Skip to content

Chains

Chaining tools

Stack the steps in the order you want them to run, tune each one, and press Run once.

Drop tools into a chain frame in the order you want them to run. The first step takes your input, and every step after it takes what the one above produced.

  1. Start from the sequence you already ran by hand. The settings you tuned on those cards are the settings the chain wants
  2. Add the steps in order. Position is execution order, so the top row runs first
  3. Set each step’s model and settings. Every step keeps its own model pill and its own gear, exactly as the standalone card would
  4. Check the summed price on Run before pressing it
  5. Press Run once

The result lands on the canvas the way any single generation does, as one asset from one press.

While it runs, the panel under the top bar carries the chain’s name and the step it is on, so a four-step chain reads step 2 of 4 rather than sitting on one number for several minutes. Clicking that row takes the canvas to where the result will land.

Only the first step takes anything from you. Every step after it is fed by the one above, so the top of the frame is the only place anything goes in.

When the first step is a tool that takes a picture or a clip, the frame shows a drop tile above the steps. Drag an image or a video reference onto it from the canvas, or click it to pick a file. The line beside the tile is the first step’s own request, so it reads “Drop an image” or “Drop a video” depending on what that tool wants, and it stops asking once the tile is filled. A first step that wants a prompt rather than media is set with its pencil, which opens that tool’s card.

A chain does not have to be built from nothing. Right-click a generated asset and Plnty reads back the tools that made it, offering that sequence as a chain you can point at something else. It lands unsaved and unrun, so nothing is spent until you press Run.

What comes back is what was recorded at the time: the tools, their order, their models, their settings, the prompt, and the LoRA a step ran on. Where something was never recorded, the chain names the step that is missing it rather than quietly starting that step from its defaults.

The order that works is almost always cheap-and-structural first, expensive-and-final last.

Put early Put late
Generation, which decides what the thing is Upscaling, which multiplies the cost of everything before it
Background removal, which changes the composition Grading and finishing, which only makes sense on the final frame
Remeshing, which changes the topology everything else sits on Texture regeneration at 4K or 8K

Upscaling early is the classic expensive mistake. Every step after it works on a bigger image, which usually costs more and always takes longer, and you have paid to enlarge something you were about to change anyway.

Order is set when you build a chain and is not fixed there. Drag a step by the number on its left: the number becomes a stack handle when you point at it, the row lifts as you carry it, and the list shows where the step will land while you are still moving.

Changing the order changes what feeds what, so any results the chain had saved from the moved position onward are cleared. A chain part way through a sequence runs those steps again rather than replaying results that no longer match the order they now sit in.

Each row carries the same controls its card would: the model pill, and the settings behind the gear. Anything you change shows in the grey line under the step’s name, so you can see what a row is set to without opening it. Swapping a model inside a chain changes that step’s price, and therefore the total.

A chain puts its last step on the canvas and leaves the middle ones off it. They are not discarded. Every step is saved to your asset library like any other generation, so the middle of a run stays searchable even when it never appeared on the board.

Turn on Keep steps and each step’s result lands on the canvas as well, in a row beside the chain, in the order they ran.

It is off by default and only offered once a chain has more than one step, because the usual case is that the middle is scaffolding. Turn it on when you are still deciding whether the order is right, since seeing what step two handed step three is the fastest way to find the step that is letting the chain down.

A chain runs once per press and reopens when it is done, which is what you want while you are still working the sequence out. When the sequence is settled and you want several passes of it, lock the chain instead.

The padlock in the chain’s header keeps the frame open through a run and gathers every result into one frame beside it, instead of leaving them loose on the board. The number next to Run says how many times a single press fires. Raising it above 1 locks the chain for you, because a batch is a locked session rather than a separate mode, and a locked chain keeps its Run button reading which run and which step it is on.

Runs go one after another rather than all at once. A chain is several paid steps, and firing five four-step chains simultaneously would reach the provider’s in-flight limit partway through with credits already reserved on the runs behind it.

A batch stops at the first run that fails instead of firing the rest into the same wall. Whatever landed before it stays, and pressing Run again starts a fresh batch.

The chain runs as one operation, so a failure part way through stops the sequence rather than delivering something half-processed. The frame names the step that stopped it.

Every step that finished before it is kept and pinned, so running the chain again pays only for what did not happen. The furthest of them also lands on the canvas, which gives a stopped run something to carry on from instead of leaving you to go looking for it.

The failure reports the same way any run does, with a code and a tier telling you whether it was your input, a transient provider problem, or something on our side. See when a run fails.

Refreshing the page or closing the tab mid-run is not the same as a failure. The steps themselves run on our servers and carry on finishing there. What stops is the part in your browser that starts the next one.

Reopen the board and the chain picks itself back up. Before it continues it checks what became of the step that was in flight when you left, because that is the one it has no record of:

What it finds What happens
The step finished while you were away Its result is kept, and the chain continues from the next step
The step is still running The chain waits for it rather than starting it again
The step failed The chain stops there and names it

You are not charged twice for a step that already finished. That is what the check is for: a chain restarting from the last step it saw finish would re-run one the provider had already completed and billed for.

Only the browser that started the run picks it up. A collaborator opening the same board is not made to finish somebody else’s work, and neither are you on a different machine. A run abandoned more than about six hours ago is left where it is, on the grounds that it is not one you are still waiting on.

A chain you have got right can be saved instead of rebuilt. Reusing a chain covers how.