2025-11-22 12:22:54 +00:00
2025-11-22 12:14:05 +00:00
2025-11-22 12:14:05 +00:00
2025-11-22 12:14:05 +00:00
2025-11-22 12:22:54 +00:00
2025-11-22 12:14:05 +00:00
2025-11-22 12:14:05 +00:00
2025-11-22 12:14:05 +00:00

ComfyUI Text Line Selector Node

The Line Selector Node allows you to process multiline text one line at a time, with intelligent control over how the line index advances after each generation. This is particularly useful for:

  • Batch processing multiple prompts
  • Cycling through different parameters or settings
  • Processing lists of data entries sequentially or randomly
  • Creating dynamic workflows that iterate through text-based inputs

Inputs

Parameter Type Description
multiline_text STRING (multiline) The input text containing multiple lines
line_index INT Line number to extract (1 = first line, 2 = second line, etc.)
control_after_generate SELECTION How to modify the index after generation

Control After Generate Options:

  • fixed: Keep the same line index (no change)
  • increment: Move to the next line (wraps to first line after the last)
  • decrement: Move to the previous line (wraps to last line before the first)
  • randomize: Select a random line (avoids repeating the current line when possible)

Outputs

Output Type Description
selected_line STRING The extracted line of text
index INT The line index used

Installation

  1. Copy the entire folder to your ComfyUI custom nodes directory:

    ComfyUI/custom_nodes/line_selector_node/
    
  2. Restart ComfyUI to load the new node

  3. The node will appear in the utils category in ComfyUI's node browser as "Text Line Selector"

Important Usage Notes

The control_after_generate functionality is only triggered once per batch.

Due to the way ComfyUI caches node execution results within a batch, this means that all batch items will use the same line.

  • ✅ Correct: Set Batch Count to 1, click "Run" multiple times to progress through lines
  • ❌ Incorrect: Set Batch Count > 1 expecting different lines per batch item
S
Description
No description provided
Readme MIT
34 KiB
Languages
JavaScript 52.7%
Python 47.3%