Best Practices

Maximize the quality of your results on CoinLab by following these proven practices.

Image Preparation

Original Image Quality

• Resolution: Recommended 1024px - 2048px

• Sharpness: Well-focused image and no blur

• Lighting: Even light, no extreme shadows

• Contrast: Good contrast between character and background

Ideal Composition

• Single Character: Only 1 character per image

• Natural pose: Standing, frontal or 3/4

• Simple background: Neutral, without distracting elements

• Framing: Full body or half body (AI generates full body aesthetically consistent)

Trait Writing

Effective Structure

• Specific: "girl in hat", "muscular man"

• Simple: Up to 2 characteristics at a time

• Clear: Avoid ambiguity

• Direct: Get to the point in 35 characters

Trait Categories

• Predominant Appearance: Dog, Cat, Man, Pig, Woman etc...

• Clothes: shirt, pants, shoes, accessories

• Accessories: glasses, hat etc...

final example: "woman with yellow shirt"

Style Selection

Understanding the Models

• Rich: Luxurious lifestyle

• Rocket: Go to the moon / Universe

• Beach: Relaxing and enjoying

Optimized Workflow

Sequential Process

1. Prepare the image with quality

2. Choose the appropriate style

3. Write a specific trait

4. Generate and evaluate the result

5. If it doesn't come out as expected try up to 3 times using different images and Traits

Iteration Strategy

• First generation: Main trait (e.g. "Lumberjack")

• Second generation: Secondary trait (e.g. "Bearded Woodcutter")

• Third generation: Style trait (e.g. "Bearded Lumberjack with Axe")

Credit Management

Usage Planning

• Budget: Set a Buzz limit per session for attempts

• Prioritization: Focus on more important variations

• Testing: Use quality images to avoid rework

Cost Optimization

• Quality first: Better image = better result

• Specific features: Avoid vague attempts

Monitoring

• Balance: Track available Buzz

• History: Review previous generations

Results Organization

Favorites System

• Mark the best creations

• Use for future reference

Efficient Search

• Mental tags: Remember the traits used

• Filters: Use prompt search

Preventive Troubleshooting

Common Issues

• Blurred image: Use a sharper image

• Inconsistent result: Be more specific in the trait

• Generation failure: Check connection and try again

Pre-Generation Checks

• Balance: Confirm sufficient Buzz

• Image: Check quality and format

• Trait: Review if it is specific and clear

• Connection: Confirm network stability