Apple MLX native implementations of state-of-the-art generative image & video models
-
Updated
Oct 14, 2026 - Python
Apple MLX native implementations of state-of-the-art generative image & video models
A professional, high-end AI Video & Image Upscaling GUI for SeedVR2 — no command line needed.
SeedVR2 1.4B image restoration and upscaling nodes for ComfyUI
SeedVR2 视频/图像超分辨率修复工具箱:Tauri 桌面版(一键安装+增量更新)与网页版/便携包,支持 3B/7B/7B-Sharp、FP8、GPU Block Swap 省显存、一键与批量修复、5 种语言界面,无需 ComfyUI | SeedVR2 video & image super-resolution toolkit: Tauri desktop (one-click install + incremental updates) + web/portable, 3B/7B/7B-Sharp & FP8, GPU Block Swap, one-click & batch restore, 5-language UI, no ComfyUI
Cog wrapper for SeedVR2 (3B/7B) video & image restoration with optional color fix
AIEZO UpScale - Làm nét & phóng to ảnh lên 2K/4K/8K bằng AI (SeedVR2), chạy trên máy, miễn phí. https://aiezo.netlify.app
Custom nodes for ComfyUI - image resizing, batch processing & prompt automation for LoRA training
Compare Muapi image enhancement APIs with runnable examples
AI PBR 贴图生成工具:一句提示词跑完 文生图→无缝→放大→PBR材质分解,一次输出 7 张贴图。ComfyUI 无头封装 + 苹果风网页界面 + Three.js 光影 3D 预览。
ComfyUI SeedVR2 video/image upscaler fork with experimental Intel XPU / PyTorch XPU support for Arc and Core Ultra iGPU
Shot-aware SeedVR2 video remaster pipeline (per-shot tuning, frame-exact, resume-safe; CUDA + AMD ROCm) + a full real-world writeup
Muapi Image Upscaler API endpoint guide, comparison, and runnable API examples.
A MAC ONLY-MLX optimized Pinokio app for SeedVR2 upscale model using native MLX inference
⚡ Production AI Studio: Qwen 2.1 Turbo (NVFP4 ~1.7s T2I), Qwen3-VL 8B Editing & SeedVR2 4K Super-Resolution
Native macOS image-generation workspace for Apple Silicon, built with SwiftUI, MFLUX, and MLX.
🎞️ Batch AI image upscaler for Windows built on ByteDance SeedVR2-3B (one-step diffusion SR). Wraps the official inference scripts — Gradio FR/EN UI, GGUF/FP8/FP16/.pth models, exact filename preservation, tiling, pause/resume, ETA, EXIF kept.
Run SeedVR2 video upscalers on Tesla V100 (Volta) — fixes the silent black-video fp16 bug with an FP16Safe-style patch.
A local, private roleplay studio: a Character you talk to (Ollama or any OpenAI-compatible server), every scene painted by local image models (mflux on a Mac, ComfyUI on Windows/NVIDIA), voiced with cloned TTS and made into 3D Gaussian splats (SHARP, TripoSplat, LiTo). Runs on a Mac, a Windows PC or both. Deno + Vue 3, no cloud needed.
To associate your repository with the seedvr2 topic, visit your repo's landing page and select "manage topics."