🚀 RIPLEY SPACE ENGINE v5.0 RELEASE NOTES (UNIVERSAL EDITION)The definitive milestone of the local Event-Driven AI Sandbox. Upgraded to Version 5.0 with full monolithic integration.🛸 CORE SYSTEM ARCHITECTURE: Version 5.0 introduces a seamless, unified environment linking the custom Web UI, the local Ollama text layer (Gemma 3), and advanced 4-bit tensor-core rendering pipelines (<code>nvfp4</code>). Powered by a 10x RIFE VFI temporal motion interpolation layer, the engine delivers fluid, seamless cinematic loops directly on budget-tier hardware setups.⚡ PERFORMANCE NOTICE: Transitioning to a pure base-model baseline architecture has successfully doubled (2x) the execution and rendering speed of the temporal LCM loops! The runtime environment operates exclusively on direct base-model matrix calculations for peak hardware optimization. Loading third-party style modifiers will sever the speed advantage and break runtime stability.<hr />🛡️ MANUAL WEIGHTS DEPLOYMENT (MANDATORY PRE-REQUISITE)To ensure absolute path stability across custom layouts, you can now use ANY custom weights for the Krea2 pipeline. You can either download the official baseline files or find alternative compatible versions on Civitai/HuggingFace by searching for their names.However, you MUST manually rename your chosen weight files exactly to match the engine's hard-locked configuration strings before launching the setup:<ol><li>📂 krea2_turbo_nvfp4.safetensors (Diffusion Model)<ul><li>Action: Take your favorite working Krea2 checkpoint (such as your previous <code>diffusion_pytorch_model.safetensors</code> or any other custom Krea2 turbo variant) and rename it exactly to <code>krea2_turbo_nvfp4.safetensors</code>.</li><li>Search Keywords: Look for <code>Krea2 Turbo</code> or <code>Krea 2 NVFP4</code>.</li><li>Path: Move to <code>ComfyUI/models/diffusion_models/</code></li></ul></li><li>📂 qwen3-vl-4b-heretic_nvfp4.safetensors (CLIP Text Encoder)<ul><li>Action: Download the required CLIP weights for Krea2, or search Civitai/HuggingFace for this exact model type, then rename it exactly to <code>qwen3-vl-4b-heretic_nvfp4.safetensors</code>.</li><li>Search Keywords: Look for <code>qwen3-vl-4b-heretic</code> or <code>Qwen VL CLIP Krea2</code>.</li><li>Path: Move to <code>ComfyUI/models/clip/</code></li></ul></li><li>📂 qwen_image_vae.safetensors (VAE Decoder)<ul><li>Action: Download the compatible image VAE decoder for the pipeline, or find it by its original name, then rename it exactly to <code>qwen_image_vae.safetensors</code>.</li><li>Search Keywords: Look for <code>qwen_image_vae</code> or <code>Qwen Image VAE for Krea</code>.</li><li>Path: Move to <code>ComfyUI/models/vae/</code></li></ul></li></ol>(Note: The runtime environment operates exclusively on direct matrix calculations matched to these filenames. Failing to rename your files will break the launch script stability).<hr />🚀 HOW TO INSTALL &amp; RUN<ol><li>COMPILATION: Run <code>1_Setup_Studio.bat</code>. The engine will check the local environment configurations and automatically compile the active <code>ripley-brain</code> layer inside your local Ollama instance.</li><li>LAUNCH: Ensure your ComfyUI server is active on its standard port (<code>8188</code>). Execute <code>2_Run_Studio.bat</code> to mount the local FastAPI app container and spin up your borderless standalone workspace window (1400x900).</li></ol><hr />⚠️ IMPORTANT HARDWARE NOTICE FOR 8GB VRAM / RTX 5050 USERSPlease close your main system browsers (Chrome/Edge/Firefox) before running the studio core launcher. This gives your RTX Tensor Cores maximum hardware headroom to weave the 3-frame LCM outputs into smooth 60fps cinematic video feeds without memory fragmentation!<hr />

RIPLEY SPACE ENGINE v5.0 [The Monolithic Universal AI Sandbox]

🪐 RIPLEY PLANETARY ENGINE v6.0 (THE SURVIVAL HUB)The definitive milestone of the local Event-Driven AI Sandbox. Upgraded to Version 6.0 — The Official Story Sequel.🛸 STORYLINE CONTEXT: The starship storyline has officially concluded. Ellen Ripley's escape pod has violently crash-landed on an uncharted, highly humid alien planet. To the user's surprise, the planetary environment closely resembles a normal Earth-like rustic countryside. Left with minimal clothing and accompanied only by a glitching, poorly functioning helper drone, she must navigate the wild elements and survive.⚡ SPEED &amp; CLARITY ACCELERATION: Version 6.0 integrates a brand-new, ultra-optimized 4-bit transformer matrix (<code>krea2_turbo_nvfp4</code>). By combining strict NVFP4 quantization with advanced weight-clipping, the rendering pipeline delivers an immediate execution speed boost. It generates highly precise surface textures—including complex dirt accumulation, skin perspiration, and intricate fabric degradation—entirely within the budget 8GB VRAM boundary.🛡️ MANUAL WEIGHTS DEPLOYMENT (MANDATORY PRE-REQUISITE)To ensure absolute path stability across custom or external disk drive layouts, you can now use ANY custom weights for the Krea2 pipeline. You can either download the official baseline files or find alternative compatible versions on Civitai/HuggingFace by searching for their names.However, you MUST manually rename your chosen weight files exactly to match the engine's hard-locked configuration strings before launching the setup:<ol><li>📂 krea2_turbo_nvfp4.safetensors (Diffusion Model)<ul><li>Action: Take your favorite working Krea2 checkpoint (such as your previous <code>diffusion_pytorch_model.safetensors</code> or any other custom Krea2 turbo variant) and rename it exactly to <code>krea2_turbo_nvfp4.safetensors</code>.</li><li>Search Keywords: Look for <code>Krea2 Turbo</code> or <code>Krea 2 NVFP4</code>.</li><li>Path: Move to <code>ComfyUI/models/diffusion_models/</code></li></ul></li><li>📂 qwen3-vl-4b-heretic_nvfp4.safetensors (CLIP Text Encoder)<ul><li>Action: Download the required CLIP weights for Krea2, or search Civitai/HuggingFace for this exact model type, then rename it exactly to <code>qwen3-vl-4b-heretic_nvfp4.safetensors</code>.</li><li>Search Keywords: Look for <code>qwen3-vl-4b-heretic</code> or <code>Qwen VL CLIP Krea2</code>.</li><li>Path: Move to <code>ComfyUI/models/clip/</code></li></ul></li><li>📂 qwen_image_vae.safetensors (VAE Decoder)<ul><li>Action: Download the compatible image VAE decoder for the pipeline, or find it by its original name, then rename it exactly to <code>qwen_image_vae.safetensors</code>.</li><li>Search Keywords: Look for <code>qwen_image_vae</code> or <code>Qwen Image VAE for Krea</code>.</li><li>Path: Move to <code>ComfyUI/models/vae/</code></li></ul></li></ol>(Note: The generation engine operates on hard-locked perspective prompts—such as <code>three-quarter view</code>—to completely eliminate image-flipping issues during ultra-fast KSampler sampling).🚀 HOW TO INSTALL &amp; RUN / LEGACY ROLLBACK<ol><li>COMPILATION: Run <code>1_Setup_Studio.bat</code>. The engine will verify local environment configurations and automatically compile the active planetary-survival <code>ripley-brain</code> logic layer inside your local Ollama instance.</li><li>LAUNCH: Ensure your ComfyUI server is active on its standard port (<code>8188</code>). Execute <code>2_Run_Studio.bat</code> to mount the local FastAPI app container, update the event-intercept subsystem, and spin up your borderless standalone workspace window (1400x900).</li><li>🕒 RESTORING VERSION 5.0 (LEGACY ROLLBACK): If your hardware configuration encounters structural stability issues with the 4-bit quantum matrix, or if you prefer the classic starship narrative architecture, you can fully downgrade the pipeline back to Version 5.0:<ul><li>Step A: Open your local terminal in the root engine directory and execute: <code>git checkout tags/v5.0</code> (or run <code>1_Rollback_To_v5.bat</code> if available).</li><li>Step B: Replace your active <code>modified_graph</code> payload with the backup configuration layout located at <code>comfy_helpers/backups/v5_linear_graph.json</code> to restore the classic linear KSampler pathing.</li><li>Step C: Rerun <code>1_Setup_Studio.bat</code> to re-align the active local LLM layer with the legacy starship-context weights.</li></ul></li></ol>🎓 CREDITS, COMPLIANCE &amp; LEGAL NOTICES<ul><li>Core Architecture: Autonomous Event-Driven AI Sandbox Subsystem.</li><li>Optimization Layer: Custom 4-bit Quantization Baseline Matrix calculations.</li><li>Licensing Agreement: Free utilization is fully granted under the Community Tier for independent creators/studios with less than 50 employees and under $1M annual revenue.</li><li>Disclaimers: Independent middleware architecture execution layer. All source-weights remain the proprietary property of their respective creators.</li></ul>

Files with SHA256: da1f1382b2a686c222b70ca375743d017e592a319fc2c7b0965414e75b6633b2