Skip to content
Code as Medium
2.7 External Influence

Session 2 · AI Art Crafting · Lesson 07

External Influence

Workflow D

Text fragments, gesture, sound, environmental conditions — real or simulated. Something enters the system that you don't control.

Concept

  • External inputs break the closed loop and admit the world.
  • Simulated influence (weather, movement, audience) counts in V1.
  • The uncontrolled input relocates some authorship outside you.

What enters your system that you don't control?

Slides

  1. 14

    Workflow D: External Influence

    • Text fragments
    • Gesture or movement
    • Sound or rhythm
    • Environmental conditions (real or simulated)

Student activity

15:00
  1. 01Choose one secondary input for your system and define how it modulates the rule.

Text, gesture, sound, or environment as interpretive signals — interpretation over causality.

  1. 01Choose an external signal

    Pick one thing that enters your system from outside — something you do not control. It reads the image; it does not cause it. In V1 the signal is declared and simulated, not wired to a live source.

  2. 02Build your influence prompt

    The specific text fragment entering your system.

    The image or images the signal will press on.

    Your influence prompt

    You are an interpretive system. An external signal is entering you: text fragment. Treat it as a lens for reading, not a cause — you interpret the relationship between the signal and the image, you do not explain one by the other. I will send the image I send you. 1. First, describe the image plainly in two or three sentences — only what is there. 2. Now re-read the same image as if text fragment were pressing on it. What do you attend to now that you ignored before? What mood, detail, or story surfaces? 3. Name one thing the signal made you invent — something not in the image that the influence pulled into your reading. 4. Then write a single image-generation prompt that renders this influenced reading — the image as text fragment made you see it, inventions included, not the plain image. Return only that prompt, labelled "IMAGE PROMPT:", ready to paste into an image generator. Do not resolve the tension between the plain reading and the influenced reading. The gap between them is the material — and the image prompt should carry that gap forward, not erase it.

  3. 03Read it in ChatGPT

    ChatGPT

    Multimodal chat with a free tier. The workshop's default interpreter — give it an image and it will describe, read, and re-read what it sees.

    Not started

    Open ChatGPT, paste your influence prompt, then send the image. It will end with an IMAGE PROMPT: — the influenced reading written as a fresh generation prompt. Copy that line for the next step.

    Open in new tab(opens ChatGPT in a new tab)

    External tool — it has its own privacy policy and may change or require an account.

  4. 04Generate the influenced image in Leonardo

    Paste the IMAGE PROMPT the reader wrote — unedited — into an image generator. The result is your image as the signal made it seen: the interpretation turned back into a picture. Do not correct it toward the original; the drift is the point.

    Leonardo

    Image generation with a free daily token allowance. The workshop's default generator for the image → text → image translation step.

    Not started

    Paste the IMAGE PROMPT into Leonardo exactly as written and generate. Bring the output back to your seed image and compare — the gap between them is what the external signal authored.

    Open in new tab(opens Leonardo in a new tab)

    External tool — it has its own privacy policy and may change or require an account.

Read the influence