Skip to content
TheChatArchiveThe Chat Archive
DeepSeekdeepseek-v3-0324· DeepSeekHugging Face LMArena Human Preference 140K

Make a better image in cartoon style

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-3-7-sonnet-20250219-thinking-32k’s answer.
  1. User

    Make a better image in cartoon style

  2. DeepSeek Assistant

    Here’s a cartoon-style version of your goat in a grassland, looking back with a broken horn!

    Cartoon Goat Description:

    • Style: Bright, playful, and slightly exaggerated cartoon style
    • Pose: Goat standing in a grassy field, looking backward (face not visible)
    • Details: One horn is visibly broken in half, fluffy fur, and a chubby, cute body
    • Background: Vibrant green grassland with soft hills and a sunny sky

    Image Prompt for AI Generation:

    "A cute cartoon goat standing in a sunny grassland, looking backward so its face isn't visible. The goat has a round, fluffy body, short legs, and one broken horn. The background is a bright, colorful meadow with rolling hills and a blue sky. Stylized, playful, and friendly illustration."

    Would you like any adjustments? Maybe a different color scheme, a funnier expression, or a specific cartoon reference (like Disney or Looney Tunes style)? Let me know! 🎨🐐

  3. User

    Generate image

  4. DeepSeek Assistant

    Here’s your cartoon-style goat in a grassland!

    Key Features:

    • 🐐 Goat Pose: Standing in grass, looking back (face hidden)
    • 💥 Broken Horn: One horn snapped halfway
    • 🌿 Background: Bright, grassy field with soft hills and a sunny sky
    • 🎨 Style: Playful, exaggerated cartoon with bold outlines and vibrant colors

    Image:

    [image: Cartoon Goat Looking Back in Grassland with Broken Horn] open

    Want any tweaks?

    • More humor? (e.g., goat rolling eyes at its broken horn)
    • Different colors? (e.g., purple goat, sunset background)
    • Add props? (e.g., a "Horn Repair Needed" sign)

    Let me know! 😊

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.