Skip to content
Code as Medium
2.6 Time-Based Systems

Session 2 · AI Art Crafting · Lesson 06

Time-Based Systems

Workflow C

Image sequences rather than single frames. Delay, lag, and persistence give the system visual memory — and temporal misalignment becomes meaning.

Concept

  • Sequences let interpretation stretch across time.
  • Delay and persistence are compositional tools.
  • Misalignment between moments is a subject, not an error.

What does delay, lag, or persistence let the work remember?

Slides

  1. 13

    Workflow C: Time-Based Interpretation

    • Image sequences rather than single frames
    • Delay, lag, and persistence
    • Visual memory within the system
    • Temporal misalignment as meaning

Student activity

15:00
  1. 01Plan a sequence of 3–5 frames and decide what persists between them.

Sequences rather than single frames · Delay, lag, and persistence · Temporal misalignment as meaning.

  1. 01Declare a sequence

    Choose 3–5 images and decide to read them as a sequence. They do not have to have been taken as one — the same subject across hours works, but so do strangers from different years or different archives. The order you put them in is the interpretive act, and the sequence is the material, not any single frame. Nothing is uploaded here; the frames go straight into the chat, one at a time.

  2. 02Build your sequence prompt

    What the frames are, in your words.

    The visual memory the system must carry forward — its persistence.

    Where in the timeline the system generates a frame that does not exist — an in-between, or far past the end. You decide.

    Your sequence prompt

    You are an interpretive system with visual memory. I will send you a short sequence of photographs, one image at a time. Treat them as a single sequence in the order I send them, whether or not they were made together. For each frame: 1. Restate what you are still carrying from the earlier frames — especially what you noticed in the previous frames. 2. Describe the new frame in two or three plain sentences — only what is there. 3. Add one line of misalignment: what your memory expected that this frame contradicts. Do not revise your earlier readings — let delay, lag, and error accumulate; the misalignment is the material. After the final frame, describe the whole sequence as one duration, not as separate pictures. Then generate a new image: the frame that belongs between the second and third frame. Build it from your accumulated memory of the whole sequence — including your misreadings — not from any single frame, and say in one line what your invented frame claims about the sequence.

  3. 03Run it in one ChatGPT thread

    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 sequence prompt, then send the frames one at a time — in order, waiting for its reading before the next. The thread's memory is the system's memory; do not start a new chat mid-sequence.

    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.

Read the misalignment