<h2 id="quick-links">🔗 Quick Links</h2>🌸 <a target="_blank" rel="ugc" href="https://civarchive.com/collections/17587706">Browse My Krea 2 LoRA Collection</a> 🤖 <a target="_blank" rel="ugc" href="https://t.me/drgirls_ai_bot">Try the Free Telegram AI Generator</a> ⭐ <a target="_blank" rel="ugc" href="https://civarchive.com/user/NovaAILabs">Follow NovaAILabs</a> ⚡ <a target="_blank" rel="ugc" href="https://civarchive.com/models/2931118/krea-2-turbo-comfyui-workflow-uncensored">Krea 2 Turbo Workflow — Uncensored</a> ⚡ <a target="_blank" rel="ugc" href="https://civarchive.com/models/2940587/krea-2-turbo-or-dual-lora-comfyui-workflow">Krea 2 Turbo Dual-LoRA Workflow</a> 💎 <a target="_blank" rel="ugc" href="https://boosty.to/diexell">Premium Releases &amp; Custom LoRA Commissions</a><hr /><h2 id="krea-2-turbo-triple-lora-workflow">⚡ Krea 2 Turbo — Triple LoRA Workflow</h2>A clean and ready-to-use ComfyUI workflow for Krea 2 Turbo, designed to work with up to three LoRAs in the same generation.Use one, two or three LoRAs, assign an individual trigger word and strength to every slot, write your scene and generate.The workflow is useful for combining multiple characters, mixing a character with style or concept LoRAs, and building more complex Krea 2 scenes without manually rebuilding your graph.<hr /><h2 id="experimental-workflow-have-fun!">🧪 Experimental Workflow — Have Fun!</h2><blockquote>This is an experimental Triple LoRA workflow. Mixing three LoRAs can produce exciting, unusual and sometimes unpredictable results — and that is part of the fun!Try different LoRA combinations, strengths, trigger words, character positions, prompts and seeds. If identities begin to blend, lower the LoRA strengths and describe each subject’s position more clearly.Experiment, create and have fun, friends! Share your best and most unexpected results in the gallery — I would love to see what you make.</blockquote><h2 id="main-features">✨ Main Features</h2>✔ Supports up to 3 Krea 2 LoRAs simultaneously ✔ Independent enable/disable switch for every LoRA ✔ Separate strength control for LoRA 1, LoRA 2 and LoRA 3 ✔ Individual trigger word for every LoRA ✔ Clean empty prompt and LoRA slots — ready for your own models ✔ Optional AI prompt enhancement ✔ Prompt preview before generation ✔ Krea 2 Conditioning Rebalance included ✔ Automatic aspect-ratio and resolution selection ✔ Automatic 2× output scaling ✔ Zeroed negative conditioning — no negative prompt required ✔ Optimized for fast Krea 2 Turbo generation ✔ Model links and bilingual instructions included inside the workflowYou can use the workflow with only one or two LoRAs by keeping the unused slots disabled.<hr /><h2 id="how-to-use">🚀 How to Use</h2><ol><li>Enter your scene in the Prompt field.</li><li>Select a file for LoRA 1.</li><li>Enter its trigger word and set the strength.</li><li>Repeat for LoRA 2 and LoRA 3 if needed.</li><li>Enable only the LoRA slots you want to use.</li><li>Enable Prompt Enhance if you want the workflow to expand your prompt automatically.</li><li>Select your aspect ratio and resolution.</li><li>Click Queue Prompt.</li></ol><h3 id="three-character-prompt-example">Three-character prompt example</h3><pre><code>text</code></pre><pre><code>Milena Novak, Seo Yuna and Kira Laurent standing together in a luxury fashion studio, balanced full-body group composition, each woman clearly separated, elegant editorial outfits, natural poses, dramatic softbox lighting, photorealistic magazine photography</code></pre>For scenes with multiple characters, clearly describe each character’s position, outfit and action. Phrases such as <code>on the left</code>, <code>in the center</code> and <code>on the right</code> help Krea 2 keep the composition organized.<hr /><h2 id="suggested-lora-strength">🎚️ Suggested LoRA Strength</h2><ul><li>1 LoRA: <code>0.85–1.0</code></li><li>2 LoRAs: <code>0.70–0.90</code> each</li><li>3 LoRAs: <code>0.55–0.80</code> each</li></ul>When combining several character LoRAs, start with lower values. If identities begin to blend, reduce all LoRA strengths and make the character positions more explicit in the prompt.<hr /><h2 id="default-settings">⚙️ Default Settings</h2><ul><li>Base model: <code>krea2_turbo_fp8_scaled.safetensors</code></li><li>Text encoder: <code>qwen3vl_4b_uncensored_int8_convrot.safetensors</code></li><li>VAE: <code>qwen_image_vae.safetensors</code></li><li>Default LoRA strength: <code>0.9</code></li><li>Steps: <code>8</code></li><li>CFG: <code>1.0</code></li><li>Sampler: <code>Euler</code></li><li>Scheduler: <code>Simple</code></li><li>Denoise: <code>1.0</code></li><li>Krea 2 Rebalance multiplier: <code>4.0</code></li><li>Per-layer weights: <code>1</code></li><li>LLM max tokens: <code>512</code></li><li>Default resolution: <code>1:1 — 3 MP</code></li><li>Output scaling: <code>2×</code></li><li>Prompt Enhance: <code>Off</code></li></ul>All three LoRA slots are disabled and empty by default. Select a valid LoRA file before enabling a slot.<hr /><h2 id="required-models">📦 Required Models</h2><h3 id="krea-2-turbo-checkpoint">Krea 2 Turbo checkpoint</h3><code>krea2_turbo_fp8_scaled.safetensors</code><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Krea-2/resolve/main/diffusion_models/krea2_turbo_fp8_scaled.safetensors">Download from Comfy-Org/Krea-2</a><h3 id="text-encoder">Text encoder</h3>The workflow is currently configured with:<code>qwen3vl_4b_uncensored_int8_convrot.safetensors</code>This is a community uncensored build. Place it in:<code>ComfyUI/models/text_encoders/</code>Official alternative:<a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Krea-2/resolve/main/text_encoders/qwen3vl_4b_fp8_scaled.safetensors">qwen3vl_4b_fp8_scaled.safetensors</a><h3 id="vae">VAE</h3><code>qwen_image_vae.safetensors</code><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Krea-2/resolve/main/vae/qwen_image_vae.safetensors">Download from Comfy-Org/Krea-2</a><hr /><h2 id="try-my-krea-2-loras">🌸 Try My Krea 2 LoRAs</h2>The workflow works especially well with original character LoRAs from my collection.Try combining different characters, styles, locations and lighting conditions. You can use three characters together, mix two characters with one style LoRA, or use a single LoRA while keeping the other slots disabled.👉 <a target="_blank" rel="ugc" href="https://civarchive.com/collections/17587706">Browse the Complete Krea 2 LoRA Collection</a><hr /><h2 id="included">📦 Included</h2>⚡ Ready-to-use Triple LoRA ComfyUI workflow 🎨 Support for one, two or three Krea 2 LoRAs 🔘 Independent LoRA enable/disable switches 🔑 Individual trigger-word handling 🧠 Optional AI prompt enhancement 📐 Resolution selector and automatic 2× scaling ⚙️ Tested Krea 2 Turbo settings 📝 English and Russian instructions inside the workflow 🔗 Direct model and NovaAILabs links<h2 id="enjoy-and-share-your-results">❤️ Enjoy &amp; Share Your Results</h2>If you enjoy this workflow:⭐ Leave a rating 💬 Share your experience 📸 Post your generations 💜 Follow NovaAILabsI would love to see which three LoRAs you combine!Which workflow should I build next?1 — Four LoRAs 2 — LoRA + ControlNet 3 — Image-to-image 4 — Character consistency workflow<h1 id="cover-image-prompt">COVER IMAGE PROMPT</h1><h2 id="main-prompt-english">Main prompt — English</h2><pre><code>A premium futuristic technology poster for an AI image-generation workflow, landscape 3:2 composition, matching a polished cyberpunk software advertisement. Dark navy and black high-tech studio background with subtle grid floor, electric cyan, neon green and vivid magenta accent lights. On the left, a clean large typography area with the exact headline: “KREA 2 TURBO” on the first line, “TRIPLE LoRA” on the second line, and “WORKFLOW” on the third line. “TRIPLE” in bright white, “LoRA” in a luminous cyan-to-green-to-magenta gradient, bold condensed italic sans-serif lettering. Above the headline, small spaced text: “NOVA AI LABS”. Below it, exact subtitle: “3 LoRAs • 1 WORKFLOW”. Under the subtitle, three separate glowing hexagonal module icons stacked vertically, each labeled exactly “LoRA”, the first outlined in electric cyan, the second in neon green, the third in vivid magenta; three luminous cables flow from the modules toward the workflow diagram. On the right, a large sharp glass holographic panel showing a complex ComfyUI node graph with three clearly visible LoRA loader branches, colorful connection wires, prompt nodes, model loader, sampler, conditioning rebalance and image output. Add a glowing “COMFYUI” badge at the lower left. Small top-right text: “AI × CREATIVITY × MORE POSSIBILITIES”. Small lower-right text: “BUILD FASTER • CREATE BOLDER”. Professional product-launch design, crisp edges, realistic glass reflections, high contrast, ultra-detailed, clean spacing, no people, no extra logos, no misspelled words, no duplicated modules, exactly three LoRA icons, SFW, 3:2 landscape poster.</code></pre><h2 id="short-version-for-krea-2">Short version for Krea 2</h2><pre><code>Futuristic cyberpunk software poster, dark navy high-tech studio, neon cyan, green and magenta lighting. Large clean title on the left: “KREA 2 TURBO — TRIPLE LoRA WORKFLOW”, subtitle “3 LoRAs • 1 WORKFLOW”. Three glowing hexagonal LoRA modules stacked vertically, cyan, green and magenta, with luminous cables leading to a large glass holographic ComfyUI workflow diagram on the right, three visible LoRA branches, colorful node connections, COMFYUI badge at bottom left, premium AI product advertisement, crisp professional layout, SFW, landscape 3:2.</code></pre>

Files with SHA256: 7c0b87c4716843abab0f8a51a3f6aae503f1aadc50ea36511f3f1ba9d7164a28