Two versions of a prompt with one switch

<!-- lesson: switch -->

Lesson "switch", in the group Packs and tools of the lesson list. Steps: 9.

#What you will be able to do

You keep two prompts in one graph and choose between them with one yes/no value, the way ComfyUI's templates do.

#What you need

• It goes on from A first picture from text. The card says so; it is a hint, not a lock.
• Nodes: CLIP Text Encode (Prompt), KSampler, Text (Multiline), If/Else Switch, Boolean.
• The app downloads and installs nothing. The step texts below say which files and folders, and which ComfyUI template names them.

#The steps

The words of each step, as the lesson shows them. Node names are in bold.

1. Compare without rewiring (reading)

   To compare two prompts you would retype the text or move links each time. ComfyUI's templates do it with a switch: both versions stay in the graph, and one yes/no value chooses which one goes on.

   This lesson builds that on the graph of "A first picture from text". The same switch later chooses between your prompt and one a language model rewrote.

2. The picture graph is needed

   The canvas must have the graph of "A first picture from text": the CLIP Text Encode (Prompt) node whose output goes into the positive input of KSampler.

   If it is not there, do that lesson first.

3. The first version

   Add a Text (Multiline) node and write a prompt into its value field. For example: a lighthouse on a cliff at sunset, warm light, calm sea.

   This node holds the text in one visible place. It goes on to the switch, not straight into CLIP Text Encode (Prompt).

4. The second version

   Add a second Text (Multiline) node and write another version into its value field: the same subject in another light, another style or with one more detail.

   Change one thing only. Then you see what that one change did.

5. The switch

   Add a If/Else Switch node. Connect the output of Text (Multiline) to its on_false input and the output of Text (Multiline) to its on_true input.

   The switch passes on one of the two. Only the chosen side is worked out.

6. Into the prompt

   Connect the output of If/Else Switch to the text input of CLIP Text Encode (Prompt).

   The text field of CLIP Text Encode (Prompt) now takes its words from the link, and the text typed in it is no longer used. The rest of the graph stays as it was.

7. The yes/no value

   Add a Boolean node, connect its output to the switch input of If/Else Switch and set its value to true.

   true chooses on_true, the second version; false chooses on_false, the first. The choice now sits in one small node you can find at once.

8. How to compare (reading)

   In ComfyUI set control_after_generate under the seed of KSampler to fixed. Run once with true and once with false.

   With the same seed the two pictures differ only by the prompt. If they also differed by the noise, you could not tell which change did what.

9. The graph is ready (reading)

   Save the graph with "File ▸ Save workflow for ComfyUI…", open the file in ComfyUI and run it there. This app runs nothing in ComfyUI itself.

   The same switch works for anything of one type: two models, two LoRA settings, two sizes. Both sides must carry the same type.

   Next: "A language model writes the prompt" puts a language model on the on_true side.

#What the lesson checks

A reading step is done when you press Got it, next. A doing step ticks itself off when the canvas holds:

• Step 2, The picture graph is needed:
  • a CLIP Text Encode (Prompt) node.
  • a KSampler node.
  • CONDITIONING of CLIP Text Encode (Prompt) into positive of KSampler.
• Step 3, The first version:
  • a Text (Multiline) node.
  • value of Text (Multiline) (first version) holds at least 10 characters.
• Step 4, The second version:
  • a Text (Multiline) node.
  • value of Text (Multiline) (second version) is not the same file as value of Text (Multiline) (first version).
• Step 5, The switch:
  • a If/Else Switch node.
  • STRING of Text (Multiline) (first version) into on_false of If/Else Switch.
  • STRING of Text (Multiline) (second version) into on_true of If/Else Switch.
• Step 6, Into the prompt:
  • output of If/Else Switch into text of CLIP Text Encode (Prompt).
• Step 7, The yes/no value:
  • a Boolean node.
  • BOOLEAN of Boolean into switch of If/Else Switch.
  • value of Boolean is true.

#What next

These lessons go on from this one: A language model writes the prompt.

#See also

• Notes, reroutes and primitives
• The prompt: how models read text
• How lessons work