Complete guide to all features in DisComfy v2.0.

Generate stunning AI images using advanced ComfyUI workflows directly from Discord.

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

Basic:

/generate prompt:a majestic dragon soaring through clouds

With Parameters:

/generate prompt:cyberpunk cityscape at night width:1024 height:768 steps:30 cfg:7.5 batch_size:2

With LoRA:

/generate prompt:anime character in magical forest lora:anime_style_v2 lora_strength:0.8

When you run /generate, you'll see an interactive setup view:

1. Select Model - Choose between Flux, Flux Krea, or HiDream
2. Select LoRA - Pick from available LoRAs (organized by model)
3. Adjust Settings - Fine-tune parameters via modal
4. Generate - Click to start generation

DisComfy supports two AI editing models with different strengths.

Use Case: High-quality detailed editing
Speed: 1-3 minutes
Command: /editflux

Features:

• High-quality results
• Detailed edits
• Better for complex changes
• 10-50 sampling steps

Usage:

/editflux image:<upload> prompt:add sunglasses and a hat steps:25

Use Case: Ultra-fast editing and iteration
Speed: 30-60 seconds
Command: /editqwen

Features:

• Lightning-fast results
• Good quality
• Perfect for rapid iteration
• 4-20 sampling steps

Usage:

/editqwen image:<upload> prompt:change the background to mountains steps:8

Every generated image has two edit buttons:

• ✏️ Flux Edit - High-quality editing
• ⚡ Qwen Edit - Ultra-fast editing

Anyone can click these buttons to edit any image!

Upscale images using AI super-resolution for incredible detail enhancement.

• 2x - Quick upscale (512→1024)
• 4x - Standard upscale (512→2048)
• 8x - Maximum upscale (512→4096)

Via Button:

1. Generate or upload an image
2. Click 🔍 Upscale button
3. Select parameters in modal
4. Confirm to start upscaling

Automatic Features:

• Original prompt automatically preserved
• Smart parameter defaults
• Real-time progress tracking

Convert images to smooth, high-quality MP4 animations.

• 81 frames - ~2 seconds (quick animation)
• 121 frames - ~3 seconds (standard)
• 161 frames - ~5 seconds (longer animation)

Via Button:

1. Generate or upload an image
2. Click 🎬 Animate button
3. Customize animation settings
4. Confirm to start video generation

Custom Animation Prompts: Modify how your image animates with natural language:

• "slowly pan across the scene"
• "camera zooms in dramatically"
• "gentle rotation and zoom"
• "dynamic movement with parallax"

• Original prompt pre-filled (editable)
• Extended 15-minute timeout
• Real-time progress tracking
• High-quality MP4 output

DisComfy provides detailed, accurate progress information using WebSocket integration with ComfyUI.

🎬 Generating Video
📊 87.5% ████████████████████░░░
🔄 Sampling (315/321)
⏱️ Elapsed: 4m 32s | ETA: 42s
🎯 Node: WanVaceToVideo
Settings: 720x720 | 161 frames | Strength: 0.7

• Step-Based Accuracy - Progress calculated from actual sampling steps
• WebSocket Integration - Real-time updates directly from ComfyUI
• Queue Monitoring - Live updates when waiting in queue
• Node Tracking - See which ComfyUI node is executing
• Time Estimates - Elapsed time and ETA
• Multi-Phase Support - Handles complex workflows
• Automatic Fallback - HTTP polling when WebSocket unavailable

When ComfyUI is busy:

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

Key Feature: Anyone can use action buttons on ANY generation!

Available buttons on every generated image:

• 🔍 Upscale - AI upscaling
• ✏️ Flux Edit - High-quality editing
• ⚡ Qwen Edit - Ultra-fast editing
• 🎬 Animate - Video generation
• 🔄 Variations - Generate variations (if available)

Interactive Discord modals for parameter customization:

Upscale Modal:

• Upscale ratio dropdown
• Denoise strength input
• Sampling steps input
• Pre-filled defaults

Video Modal:

• Frame count selection
• Animation strength
• Quality settings
• Animation prompt editing

Edit Modal:

• Edit prompt input
• Sampling steps
• Model-specific ranges
• Validation

• Infinite Usage - Buttons never expire
• Pre-filled Defaults - Sensible starting values
• Input Validation - Prevents errors
• Smart Descriptions - Parameter help text
• Prompt Preservation - Original prompts auto-filled

Smart rate limiting prevents abuse while allowing normal usage:

• Per-User Limits - Configurable requests per minute
• Global Limits - Server-wide protection
• Sliding Window - Fair distribution
• Informative Messages - Clear rate limit feedback

Automatic cleanup and organization:

• Auto-Cleanup - Old files removed after limit
• Unique Filenames - Timestamp-based naming
• Organized Storage - Structured output directory
• Size Management - Configurable file retention

Robust error recovery:

• Graceful Degradation - Fallback to HTTP polling
• User-Friendly Messages - Clear error descriptions
• Detailed Logging - Debug information
• Recovery Options - Automatic retries
• Validation - Input checking before processing

Multiple users can generate simultaneously:

• Session Management - Proper HTTP connection pooling
• WebSocket Multiplexing - Single persistent connection
• Queue Management - Fair processing order
• Resource Cleanup - Proper session disposal

Dynamic LoRA loading with flexible strength control.

• Auto-Discovery - Automatically finds LoRAs in ComfyUI
• Model Organization - LoRAs grouped by compatible model
• Strength Control - Adjust LoRA influence (0.0-2.0)
• Interactive Selection - Choose LoRAs from dropdown
• Multiple LoRAs - Support for workflow-specific LoRA combinations

List Available LoRAs:

/loras

Use in Generation:

/generate prompt:your prompt lora:lora_name lora_strength:0.8

Via Interactive Setup:

1. Run /generate
2. Select LoRA from dropdown
3. Adjust strength if needed
4. Generate

1. Be Specific - Detailed prompts yield better results
2. Use Appropriate Steps - 20-30 steps for most generations
3. Start Simple - Test with basic parameters first
4. Choose Right Tool - Qwen for speed, Flux for quality
5. Experiment with LoRAs - Find styles that work for you

1. Batch Wisely - Use batch_size for variations
2. Optimize Steps - More steps ≠ always better
3. Use Seed Control - Reproduce good results with same seed
4. Monitor Queue - Generate during off-peak times
5. Clean Outputs - Regularly clear old files

1. Generate First - Start with basic generation
2. Iterate with Qwen - Fast edits for refinement
3. Final Polish with Flux - High-quality final edits
4. Upscale Last - Upscale only final results
5. Animate Select Results - Videos take time, be selective

• ✅ Refactored command handlers
• ✅ Improved progress tracking accuracy
• ✅ Better error messages
• ✅ Enhanced validation
• ✅ Optimized performance
• ✅ Cleaner architecture
• ✅ 99% test coverage

See Changelog for complete release history.

🎨 Ready to explore all features? Check out Usage Examples for practical demonstrations!