nodes\nodes.py", line 179, in execute
    raise ValueError(f"Image with ID {id} not found in cache.")

ValueError: Image with ID b64e24cc-f045-4c27-bcca-87bc34598270 not found in cache.

When the queue is limited to 4, everything works normally.
If the queue drops to 1 remaining and I press Generate again, the newly added tasks finish correctly.

What I’ve tried:
Deleted the server directory and reinstalled the LocalServer.
Restarted Krita and the LocalServer multiple times.
The issue still persists.

Additional note:
This issue did not occur before the latest update.

[Acly]

What do you mean with Reference mode - using a "Reference" control layer, or just strength less than 100%?

How long does precessing 4 images take for you?

I've tried to reproduce it with multiple input images and queuing up to 16 images, but didn't encounter the issue so far.

[abrageddon]

I'm having the same issue. Indeed, It is when using strength less than 100%.

[usr7072]

I apologize if my explanation is unclear; I’m using a translation tool.
The prompt on the top-left is set to “Generate.”
The prompt on the bottom-left is set to “Reference,” and I’ve selected the layer to reference.
Strength is set to 100%.
I set Batches to a number greater than 4.
When I click the Generate button to start, the process stops with an error right after completing 4 images.

I’ve attached a screenshot—please refer to the image to understand the environment I’m using.

When the intensity was reduced to 96%, it functioned up to generating the third image, but an error occurred when generating the fourth image.

[hjmeneses]

Same here, using flux, get an error about image missing in cache

[sfisher]

This isn't a huge deal to me, but FWIW, I get this occasionally recently after updating both the Krita Ai Diffusion code and the Comfy UI Tooling node.

It gives an error like Server error: Image with ID 04ae8cab-dabe-4b6c-b031-127aa912adfa not found in cache. and the trace I see in ComfyUI looks like the one below. I'm running my own custom ComfyUI server not the embedded one.

It doesn't happen consistently for me but there are some situations that seem to bring it on--mostly having to do with queueing a bunch of refine ("img2img") generations. I don't know if I've seen it from a fresh images with denoising strength of 1. I don't think it is tied to any particular safetensors model, though I see it most with Flux-1.dev-q8, but probably just because I use that model more frequently. I'm running a 5070ti and 64GB RAM, but I don't believe it's related to the video card or any system problems that I'm aware of.

If I had to guess, it may be related to some kind async process or a weird race condition since it seems fairly inconsistent and doesn't always happen after the same number of generations or same number in the queue and the behavior isn't fully predictable. It feels sort of random when it happens to me but tends to come up when there are a lot of items in the queue.

I can probably get it to happen more than 50% of the time with some elaborate steps like these:

1. Generate around 10-20 images from scratch (fast Stable Diffusion models are fine) at around 1024x1024 resolution and click the ✅ in the generation results on each image to add them as layers.
2. Click the dropdown arrow for "batches" and set it to 2 and change checkpoint to "Flux.1-dev" (though would probably happen with other checkpoints, also). Set the denoising strength to about 87% (The exact number probably doesn't matter, either).
3. Click the Refine button.
4. Click the visibility on the top layer to make it invisible.
5. Repeat steps 3 & 4 until you have 2 refined images in the queue for each layer (there should be between 20-40 images in the queue now).
6. Let it run for a while.
7. After about 15-30 images there is a good chance it will have the error and it will stop processing the rest of the queue at that point.

I've also had it happen by just refining the same image a lot of times. (Set a denoising strength, set batches to 2, and click the "Refine" button a bunch of times until you have 20-50 items in the queue.)

This seems to be a new things since I could queue up at least 100 things in the past and it wouldn't error out randomly after generating 20-40 (which seems to be the usual time frame for me if I queue a lot of things).

Example of the output I'm seeing.

# ... a bunch of generations work . . .
Prompt executed in 33.50 seconds
loaded completely 13357.371 12245.509887695312 True
100%|█████████████████████████████████████████████████████████████████████████████████████| 26/26 [00:31<00:00,  1.21s/it]
Requested to load AutoencodingEngine
loaded completely 190.02435302734375 159.87335777282715 True
Prompt executed in 32.84 seconds


!!! Exception during processing !!! Image with ID 84e22cf5-fc57-4b62-8ad3-2846114e61a2 not found in cache.
Traceback (most recent call last):
  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 496, in execute
    output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, hidden_inputs=hidden_inputs)
                                                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 315, in get_output_data
    return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, hidden_inputs=hidden_inputs)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 289, in _async_map_node_over_list
    await process_inputs(input_dict, i)
  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 277, in process_inputs
    result = f(**inputs)
             ^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\comfy_api\internal\__init__.py", line 149, in wrapped_func
    return method(locked_class, **inputs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\comfy_api\latest\_io.py", line 1231, in EXECUTE_NORMALIZED
    to_return = cls.execute(*args, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-tooling-nodes\nodes.py", line 179, in execute
    raise ValueError(f"Image with ID {id} not found in cache.")
ValueError: Image with ID 84e22cf5-fc57-4b62-8ad3-2846114e61a2 not found in cache.

Prompt executed in 0.02 seconds
!!! Exception during processing !!! Image with ID 04ae8cab-dabe-4b6c-b031-127aa912adfa not found in cache.
Traceback (most recent call last):
  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 496, in execute
    output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, hidden_inputs=hidden_inputs)
                                                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 315, in get_output_data
    return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, hidden_inputs=hidden_inputs)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 289, in _async_map_node_over_list
    await process_inputs(input_dict, i)
  File "D:\ComfyUI_windows_portable\ComfyUI\execution.py", line 277, in process_inputs
    result = f(**inputs)
             ^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\comfy_api\internal\__init__.py", line 149, in wrapped_func
    return method(locked_class, **inputs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\comfy_api\latest\_io.py", line 1231, in EXECUTE_NORMALIZED
    to_return = cls.execute(*args, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-tooling-nodes\nodes.py", line 179, in execute
    raise ValueError(f"Image with ID {id} not found in cache.")
ValueError: Image with ID 04ae8cab-dabe-4b6c-b031-127aa912adfa not found in cache.

Prompt executed in 0.01 seconds

[Acly]

Thanks for the details. I suspect it has to do with how long the jobs are running in total, images are removed from cache after 10 minutes. They should be uploaded when needed, not all at the start, probably something is going wrong there.

[Scarlet-Raine]

Can confirm same issue here since the last update or 2, refining 60% str batch 10 stops after 3. It's really annoying cause I queue up a bunch of refinements and pick the best one, but now I can't even leave it alone cause it crashes. Hoping for a fix soon

Server error: Image with ID 81926d2b-3916-45c0-8900-5728a3149acc not found in cache.