Workflow Inputs are nodes that allow you to feed three types of content - prompts, images, and videos - into the other nodes in your graph. Similar to when you normally use a prompt or reference image in our inference form, a Workflow Input allows you to modify what goes into your workflow so you can vary what comes out of it.
A workflow input can either be static (unable to be changed at runtime) or dynamic (able to be changed at runtime). When you're building a workflow, ask yourself whether you will want the image, video, or prompt to be changed when the workflow is run by another person. If you don't want it to be changed, make it static; If you do want it to be changed, make it dynamic. You can do that by clicking on this icon here:
This button allows you to convert the input into a dynamic input, allowing the person running the workflow to replace it with whatever they would like. Instead of a mermaid, they could add in a pegasus or a knight, or both.
The image below shows what the inputs look like when they are converted into dynamic inputs; Their images or text is collapsed to indicate they will be changed or set when the workflow is run.
Workflow Inputs: Not Just One but Many
One of the most powerful features of Workflows is the ability to add as many inputs as you would like. For example, once you have a workflow set up for changing the pose of a character, you can add as many characters as you'd like to change the pose in the same way:
When you add additional inputs, you will typically increase the number of generations you'll receive in the workflow. For example, if your workflow has a 2D Generator and you add in 3 prompts feeding into it, it will activate that generator 3 times. If that generator creates 4 images, then you'll create 3x4 = 12 images with those 3 prompts.
Workflow inputs aren't just used with generators. For example, a Reference Image node may feed an image into a 2D Generator but can also provide an image for a Composition Node. A Reference Video node can be used within a Stitch Video node as well as a Kling O1 base video input. We highly recommend you check out our example workflows for tips and tricks on how to use these nodes in your own workflows!
Workflow Inputs: Default State
When you create an input node and connect it, you will need to provide a default value or file within the node. For prompts this is text, and it is an image or video in their respective nodes. You cannot leave an input node empty on the workflow, or else the workflow will be put in an invalid state.
When a user runs the workflow, they will be shown the input nodes in their default state, which is the state they were last placed in while you were creating the workflow. When the workflow is run, the inputs are shown in the Run Workflow UI so the user may know what they need to input in their flow. When they prepare their workflow, they will see the default values and may replace them or add onto them for their run.
It is common for more complex workflows to use a combination of static workflow inputs as well as dynamic inputs to reduce the number of inputs a user has to assess before running the workflow.
