An upgraded version of my original Qwen Cockpit workflow that adds several features and optimizations. Same philosophy as the first version in that all the complexity of Comfyui is removed and all that's left is a clean, easy to read, and completely modular workflow. All loaders have moved to the backend including the Lora Loader. Just collapse the backend to access them.

Pipelines included:

1. Text2Image

2. Image2Image

3. Qwen Edit

4. Inpaint

5. Outpaint

-ControlNet

All of these are controlled with the "Mode" node at the top left. Just switch to your desired workflow and the whole workflow accommodates. The ControlNet runs parallel to all modes so it can be enabled in any pipeline.

Features Included:

- Refining

- Upscaling

- Resizing

- Image Stitch

Features work as they did before, Just enable whichever one you need and it will be applied. Image stitch is new and only works in mode 3 Qwen Edit as it allows you to add an object or person to an existing image.

I've tested everything on my 8 gb VRAM 3070 and every feature works as intended. Base generation times take about 20-25 seconds with the lightning 4 step lora which is currently the default of the workflow.

If you run into any issues or bugs let me know and I'll try to sort them out. Thanks and I hope you enjoy the workflow.