Complete guide to using DisComfy v2.0.

DisComfy is a powerful Discord bot that integrates ComfyUI for professional AI image and video generation directly in Discord.

1. Ensure bot is invited to your server
2. Type / in any channel to see commands
3. Run /help for quick overview
4. Try your first generation: /generate prompt:a beautiful sunset

1. Generate - Create base image
2. Refine - Edit with Qwen (fast) or Flux (quality)
3. Enhance - Upscale for higher resolution
4. Animate - Convert to video (optional)

Generate AI images with interactive setup.

Usage:

/generate prompt:your description here

Parameters:

• prompt - Description of what to generate (required, 1-500 characters)

Interactive Setup: After running the command, you'll see:

1. Model Selection - Choose between:

   • Flux - High-quality general purpose
   • Flux Krea - Enhanced Flux with improved quality
   • HiDream - Alternative model

2. LoRA Selection - Choose optional LoRA:

   • LoRAs organized by compatible model
   • Adjust strength with modal

3. Settings Button - Customize:

   • Width (256-2048, default: 1024)
   • Height (256-2048, default: 1024)
   • Steps (1-50, default: 30)
   • CFG Scale (1.0-20.0, default: 7.5)
   • Seed (-1 for random)
   • Batch Size (1-4, default: 1)

4. Generate Button - Start generation

Example:

/generate prompt:cyberpunk cityscape at night, neon lights, rain, detailed

High-quality image editing using Flux Kontext AI.

Usage:

/editflux image:<upload> prompt:edit description

Parameters:

• image - Image to edit (required, upload file)
• prompt - How to edit the image (required)
• steps - Sampling steps (optional, 10-50, default: 20)

Best For:

• Detailed, high-quality edits
• Complex changes
• Final polishing
• When you have time (1-3 minutes)

Examples:

/editflux image:photo.jpg prompt:add sunglasses and a hat
/editflux image:photo.jpg prompt:change background to mountains steps:30
/editflux image:photo.jpg prompt:make it nighttime instead of day

Ultra-fast image editing using Qwen 2.5 VL AI.

Usage:

/editqwen image:<upload> prompt:edit description

Parameters:

• image - Image to edit (required, upload file)
• prompt - How to edit the image (required)
• steps - Sampling steps (optional, 4-20, default: 8)

Best For:

• Rapid iteration
• Testing ideas
• Quick changes
• Speed priority (30-60 seconds)

Examples:

/editqwen image:photo.jpg prompt:change shirt color to blue steps:8
/editqwen image:photo.jpg prompt:add flowers in background steps:12
/editqwen image:photo.jpg prompt:make the character smile

Check bot and ComfyUI status.

Usage:

/status

Shows:

• Bot version
• Discord connection status
• ComfyUI connection status
• System information

Example Output:

✅ Bot Status
Version: v2.0.0
Discord: Connected
ComfyUI: Connected (http://localhost:8188)
Uptime: 2 hours, 15 minutes

Get command help and information.

Usage:

/help

Shows:

• Command reference
• Parameter information
• Quick tips
• Links to documentation

List available LoRA models.

Usage:

/loras

Shows:

• All available LoRAs
• Organized by model type
• How to use them in generations

Example Output:

📚 Available LoRAs

Flux LoRAs:
• anime_style_v2
• photorealistic_enhance
• art_deco

HiDream LoRAs:
• landscape_enhance
• portrait_v3

Every generated image has action buttons that anyone can use:

Upscale image using AI super-resolution.

Options:

• Upscale Ratio - Choose 2x, 4x, or 8x
• Denoise Strength - 0.1-1.0 (detail enhancement)
• Sampling Steps - 10-50 (quality control)

When to Use:

• Final images for high resolution
• Printing or large format
• Detail enhancement

Tips:

• 2x: Quick upscale for social media
• 4x: Standard printing quality
• 8x: Large format, maximum detail
• Denoise 0.5: Balanced
• Denoise 0.7-1.0: Strong enhancement

High-quality editing using Flux Kontext.

Opens Modal:

• Edit prompt input
• Sampling steps (10-50)

Best For:

• Final quality edits
• Complex changes
• Professional results

Example Prompts:

• "add dramatic lighting"
• "change hair color to blue"
• "add magical effects"

Ultra-fast editing using Qwen 2.5 VL.

Opens Modal:

• Edit prompt input
• Sampling steps (4-20)

Best For:

• Quick iterations
• Testing ideas
• Rapid changes

Example Prompts:

• "make background darker"
• "add smile"
• "change shirt color"

Convert image to video animation.

Options:

• Frame Count - 81, 121, or 161 frames
• Strength - 0.1-1.0 (animation intensity)
• Steps - Quality control
• Animation Prompt - Customize motion

Frame Counts:

• 81 frames: ~2 seconds (quick loop)
• 121 frames: ~3 seconds (standard)
• 161 frames: ~5 seconds (longer, dramatic)

Animation Prompt Examples:

• "slowly pan across the scene"
• "camera zooms in on subject"
• "gentle rotation"
• "dynamic movement with parallax"

Tips:

• Test with 81 frames first
• Simple movements work best
• Videos take 2-10 minutes
• Original prompt pre-filled (editable)

DisComfy provides real-time progress updates during generation:

🎨 Generating Image
📊 65.0% █████████████░░░░░░░
🔄 Sampling (195/300)
⏱️ Elapsed: 1m 23s | ETA: 48s
🎯 Node: KSampler
Settings: 1024x1024 | Steps: 30 | CFG: 7.5

When ComfyUI is busy:

⏳ Queued - Position #2 in queue
🔄 Other generations ahead: 1
⏱️ Elapsed: 15s

• Real-time Updates - Live percentage and ETA
• Step Tracking - Current step / total steps
• Node Execution - Which ComfyUI node is running
• Queue Position - Your place in line
• Time Estimates - Elapsed and remaining time
• Settings Display - Your generation parameters

Be Specific:

❌ "a dog"
✅ "a golden retriever puppy playing in a sunny garden, detailed fur, photorealistic"

Use Style Descriptors:

• "photorealistic"
• "anime style"
• "oil painting"
• "digital art"
• "8k, highly detailed"

Separate Concepts:

"subject, action, environment, lighting, style, quality"
"dragon, flying, mountains, sunset, fantasy art, detailed"

Steps:

• 15-20: Fast, good quality
• 25-30: Balanced (recommended)
• 35-50: Maximum quality (slower)

CFG Scale:

• 3-5: More creative, less adherence
• 6-8: Balanced (recommended)
• 9-15: Strict adherence (can oversaturate)

Batch Size:

• 1: Single image
• 2-4: Variations (different seeds)

Seed:

• -1: Random (default)
• Specific number: Reproducible results
• Same seed + params = similar result

Quick Iteration:

1. Generate base image
2. Edit with Qwen (steps: 8) - Test ideas quickly
3. Edit with Qwen again if needed
4. Final edit with Flux (steps: 25) - Polish
5. Upscale 4x - Final result

Quality Priority:

1. Generate with high steps (35-40)
2. Edit with Flux (steps: 30-40)
3. Upscale 8x with denoise 0.7
4. Animate if desired

When to Upscale:

• Only upscale final images
• After all edits complete
• For printing or high-res needs

Denoise Guide:

• 0.1-0.3: Minimal change
• 0.4-0.6: Balanced (recommended)
• 0.7-1.0: Strong enhancement

Best Practices:

• Start with 81 frames to test
• Use simple animation prompts
• Generate videos during off-peak
• Videos take 2-10 minutes
• Don't spam video generation

Animation Prompts:

• Keep it simple
• Describe camera movement
• Avoid scene changes
• Test before long renders

1. Generate base: /generate prompt:character concept, detailed
2. Refine features: Qwen edits for quick changes
3. Final details: Flux edit for quality
4. Upscale: 4x for portfolio

1. Generate: Wide aspect (1344x768)
2. Iterate: Qwen for composition changes
3. Polish: Flux for detail enhancement
4. Animate: Slow pan across scene

1. Generate: Simple, clear prompt
2. Iterate: Multiple Qwen edits
3. Final: Flux edit for perfection
4. Upscale: 4x for print

1. Generate: Multiple variations (batch_size:4)
2. Pick best: Choose favorite
3. Refine: Edit to add details
4. Upscale: 8x for presentation

• Configurable per-user limits
• Prevents spam and abuse
• Contact admin if blocked
• Check /status for limits

• Images: Up to 25MB upload
• Generated: Varies by resolution
• Videos: Can be large (50-200MB)
• Auto-cleanup after 100 files

• Images: ~30 seconds
• Edits (Qwen): 30-60 seconds
• Edits (Flux): 1-3 minutes
• Upscale: ~45 seconds
• Video: 2-10 minutes
• Queue times: Varies

• Must have required models
• Custom nodes needed for some workflows
• Enough VRAM for generation
• Stable network connection

Solution: Increase timeout in config or reduce complexity

Solution: Wait 1-2 minutes or restart bot

Solution: Wait for window to reset or contact admin

Solution: Check ComfyUI has required models

Solution: Normal if queued, otherwise check ComfyUI

See Troubleshooting for detailed solutions.

• Usage Examples - Practical examples and workflows
• Features - Detailed feature documentation
• Troubleshooting - Common issues and solutions
• Configuration Guide - Customize bot settings
• Custom Workflows - Create your own workflows

🎨 Ready to create amazing art? Start with /generate and explore!