<h1 id="ideogram-4-gguf-workflow-for-comfyui">Ideogram 4 GGUF Workflow for ComfyUI</h1>A simple workflow for running Ideogram 4 in GGUF quantized format on ComfyUI, optimized for systems with limited VRAM.Tested on RTX 3060 12GB with 16GB RAM (and 32GB Virtual RAM)⚠️ Don't forget to update your ComfyUI first!Available Versions:- v1.0 - Standard (2 UNET GGUF models) - Default traditional workflow- v2.0 - TurboTime (Single UNET GGUF) - ⚡ Faster, better quality, super low VRAM---<h2 id="what's-new-in-v2.0">🆕 What's New in v2.0</h2>- ⚡ Optimized workflow: Removed redundant nodes (Dual Model CFG Guider, CFG Override)- 🚀 TurboTime LoRA support: 2-step generation with CFG=0.0- 💾 Low VRAM mode: Works on 8GB+ GPUs with Q4_0 quantization- 🎯 Cleaner structure: Added BasicGuider for proper CFG handling- 📝 Faster generation: ~1-2 seconds per image on modern GPUs---<h2 id="required-models">📦 Required Models</h2>1. UNET Models → ComfyUI/models/unet/Download unet files from HuggingFace (Recommended):- Repository: <a target="_blank" rel="ugc" href="https://huggingface.co/molbal/ideogram-4-gguf">https://huggingface.co/molbal/ideogram-4-gguf</a>- Main Model (choose desired quantized version):- <code>ideogram4-transformer-q4_0.gguf</code> (5.64 GB) - ⭐ Best for low VRAM- <code>ideogram4-transformer-q4_1.gguf</code> (6.21 GB)- <code>ideogram4-transformer-q5_0.gguf</code> (6.77 GB)- <code>ideogram4-transformer-q5_1.gguf</code> (7.33 GB)- <code>ideogram4-transformer-q8_0.gguf</code> (10.1 GB) - Best quality- Unconditional Model (not needed in workflow v2.0):- <code>ideogram4-unconditional_transformer-q4_0.gguf</code> (5.64 GB)- <code>ideogram4-unconditional_transformer-q4_1.gguf</code> (6.21 GB)- <code>ideogram4-unconditional_transformer-q5_0.gguf</code> (6.77 GB)- <code>ideogram4-unconditional_transformer-q5_1.gguf</code> (7.33 GB)- <code>ideogram4-unconditional_transformer-q8_0.gguf</code> (10.1 GB)💡 Note: The repository includes inference speed and memory usage charts to help you choose the best quantization for your system.Alternative link for unet files (Civitai - Q4_0 only):- Download from: <a target="_blank" rel="ugc" href="https://civarchive.com/models/2681714/ideogram-4-gguf">https://civarchive.com/models/2681714/ideogram-4-gguf</a>- Note: File names may differ slightly from HuggingFace2. Text Encoder → ComfyUI/models/clip/- Qwen3-8B-Q4_K_M.gguf- Source: <a target="_blank" rel="ugc" href="https://huggingface.co/Qwen/Qwen3-8B-GGUF">https://huggingface.co/Qwen/Qwen3-8B-GGUF</a>3. Prompt Enhancer (optional) → ComfyUI/models/clip/- gemma4_e4b_it_fp8_scaled.safetensors- Source: <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/gemma-4">https://huggingface.co/Comfy-Org/gemma-4</a>- Automatically converts natural language to JSON format4. VAE → ComfyUI/models/vae/- flux2-vae.safetensors- Download: <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/vae/flux2-vae.safetensors">https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/vae/flux2-vae.safetensors</a>---<h2 id="custom-node-required">⚙️ Custom Node Required</h2>Single GGUF custom node by <code>molbal</code> (for both UNET and CLIP)- Repository: [molbal/ComfyUI-GGUF](<a target="_blank" rel="ugc" href="https://github.com/molbal/ComfyUI-GGUF">https://github.com/molbal/ComfyUI-GGUF</a>)- Install: <code>git clone https://github.com/molbal/ComfyUI-GGUF.git</code>- Nodes: UnetLoaderGGUF + GGUFCLIPLoader⚠️ Important: This is a fork of city96/ComfyUI-GGUF with Ideogram 4 support. If you have city96 version installed, remove it first (same folder name causes conflicts).Installation steps:1. Stop ComfyUI2. Delete the entire <code>ComfyUI-GGUF</code> folder in custom_nodes (not just the files inside it)3. Clone: <code>git clone https://github.com/molbal/ComfyUI-GGUF.git</code>4. Restart ComfyUI💡 Optional: If you encounter any import errors, install requirements:<pre><code>pip install -r ComfyUI-GGUF/requirements.txt</code></pre>✅ That's it! This single extension handles both UNET and CLIP loading.---<h2 id="choosing-the-right-quantization">📊 Choosing the Right Quantization</h2>The HuggingFace repository includes helpful charts for:- Inference Speed: How fast each quant generates images- Memory Usage: How much VRAM each quant requiresGeneral Guidelines:- Q4_0: Lowest VRAM (~5.5 GB), fastest, good quality- Q5_0/Q5_1: Balanced VRAM (~6.8-7.3 GB), better quality- Q8_0: Highest VRAM (~10 GB), best quality, slower💡 Check the charts in the repository for detailed comparisons!---<h2 id="json-prompting-(optional)">🗒️ JSON Prompting (Optional)</h2>Ideogram 4 works best with structured JSON prompts. This workflow uses Gemma 4 to automatically convert your natural language prompts into JSON format for:- ✅ Better text rendering and typography- ✅ More accurate composition control- ✅ Consistent results across generationsJust write a simple prompt like "A poster for a coffee shop" and let Gemma 4 handle the rest!---<h2 id="quick-tips">💡 Quick Tips</h2>- For TurboTime LoRA (v2.0): Use only <code>ideogram4-transformer-q4_0.gguf</code> (no unconditional model needed) - TurboTime LoRA: Download from: <a target="_blank" rel="ugc" href="https://huggingface.co/ostris/ideogram_4_turbotime_lora">https://huggingface.co/ostris/ideogram_4_turbotime_lora</a>- For standard mode (v1.0): Use both main and unconditional models- CFG Settings: 0.0 for TurboTime, 7.0 for standard mode- Scheduler: mu=0.5, std=1.75 for TurboTime | mu=0.0, std=1.5 for standard- Quantization: Q4_0 for lowest VRAM, Q8_0 for best quality---<h2 id="system-requirements">💻 System Requirements</h2>Minimum:- GPU: 8GB+ VRAM (12GB recommended)- RAM: 16GB+ (32GB recommended)- Virtual Memory (Pagefile): 32GB+ ⚠️Important: If you experience crashes during VAE decode, make sure your Windows Virtual Memory (Pagefile) is set to at least 32GB:- Settings → System → About → Advanced system settings → Performance → Settings → Advanced → Virtual memory- Set to 32GB or moreThis is especially important for systems with 16GB RAM running GGUF models!---<h2 id="turbotime-lora-tips">💡 TurboTime LoRA Tips</h2>Steps Guide (with TurboTime Lora):- 2 steps: Fastest (~15s), good for quick tests and drafts- 4-6 steps: Good balance (~30s), decent quality for iterations- 12-20 steps: ⭐ Recommended! Best quality/speed ratio (~1.5 min)- 30-50 steps: Maximum quality (~3-4 min), all fine details rendered- 50+ steps: Marginal improvement over 30, not worth the extra timeUnlike other Turbo LoRAs (Flux/SDXL), increasing steps does NOT overcook the image with TurboTime Lora! You can safely use 20+ steps for better anatomy and details.Pro Tip: If your prompt includes fine details (flowers, jewelry, text), use at least 20 steps for them to appear correctly.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/9dab4663-c94b-4ab4-bf63-db573f320806/original=true/9dab4663-c94b-4ab4-bf63-db573f320806.jpeg" /><hr /><h2 id="important:-custom-node-conflicts">⚠️ Important: Custom Node Conflicts</h2>If you have multiple GGUF-related custom nodes installed, they can conflict and cause "Unknown model architecture" errors.Common conflicts:- <code>ComfyUI-GGUF_KREA-2</code> - Causes GGUF loader to use wrong code- <code>ComfyUI-GGUF</code> (city96 version) - Conflicts with molbal version- Any other GGUF loader custom nodesSolution:1. Remove ALL conflicting custom nodes:- Delete <code>ComfyUI-GGUF_KREA-2</code> folder- Delete <code>ComfyUI-GGUF</code> folder (if city96 version)- Remove any other GGUF loaders2. Install only molbal/ComfyUI-GGUF:cd custom_nodesgit clone <a target="_blank" rel="ugc" href="https://github.com/molbal/ComfyUI-GGUF.git">https://github.com/molbal/ComfyUI-GGUF.git</a><h2 id="custom-node-conflicts-visual-guide">⚠️ Custom Node Conflicts - Visual Guide</h2>The image below shows when the GGUF loader works vs fails with Ideogram 4:✅ Only molbal/ComfyUI-GGUF installed → Works perfectly❌ city96 version installed → "Unknown model architecture" error❌ ComfyUI-GGUF_KREA-2 installed → Conflicts with loader❌ Multiple GGUF nodes installed → Wrong loader selectedSolution: Remove all other GGUF custom nodes and keep only molbal/ComfyUI-GGUF<h2 id="advanced:-managing-multiple-gguf-custom-nodes"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f5d08c00-bdff-4a0f-b5bd-a0cc9d3c32a4/original=true/f5d08c00-bdff-4a0f-b5bd-a0cc9d3c32a4.jpeg" /> 🔧 Advanced: Managing Multiple GGUF Custom Nodes</h2>If you work with multiple GGUF models (Ideogram 4, KREA-2, etc.) and need different custom node versions:Quick Switch Method:1. Install each custom node version2. Zip each folder with a descriptive name:- <code>ComfyUI-GGUF-molbal.zip</code> (for Ideogram 4)- <code>ComfyUI-GGUF-city96.zip</code> (for other models)- <code>ComfyUI-GGUF_KREA-2.zip</code> (for KREA-2)3. Delete the extracted folders4. When you need a specific version, just extract the corresponding zip5. Restart ComfyUIThis avoids conflicts and makes switching instant without reinstalling<h2 id="credits">🙏 Credits</h2>- Ideogram 4 Model: [Ideogram AI] <a target="_blank" rel="ugc" href="https://ideogram.ai/">https://ideogram.ai/</a>- GGUF Conversion &amp; Custom Node: [molbal] <a target="_blank" rel="ugc" href="https://github.com/molbal/ComfyUI-GGUF">https://github.com/molbal/ComfyUI-GGUF</a>- TurboTime LoRA: [ostris] <a target="_blank" rel="ugc" href="https://huggingface.co/ostris/ideogram_4_turbotime_lora">https://huggingface.co/ostris/ideogram_4_turbotime_lora</a>- Workflow: [dvdufo] <a target="_blank" rel="ugc" href="https://civarchive.com/user/dvdufo">https://civarchive.com/user/dvdufo</a>

Ideogram 4 GGUF Workflow

## v1.0 - Initial GGUF WorkflowA simple workflow to test Ideogram 4 with GGUF quantized models on ComfyUI.Features:- Works with GGUF models (Q4_0 quantization)- Tested on RTX 3060 12GB with 16GB RAM- Uses Dual Model CFG Guider for proper conditioning- Includes automatic JSON prompting via Gemma 4Requirements:- ComfyUI (updated version)- GGUF models: ideogram4-Q4_0.gguf + ideogram4_uncond-Q4_0.gguf- Text encoder: Qwen3-8B-Q4_K_M.gguf- Custom node: molbalIf you want faster generation with TurboTime LoRA, check out v2.0! 😊

Files with SHA256: e99031a06c0c6881132b0aec6a4802f7fd93f1656176929bedacfe4bb2601106