Local Workflow Configuration
A local workflow runs in your local ComfyUI environment. Before using this type of workflow, make sure ComfyUI can start normally and place the required model files in the directories specified by the workflow documentation. After configuration, you can enter prompts, adjust parameters, and run the generation task directly from the node property panel.
1. Activate the Node Property Panel
Click an image, video, text, or audio node on the canvas to activate its property panel. The panel appears below the node and lets you choose a workflow, enter prompts, adjust parameters, and run the task.
In the workflow picker at the lower-left area of the property panel, select a workflow under a local AI model category. Local workflows usually use the standard dark background. If a model category has a green background, it is an API model category and requires the corresponding API setup; it is not covered by this local workflow guide.
2. Open the Workflow Introduction
After selecting a workflow, click the question-mark button in the upper-right corner of the node property panel. DFCine will open the workflow introduction page in your browser.
On that page, focus on the model configuration section. Different workflows may require different checkpoints, LoRA files, VAE files, audio models, or auxiliary models. The introduction page lists the required model files and tells you which ComfyUI model directory each file should be placed in.
Follow the instructions to place the model files. Common directories include:
ComfyUI/models/checkpointsComfyUI/models/lorasComfyUI/models/vaeComfyUI/models/clipComfyUI/models/unet- Any additional directory specified by the workflow documentation
After placing the models, confirm that ComfyUI can detect them. Some models require restarting ComfyUI before they are loaded.
3. Set Prompts and Parameters
Return to the DFCine node property panel and enter the prompt for the current task. Prompt requirements vary by workflow, so follow the recommended wording on the workflow introduction page whenever possible.
Adjust generation parameters as needed, such as resolution, duration, seed, generation mode, or advanced settings. If you are unsure about a parameter, start with the default values and run a small test first. After the workflow runs correctly, continue tuning the settings.
The upper-right area of the property panel also includes a preset button. Some workflows provide preset prompts and parameters. You can choose one of these presets to let DFCine fill the prompt and property values automatically, then click Run. Presets are useful for testing a workflow for the first time and quickly confirming that the models, parameters, and execution chain are working.
Before running, confirm the following:
- ComfyUI is open and connectable.
- The models required by the selected workflow are in the correct directories.
- The node has the required upstream inputs connected, such as an image input for an image-to-video workflow.
- Your local VRAM, disk space, and output directory are sufficient for the workflow.
If ComfyUI is not running, the workflow cannot execute, or the model setup is incomplete, DFCine will show a pop-up or display the error message on the node.
4. Run the Local Workflow
After confirming the setup, click the run button in the lower-right corner of the node property panel. When execution starts, the current node enters the queued or running state, its border turns green, and status text is shown inside the node.
During execution, you can also watch the ComfyUI console to understand model loading, node execution, and output saving progress. If the task takes a long time, wait for ComfyUI to finish the current queue and avoid repeatedly clicking Run.
After the workflow succeeds, the generated result is returned automatically to the node that was executed. Image, video, audio, and text results are displayed inside the node in their matching formats, and they can be connected as outputs to downstream nodes.
5. Manage Historical Results
The same node can be executed multiple times. After a node has produced multiple results, right-click the node and open Generation History to view and manage earlier results.
In Generation History, you can set an older result as the current display result. The result currently displayed on the node also becomes the node's current output and can be passed to downstream nodes.
If you need to save a result elsewhere, use Save As from the node context menu, or add the result to the Asset Library for reuse in later projects.
6. Watch Video Demos
For the full workflow installation, model placement, and execution process, refer to the operation demos published by DIGITFOLD on its social media channels. These videos usually show the exact UI path, model file locations, and real execution results, making them useful references when configuring a local workflow for the first time.