add basically all inputs to one node, venice parameters etc, use dynamic combo for hiding unsupported capabilities
fix encode tensor for vision
add text model spec get to utils/catalog
it does work technically but 1024x1024 only, it requires different params, with dynamic combo widgets can be changed accordingly but thats work for future
adjust prompt field tooltip
add more optional widgets for dynamiccombo, durations, aspect ratios, resolutions, audio, and change first or default func to constraint_values
also move some code
not finished at all but will generate video now, and outputs newer "VIDEO" datatype output
ProgressBar should show more correct based on polling response
adjusted polling interval to be a bit slower
add (temporary) ability to bypass api api call for video and load one from testing_video folder instead for testing video output usage downstream, if any
the debug env var will trigger any api generated video to be saved to the same testing_video folder instaed of just being saved as temp file
my shitty description:
refactor a bunch of stuff
better model loading, removed the stupid js workaround to populating the COMBO fields
removed model loading from js and model fetching from routes/pyserver
unified model/character/styles lists internally as "catalog"
changed textbox in settings to by type: password
replaced primitive requests with client session for better/cleaner/modular request managing, added some better error logging and responses
added dry run env var VENICE_CLIENT_DRY_RUN can be set to 'true' or '1' (it just returns dummy response and doesnt hit the veniceai api)
make __init__.py module/node loading better by loading any .py file with NODE_*_MAPPING in nodes folder instead of hardcoding names
remove old redundant files
AI summarized description with some editing:
- Added `venice_client.py` for handling API requests to VeniceAI.
- Introduced `venice_config.py` for managing API key configuration.
- Created `venice_catalog.py` to manage and cache models, styles, and characters.
- Removed outdated key management logic from `pyserver/get_key_from_jssetting.py`.
- Consolidated character and model update logic into the new catalog management system.
- Move utility functions to `nodes/utils.py` containing validation stuff mostly.
- Removed legacy update scripts for characters, models, and styles.
- Added error handling and logging for API interactions.
- remove obsolete workflow JSON file.
draconicdragon