Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Imagen: Breaking LLM Intuition (By: Haleema Tal...

Imagen: Breaking LLM Intuition (By: Haleema Tallat) - DevFest Lahore 2025

Talk by Haleema Tallat (https://www.linkedin.com/in/haleema-tallat/) at DevFest Lahore 2025 by GDG Lahore.

GDG Lahore PRO

December 20, 2025

Transcript

38 slides
  1. LLMs trained us to believe: ◦ Retries fix errors and

    improve outputs ◦ Prompts specify precise outcomes ◦ Evaluation is cheap and straightforward ◦ Determinism is achievable with effort
  2. Begin with xT, completely random pixels Gradually refine towards x₀

    through learned steps All pixels update. Every denoising step modifies the entire image. Imagen generates images by reversing accumulated noise through iterative refinement:
  3. • No privileged step: Each iteration has equal importance •

    No "finishing" concept: Objects don't get completed sequentially • No symbolic structure: Imagen never knows it's drawing a hand
  4. Why Imagen Is Inherently High-Variance Imagen samples from a wide,

    multimodal distribution where language dramatically under-constraints visual outcomes.
  5. In LLMs: • Temperature controls randomness • Retries with lower

    temperature converge • Errors can be corrected • Quality improves with iterations
  6. In Imagen: • Retries don't correct errors • They resample

    the distribution • Each attempt is independent • No convergence guarantee
  7. • Spatial constraints saturate early ◦ Adding "on the left"

    or "in the centre" provides diminishing control • Detail shifts style, not structure ◦ More descriptive prompts change texture and mood, not fundamental composition • Improvements hit limits fast ◦ Prompt engineering reaches a ceiling far earlier than with LLMs
  8. Matter at Hand • Imagen 4 generates more convincing hands

    with better anatomical appearance • Finger count remains probabilistic, not enforced • No hand-level structural invariants exist in the model
  9. Mirrors, Text, Symmetry • Imagen does not model: ◦ object

    identity across space ◦ bidirectional consistency ◦ symbols as symbols • It models local pixel correlations • Realism collapses under global constraints.
  10. Realism • learns how images usually look • captures texture,

    lighting, style • makes unlikely images less likely • produces convincing surface coherence • “Does this look like something I’ve seen before?” Understanding • no enforced rules or invariants • no object identity across space • no constraint satisfaction • “Is this necessarily correct?”
  11. What Improves Every Generation • texture fidelity • lighting &

    materials • style imitation • photorealism -> surface plausibility • instruction following • compositional reasoning • long-range coherence • constraint satisfaction -> behavioral reliability Imagen LLMs
  12. Imagen in a Startup Product: Viggle • uses Imagen 3

    to generate virtual characters from text • users describe characters • Imagen creates visually plausible characters for video • motion, timing, and storytelling are handled by the product
  13. Imagen in a Startup Product: Cartwheel • uses Imagen 3

    for text-to-character visual generation • creators describe characters in natural language • Imagen generates character visuals directly in the product • characters are then animated and exported fully rigged • Imagen supports creative ideation, not animation logic
  14. Lahore LLMs trained us to expect generative models to behave

    like software. But Imagen shows us: • outputs are sampled, not computed • realism can emerge without understanding • improvement does not imply new capabilities

More Decks by GDG Lahore

Other Decks in Programming