<h1 id="zit-styler:-seedvr-vhs-and-crt-workflow">📼 ZIT STYLER: SEEDVR VHS &amp; CRT WORKFLOW</h1>SquiDavey's ZIT Workflow combines the speed of Z-Image Turbo with a local LLM to generate improve your prompts, upscales with SeedVR, then processes the image through CRT Post Process Suite.<hr /><h2 id="how-it-works">🛠️ HOW IT WORKS</h2><ol><li>LLM Expansion: A local LLM (via LM Studio) expands simple concepts into rich, photorealistic descriptions.</li><li>Generation: Z-Image Turbo creates the base image efficiently.</li><li>Analog FX: The image passes through a CRT/VHS simulation for authentic glitch and signal noise.</li><li>Upscale: The SeedVR DiT model performs a tiled upscale on the final result.</li></ol><hr /><h2 id="required-models-and-downloads">📥 REQUIRED MODELS &amp; DOWNLOADS</h2><h3 id="1.-z-image-turbo-(the-generator)">1. Z-Image Turbo (The Generator)</h3>Place these in your <code>ComfyUI/models/</code> directory:<ul><li>UNET: <code>z_image_turbo_bf16.safetensors</code> Folder: <code>models/diffusion_models/</code></li><li>VAE: <code>ae.safetensors</code> Folder: <code>models/vae/</code></li><li>CLIP: <code>qwen_3_4b.safetensors</code> Folder: <code>models/text_encoders/</code></li></ul><h3 id="2.-seedvr-(the-upscaler)">2. SeedVR (The Upscaler)</h3><ul><li>DiT Model: <code>seedvr2_ema_7b_fp16.safetensors</code> Folder: <code>models/checkpoints/</code></li><li>VAE: <code>ema_vae_fp16.safetensors</code> Folder: <code>models/vae/</code></li></ul><h3 id="3.-post-process-(crt-effect)">3. Post-Process (CRT Effect)</h3><ul><li>Model: <code>4x-ClearRealityV1.pth</code> Folder: <code>models/upscale_models/</code></li></ul><h3 id="4.-local-llm-(prompting)">4. Local LLM (Prompting)</h3><ul><li>Software: <a target="_blank" rel="ugc" href="https://lmstudio.ai/">LM Studio</a> Instructions: Download a model (e.g., Qwen 2.5 7B), go to the Developer/Server tab, and Start Server on port <code>1234</code>.</li></ul><hr /><h2 id="workflow-instructions">⚡ WORKFLOW INSTRUCTIONS</h2><h3 id="mode-a:-text-to-image-(new-generation)">MODE A: Text-to-Image (New Generation)</h3><ol><li>Ensure LM Studio server is running.</li><li>Type a simple concept in the Prompt group (Green Node).</li><li>Set the "SEEDVR / LOAD IMAGE" switch to True.</li><li>Press Queue Prompt.</li></ol><h3 id="mode-b:-image-to-upscale-(existing-image)">MODE B: Image-to-Upscale (Existing Image)</h3><ol><li>Set the "SEEDVR / LOAD IMAGE" switch to False.</li><li>(Optional) Disable the "LLM" group to save time.</li><li>Upload your image into the "UPSCALE BYPASS" node.</li><li>Press Queue Prompt.</li></ol>Note: If you run out of VRAM during the Upscale step, try reducing the tile size in the SeedVR settings and increase the blocks to swap. Enjoy 🥳

Files with SHA256: 44e299a9e20abda9e145e3dcb198c0a610282c6f39835d142f178daec7ddc020