hushed salmon
I'm back after some years, I remember back then I could see what was happening in the background, like when I change a model to another and it takes some time to prepare the data, this new version I don't see anything until the image starts being generated, so I think the interface could show this loading bar in another color other than blue to show that data is being prepared
hushed salmon
amazing!
hushed salmon
If you turn that on in the Settings modal under User Interface, you get this:
still doesn't do it... for example, if I cancel a task while the model loads and start it again, it keeps loading the model, but nothing happens in the front end, only if I see the system resources I can see that the app is still working
cinder mural
I'm not sure if the model loading ever stops until it's completed.
Probably just doesn't report anything until the loading is completed.
hushed salmon
that's what I mean, when it starts loading starts an animation, disc spinning or something, and stop only when finished, indenpendent from current task
drifting bough
Yeah, that setting only shows you the progress of denoising steps, not model loading.
cinder mural
Yeah, that setting only shows you the progress of denoising steps, not model loa...
It **does **show the model loading info as well, just not exactly how much of the model is loaded.
drifting bough
But not the progress, the percentage done as it does with denoising.
quasi hinge
the loading of the model could be observed in the launcher/cli maybe we can expose this to the UI and make it visiable
drifting bough
There currently isn't any progress displayed in the terminal, either. There'd have to be an option to turn on with that module.
quasi hinge
the loading shard thing?
drifting bough
There aren't any shards shown. It's just text like this, one per line, waiting as the model loads.
quasi hinge
that is a dev install, i dont know if this is a dev install only thing. but the info is there
drifting bough
that is a dev install, i dont know if this is a dev install only thing. but the ...
Not seeing that on mine, which is 6.12 installed from the community installer.
azure widget
it depends upon the model (and the loading method for such)
for the ones that don't have it, they should load quite quickly anyhow (and did so in the example above), so i'm wondering how useful it'd be there
drifting bough
It didn't, actually. It took over 45 seconds, and that was a smaller one.
hushed salmon
I changed model and this is what I had:
[2026-04-15 13:35:30,174]::[InvokeAI]::INFO --> Emitted public queue item event queue_item_status_chang
ed to all subscribers in queue default
Fetching 17 files: 100%|███████████████████████████████████████████| 17/17 [00:00<00:00, 16976.94it/s]
Loading pipeline components...: 100%|███████████████████████████████████| 7/7 [00:24<00:00, 3.48s/it]
[2026-04-15 13:36:10,763]::[ModelManagerService]::INFO --> [MODEL CACHE] Loaded model '7288e5e5-46bb-40
5e-a29d-7f81c239fa71:text_encoder' (CLIPTextModel) onto cuda device in 5.20s. Total model size: 234.72M
B, VRAM: 234.72MB (100.0%)
[2026-04-15 13:36:10,768]::[ModelManagerService]::INFO --> [MODEL CACHE] Loaded model '7288e5e5-46bb-40
5e-a29d-7f81c239fa71:tokenizer' (CLIPTokenizer) onto cuda device in 0.00s. Total model size: 0.00MB, VR
AM: 0.00MB (0.0%)
E:\Ai\invokeai.venv\Lib\site-packages\compel\prompt_parser.py:598: PyparsingDeprecationWarning: 'delim
ited_list' deprecated - use 'DelimitedList'
options = pp.Dict(pp.Optional(pp.delimited_list(option)))
E:\Ai\invokeai.venv\Lib\site-packages\compel\prompt_parser.py:666: PyparsingDeprecationWarning: 'delim
ited_list' deprecated - use 'DelimitedList'
[2026-04-15 13:35:30,174]::[InvokeAI]::INFO --> Emitted public queue item event queue_item_status_chang
ed to all subscribers in queue default
Fetching 17 files: 100%|███████████████████████████████████████████| 17/17 [00:00<00:00, 16976.94it/s]
Loading pipeline components...: 100%|███████████████████████████████████| 7/7 [00:24<00:00, 3.48s/it]
[2026-04-15 13:36:10,763]::[ModelManagerService]::INFO --> [MODEL CACHE] Loaded model '7288e5e5-46bb-40
5e-a29d-7f81c239fa71:text_encoder' (CLIPTextModel) onto cuda device in 5.20s. Total model size: 234.72M
B, VRAM: 234.72MB (100.0%)
[2026-04-15 13:36:10,768]::[ModelManagerService]::INFO --> [MODEL CACHE] Loaded model '7288e5e5-46bb-40
5e-a29d-7f81c239fa71:tokenizer' (CLIPTokenizer) onto cuda device in 0.00s. Total model size: 0.00MB, VR
AM: 0.00MB (0.0%)
E:\Ai\invokeai.venv\Lib\site-packages\compel\prompt_parser.py:598: PyparsingDeprecationWarning: 'delim
ited_list' deprecated - use 'DelimitedList'
options = pp.Dict(pp.Optional(pp.delimited_list(option)))
E:\Ai\invokeai.venv\Lib\site-packages\compel\prompt_parser.py:666: PyparsingDeprecationWarning: 'delim
ited_list' deprecated - use 'DelimitedList'
then after it was loaded it showed
the loading bar was the same as the blue solid bar on the UI
