Deadline Distributed Placeholder

This is the Deadline-branded alias of DistributedPlaceholder, and like the other aliases in this pack it's the same class with a name that fits Deadline workflows. No inputs, one image output that is literally a black 1×1 CPU tensor (torch.zeros(1, 1, 1, 3)), zero VRAM, zero configuration.

Its job: when the master's "master rendering enabled" setting is off, the master still has to execute a graph to coordinate a job, and that graph still wants an IMAGE to flow through nodes downstream of the sampler. Placeholder stands in for the entire generation chain on the master's copy of the workflow - black, instant, CPU-only - while the real rendering happens on Deadline workers and comes back through a collector. Without it, the master's pruned workflow fails on a missing image and the whole job dies before it starts.

Since it's an alias, everything in the DistributedPlaceholder article applies here, including the mental model: this is a structural shim for distributed execution, not a node you'd ever wire into a real output. Use the Deadline-prefixed name if you're building Deadline submission workflows and want the node names to read cleanly in the job UI; use the plain name otherwise. They're interchangeable in every way that matters - same behavior, same output type, same "when to reach for it" rule: only when the master is coordinating without rendering.