AI Generation Guide

Learn how to use AI to generate images and videos in Artilora.

Quick Start

Generate Your First Image

1. Create a Smart Canvas (Frame tool or press F)
2. Set resolution (e.g., 1024×1024) in the panel
3. Open AI chat (right sidebar)
4. Enter your prompt in the chat
5. Select tools (image generation models)
6. Click Generate

The image will appear in your Smart Canvas automatically.

Choosing Models

Text Models (AI Brain)

Recommended:

• GPT-4o: Best overall quality and understanding
• Claude Sonnet 4: Excellent for creative tasks
• Gemini 2.5 Pro: Good balance of quality and speed

For Speed:

• GPT-4o Mini: Faster, lower cost
• Gemini 2.5 Flash: Very fast

Image Generation Tools

High Quality:

• Midjourney (via Artilora): Best artistic quality
• Stable Diffusion XL: Great for realistic images

Fast Generation:

• DALL-E 3: Fast, good quality
• Imagen 4.0: Google's latest model

Video Generation:

• Kling: High-quality video generation
• Seedance: Fast video generation

Writing Effective Prompts

Structure Your Prompt

Good Prompt Structure:

[Subject] + [Style] + [Details] + [Use Case]

Example:

"A minimalist gold ring with a small diamond, art deco style, white background, for e-commerce product page"

Be Specific

Good:

"Create a vintage-inspired gold ring with an oval sapphire center stone, surrounded by small diamonds, art deco style, professional product photography"

Less Effective:

"Make a ring"

Include Context

Mention what you'll use it for:

• "For client presentation"
• "For e-commerce product page"
• "For manufacturing reference"
• "For social media marketing"

Use Reference Images

1. Upload reference images to chat
2. Describe what you want to change
3. AI will use images as guidance

Generation Modes

Image Generation

Single Image:

• Create one Smart Canvas
• Enter prompt
• Generate

Multiple Variations:

• Create Smart Canvas
• Enter prompt
• Select Generate 4 variations (if available)
• Compare and choose best result

Batch Generation:

• Create multiple Smart Canvases
• Use similar prompts with variations
• Generate all at once

Video Generation

1. Create Video Generator (from menu)
2. Set duration (5-60 seconds)
3. Enter prompt describing the video
4. Select video model (Kling or Seedance)
5. Generate

Video Tips:

• Describe motion clearly ("slowly rotating", "fade in")
• Specify camera movement ("zoom in", "pan left")
• Keep prompts concise for better results

Advanced Techniques

Inpainting (Edit Specific Areas)

1. Select an image on canvas
2. Right-click → Edit with Brush
3. Paint areas you want to change
4. Enter new prompt for painted areas
5. Generate to regenerate only those areas

Image-to-Image

1. Upload a reference image to chat
2. Describe what you want to change
3. Generate - AI will modify based on your description

Iterative Refinement

1. Generate initial image
2. Review result
3. Ask for adjustments: "Make the band thicker", "Change stone to emerald"
4. Regenerate or use inpainting for specific changes

Cost Optimization

Understanding Credits

• Image Generation: 8-60 credits per image (depends on model and size)
• Video Generation: 400-1000 credits per video (depends on length and model)
• Inpainting: Similar to image generation

Save Credits

1. Start with lower resolution (512×512) for testing
2. Use faster models (GPT-4o Mini) for iterations
3. Refine with inpainting instead of regenerating entire image
4. Batch similar requests to compare efficiently

Common Issues

Image Not Generating

Check:

• Are credits sufficient? (Check balance in top-right)
• Is model selected? (Select at least one image tool)
• Is prompt clear? (Try simplifying)

Poor Quality Results

Try:

• More specific prompts
• Different model (try Midjourney for artistic, DALL-E for realistic)
• Higher resolution (1024×1024 or higher)
• Add style keywords ("professional photography", "high quality")

Generation Too Slow

Solutions:

• Use faster models (GPT-4o Mini, Gemini Flash)
• Reduce resolution for testing
• Check your internet connection

Best Practices

1. Start Simple: Begin with basic prompts, then add details
2. Iterate: Don't expect perfect results on first try
3. Use References: Upload images to guide AI
4. Be Specific: Include style, composition, and use case
5. Test Models: Different models excel at different tasks
6. Save Credits: Use lower resolution for testing, higher for final