Quiz połączony z przewodnikiem · Średni Poziom

Quiz dotyczący renderowania tekstu w obrazach generowanych przez sztuczną inteligencję

Testuje, dlaczego tokeny podsłów CLIP ukrywały pisownię, jak kodery T5 i kodery na poziomie bajtów poprawiły pisownię oraz jak pomocne jest szkolenie i malowanie uwzględniające glify.

Powiązane ścieżki prowadząceRenderowanie tekstu w obrazach generowanych przez AI
Pytanie 1 z 8

Why did CLIP-based image models struggle to spell words correctly?

Podgląd każdego pytania8 pytania

Odpowiedzi i wyjaśnienia pojawiają się w miarę rozwiązywania quizu.

  1. Why did CLIP-based image models struggle to spell words correctly?
    • CLIP had no vocabulary at all
    • CLIP turns prompts into subword tokens that capture meaning, not letter sequences
    • CLIP processes only numbers
    • CLIP was trained only on images that contained no text
  2. Which kind of encoder did Google's character-aware study find spells better?
    • A smaller CLIP encoder
    • An audio encoder
    • An image-only encoder
    • A byte-level encoder such as ByT5 that sees individual characters
  3. Why is a small error tolerable in a generated face but not in generated text?
    • A single wrong stroke can change which letter it is
    • Faces are generated by a separate model
    • Text is always drawn at a higher resolution
    • Faces use a different kind of noise
  4. What limitation in training data made it hard to link words to letter shapes?
    • Training images used too many fonts
    • All training images were black and white
    • Captions rarely wrote out the text visible in the image
    • Only English-language images were included
  5. What did research systems such as TextDiffuser and GlyphControl add to image generation?
    • Invisible watermarks
    • Explicit layouts or pre-rendered letter images to guide the model
    • Audio narration
    • Faster sampling and nothing else
  6. Which model does the guide say paired its image generator with a T5 encoder?
    • Stable Diffusion 3
    • Stabilna dyfuzja 1.5
    • Stable Diffusion 1.4
    • The original CLIP model
  7. For production work that needs exact text, what does the guide call the most reliable approach?
    • Writing a much longer prompt
    • Obniżenie rozdzielczości
    • Adding a heavy negative prompt
    • Adding text outside the model, or inpainting a masked area with the corrected word
  8. According to the guide, how do modern image models produce letters?
    • By copying shapes from a font file
    • By running OCR on the prompt
    • By generating them as pixels, which is why long passages and small print still go wrong
    • By looking up images of words on the web