<h2 id="deprived-flux-2-klein-9b-simple-all-in-one-text-to-image-workflow">DepriveD FLux 2 Klein 9B – Simple All-in-One Text-to-Image Workflow</h2>A clean, beginner-friendly, one-stop workflow for Flux 2 Klein 9B. Press a few buttons → hit Queue (or just Ctrl + Enter) → get great images.<h3 id="what-this-workflow-is">What this workflow is</h3>This is designed to be simple and fast. Everything important is clearly labeled and color-coded so you can understand it quickly. No messy spaghetti.<h3 id="built-in-prompt-creators-(free)">Built-in Prompt Creators (Free)</h3>I made three custom prompt creators that are included:<ul><li>Scene Creator → Generates a random scene with random assets</li><li>Creature Creator → Generates a random colossal creature in a random setting</li><li>Character Creator → Generates a random character in a random setting</li></ul>Only use ONE Creator a a time. When a creator is turned on, you don’t need a positive prompt. Together they give tens of thousands of possible combinations (before even changing the seed). You can edit them later if you know how.<h3 id="optional-ai-prompt-(llm)">Optional AI Prompt (LLM)</h3>There is an optional LLM that can flesh out your prompts. You can use it:<ul><li>With a normal positive prompt</li><li>Together with the Creators</li><li>Or by itself</li></ul>It is not required.<h3 id="required-custom-nodes">Required Custom Nodes</h3><ul><li>ComfyUI-Pixaroma</li><li>rgthree-comfy (for reroute nodes)</li></ul>Install them through ComfyUI Manager if you don’t already have them.<h3 id="important-notes">Important Notes</h3><ul><li>Images do NOT save automatically. On the Compare Image node, click Show and then Save manually.</li><li>Sage Attention is enabled for speed. If you have issues, turn it off.</li><li>Recommended Steps and CFG are shown in the General Info panel on the left.</li></ul><h3 id="how-to-use-(super-simple)">How to use (super simple)</h3><ol><li>Choose your resolution</li><li>Set Steps &amp; CFG (or leave the recommended settings)</li><li>(Optional) Turn on AI Prompt</li><li>Choose an Art Style if you want</li><li>Write a prompt or turn on one of the Creators</li><li>(Optional) Add LoRAs</li><li>Hit Queue or just press Ctrl + Enter</li></ol><h3 id="required-models">Required Models</h3><ul><li>Flux-Klein-2-9B-INT8 ConvRot</li><li>Qwen_3_8b_fp8mixed (CLIP)</li><li>Full_Encoder_Small_Decoder VAE</li><li>Dolphin-2.9-Llama3-8B-Q4_K_M (for the optional AI Prompt)</li></ul>A system prompt for the LLM is also included in the download.<h3 id="about-the-loras">About the LoRAs</h3>The LoRAs currently loaded in the panel were not used for any of the example images. They are just placeholders. Feel free to replace them with whatever you want.<h3 id="happy-creating!!!!">HAPPY CREATING!!!!</h3>

-DepriveD- Flux 2 Klein 9B T2I Workflow [Built-in Creator]

About this versionAbout this version (v2.0)<ul><li>All creators cleaned up and expanded for more variety</li><li>Scene Creator now includes a brand new Neon City location</li><li>Added clean dropdowns for Upscale, Art Style, and Background (with N/A option to disable any of them)</li><li>Background dropdown also includes a new White option</li><li>Added a Run Log that shows hardware info + previous generation times</li><li>Negative prompt removed from the main UI (a simple default is now used internally and can still be adjusted inside the Conditioning panel if needed)</li><li>Parameter panel made larger and easier to read</li><li>Larger, clearer generation timer</li></ul>Overall cleaner, more compact, and easier to use while delivering slightly better results.

DeprivedFlux2Klein9BT2I_v20.txt

DeprivedFlux2Klein9BT2I_v20.json

v2.0

v1.0

<h2 id="deprived-krea-2-simple-all-in-one-text-to-image-workflow">DepriveD Krea 2 – Simple All-in-One Text-to-Image Workflow</h2>A clean, beginner-friendly, one-stop workflow for Krea 2. Press a few buttons → hit Queue (or just Ctrl + Enter) → get great images.<h3 id="what-this-workflow-is">What this workflow is</h3>This is designed to be simple and fast. Everything important is clearly labeled and color-coded so you can understand it quickly. No messy spaghetti.<h3 id="built-in-prompt-creators-(free)">Built-in Prompt Creators (Free)</h3>I made three custom prompt creators that are included:<ul><li>Scene Creator → Generates a random scene with random assets</li><li>Creature Creator → Generates a random colossal creature in a random setting</li><li>Character Creator → Generates a random character in a random setting</li></ul>Only use ONE Creator a a time. When a creator is turned on, you don’t need a positive prompt. Together they give tens of thousands of possible combinations (before even changing the seed). You can edit them later if you know how.<h3 id="optional-ai-prompt-(llm)">Optional AI Prompt (LLM)</h3>There is an optional LLM that can flesh out your prompts. You can use it:<ul><li>With a normal positive prompt</li><li>Together with the Creators</li><li>Or by itself</li></ul>It is not required.<h3 id="required-custom-nodes">Required Custom Nodes</h3><ul><li>ComfyUI-Pixaroma</li><li>rgthree-comfy (for reroute nodes)</li></ul>Install them through ComfyUI Manager if you don’t already have them.<h3 id="important-notes">Important Notes</h3><ul><li>Images do NOT save automatically. On the Compare Image node, click Show and then Save manually.</li><li>Sage Attention is enabled for speed. If you have issues, turn it off.</li><li>Recommended Steps and CFG are shown in the General Info panel on the left.</li></ul><h3 id="how-to-use-(super-simple)">How to use (super simple)</h3><ol><li>Choose your resolution</li><li>Set Steps &amp; CFG (or leave the recommended settings)</li><li>(Optional) Turn on AI Prompt</li><li>Choose an Art Style if you want</li><li>Write a prompt or turn on one of the Creators</li><li>(Optional) Add LoRAs</li><li>Hit Queue or just press Ctrl + Enter</li></ol><h3 id="required-models">Required Models</h3><ul><li>Krea 2 INT8</li><li>Qwen3VL_4B_BF16 (CLIP)</li><li>Qwen Image VAE</li><li>Dolphin-2.9-Llama3-8B-Q4_K_M (for the optional AI Prompt)</li></ul>A system prompt for the LLM is also included in the download.<h3 id="about-the-loras">About the LoRAs</h3>The LoRAs currently loaded in the panel were not used for any of the example images. They are just placeholders. Feel free to replace them with whatever you want.<h3 id="happy-creating!!!!">HAPPY CREATING!!!!</h3>

-DepriveD- Krea 2 T2I Workflow [Built-in Creator]

About this version (v2.0)<ul><li>Switched scheduler from Simple → Beta for noticeably better overall quality with almost no speed loss</li><li>All creators cleaned up and expanded for more variety</li><li>Scene Creator now includes a brand new Neon City location</li><li>Added clean dropdowns for Upscale, Art Style, and Background (with N/A option to disable any of them)</li><li>Background dropdown also includes a new White option</li><li>Added a Run Log that shows hardware info + previous generation times</li><li>Negative prompt removed from the main UI (a simple default is now used internally and can still be adjusted inside the Conditioning panel if needed)</li><li>Parameter panel made larger and easier to read</li><li>Larger, clearer generation timer</li></ul>Overall cleaner, more compact, and easier to use while delivering slightly better results.

Files with SHA256: a666138b13aca1b51ba8f8d418069a797a6dca9080e7715859dbc1ada6fe0ebd