Prompting Guide for Image Generation

1. Crafting Effective Prompts

Prompting with NoFilterGPT uses SDXL or Pony SDXL which involves providing clear, descriptive input to generate high-quality images. The model interprets your text to create visuals, so specificity and structure are key.

Tips for Clear Faces and Accurate Results

• Be Specific: Include details like "sharp facial features," "realistic skin texture," or "well-defined eyes" for clearer faces.
• Use Style Keywords: Add terms like "photorealistic," "4k resolution," or "cinematic lighting" to enhance quality.
• Order Matters: Place the most important elements (e.g., "a woman with long red hair") at the start, followed by secondary details (e.g., "wearing a blue dress, standing in a forest").
• Avoid Ambiguity: Vague terms like "nice" or "good" can confuse the model—opt for precise descriptors instead.

2. Negative Prompts

Negative prompts let you specify what you don't want in the image, such as "blurry faces," "distorted hands," or "extra limbs." Set them in Settings → Image Gen for extra control over your generations.

3. Using Weights to Emphasize Features

If you really want something featured prominently in your image—like big eyes or a specific detail—you can add weights to your prompt. Use double parentheses ((feature)) to boost its importance, or ((feature:weight)) with a number (e.g., 1.3) to fine-tune the emphasis. Higher weights (e.g., 1.5, 2.0) increase the effect based on your results.

For example, ((big eyes)) will make eyes more prominent, while ((big eyes:1.3)) gives a stronger push for really big eyes. Experiment with weights to get the balance you want!

4. Example Prompts

Here are some ideas to get you started with SDXL or Pony SDXL, including examples inspired by Pony Diffusion V6 XL. Copy them individually below:

a young man with ((big eyes)), sharp facial features and short black hair, wearing a leather jacket, standing in a neon-lit city at night, photorealistic, 4k resolution

a majestic white horse galloping through a golden field under a clear blue sky, highly detailed, ((vibrant colors:1.5))

a futuristic cyborg with intricate metallic designs, ((glowing blue eyes:1.3)), in a dark sci-fi lab, cinematic lighting

an elderly woman with ((soft wrinkles)), kind eyes, sitting on a porch with a sunset background, warm tones, realistic

score_9, score_8_up, score_7_up, source_anime, vibrant colors, volumetric lighting, kindred, lamb, furry, anthro, ((small breast:1.2)), hourglass body, mask, white see-through summer dress, long white hair

score_8, source_furry, glowing forest, mystical fox with ((silver fur:1.4)), ethereal light, fantasy art, ultra-detailed

score_9, score_8_up, source_cartoon, astronaut pony, floating in zero gravity, neon spaceship interior, ((vibrant colors)), dynamic pose

5. Fine-Tuning Your Results

Experiment with adjectives, settings, and styles to refine your output. For Pony SDXL, you can also emphasize character traits or fantasy elements (e.g., "elf ears," "glowing aura") for more creative results.

6. Troubleshooting Common Issues

If your images aren't turning out as expected, try these fixes:

• Blurred Faces: Add "high detail, sharp focus" to your prompt.
• Weird Proportions: Specify "anatomically correct" or "realistic proportions."
• Unexpected Elements: Use negative prompts, or keep your prompt concise and focused.

Looking for usage limits? See the Documentation Hub or API docs.