Experimenting with a prompt to create these different versions. Used Zimage and Qwen Edit.
If you enjoy my work and would like a version of yourself in one of the styles I use, send over an image; this is completely free of charge.
Experimenting with a prompt to create these different versions. Used Zimage and Qwen Edit.
If you enjoy my work and would like a version of yourself in one of the styles I use, send over an image; this is completely free of charge.
Random Zimage Creations
If you enjoy my work and would like a version of yourself in one of the styles I use, send over an image; this is completely free of charge.
🎨 zimage to prompt and generate image and Qwen Image Edit for final touches.👇
🖼️ Left Zimage 🖌️ Right Qwen Edited
If you enjoy my work and would like a version of yourself in one of the styles I use, send over an image; this is completely free of charge.
Z-Image Turbo ra mắt với kiến trúc S3-DiT: kết hợp dòng đơn thống nhất văn bản và hình ảnh, cho khả năng tạo ảnh đa ngôn ngữ vượt trội. Dùng GGUF, K-Quants, I-Quants để tối ưu mô hình 6B trên GPU consumer. Chạy mượt trên 6GB VRAM nhờ mmap và tải từng phần. Q3_K_S phù hợp cho VRAM thấp, Q8_0 trên 4090 đạt <1s. Dùng diffusers + CPU offload hoặc ComfyUI-GGUF plugin để tránh lỗi tensor. Tích hợp Qwen3-4B giúp hiểu sâu ngữ nghĩa qua Chain-of-Thought.
Z-Image Turbo giờ có thể chạy mượt trên GPU 6GB nhờ công nghệ GGUF! Chỉ cần card RTX 3060/4060, bạn đã tạo được ảnh AI chất lượng cao. Dùng ComfyUI + plugin GGUF, tải model Q3_K_S (6GB), Q4_K_M (8GB) hoặc Q8_0 (12GB+). Cần RAM 16-32GB, SSD, và đúng file: UNet, CLIP (Qwen3-4B GGUF), VAE. Mẹo: Dùng 8-10 bước, CFG=1.0, sampler Euler. Hỗ trợ tiếng Việt & tiếng Trung, kể cả chữ viết. Thử ngay: "Cô gái mặc Hanbok đứng bên cầu sương mù, phong cảnh thủy mặc"! #AIArt #GGUF #ZImage #ComfyUI #AIẢnh #TríTuệ
It's kinda getting hard to hate on China cos every few months or so, some new text/image/video generation model comes along that I can actually run on my hardware, and it beats whatever came before it a year ago, half a year even sometimes. Like that Z-Image Turbo and Qwen 3.
I know China's government is so fucking bad. They're always one of the top 3 countries who uses my snowflake proxy whenever I check, but yeah. I actually like where they're going with "AI".
🎨 zimage to prompt and generate image and Qwen Image Edit for final touches.👇
If you enjoy my work and would like a version of yourself in one of the styles I use, send over an image; this is completely free of charge.
🎨 zimage to prompt and generate image and Qwen Image Edit for final touches.👇
If you enjoy my work and would like a version of yourself in one of the styles I use, send over an image; this is completely free of charge.
🎨 zimage to prompt and generate image and Qwen Image Edit for final touches.👇
If you enjoy my work and would like a version of yourself in one of the styles I use, send over an image; this is completely free of charge.
Just catching up after a year of not fiddling around with local image generation.
Yeah this was made with Z-Image, the model that's been popping up after 2 weeks or so because it generates images fast, only taking about 25 seconds on my RTX 3060 power capped at 145w, with good prompt adherence.. or some thing.
Idk about prompt adherence but I'm still "training" myself to make prompts.
#zimage #localai #comfyui #bears #furry #nvidia #cuda #python
Generative #KI macht keinen Spaß, wenn man "nur" ein #Apple #MacBook Pro #M1 hat. Dieses Bild hat mit #ZImage Turbo unter #ComfyUI 40 Minuten für die Generierung gebraucht. Turbo, ja nee is klar :mastolol: Der Prompt war übrigens: "a vase with a colorful bouquet of wildflowers", falls das jemand nachmachen möchte.
🚀 Behold, the latest buzzword salad from GitHub's trendy kitchen: Z-Image, a 6-billion-parameter monstrosity that promises to churn out images faster than you can say "overhyped AI model." 🌟✨ But wait, there's more! You can now manage prompts and fix vulnerabilities like never before—because who wouldn't want their security tools tangled with image generation? 😂🔧
https://github.com/Tongyi-MAI/Z-Image #ZImage #OverhypedAI #ImageGeneration #GitHub #SecurityTools #TrendingTech #HackerNews #ngated
Z-Image: Powerful and highly efficient image generation model with 6B parameters
https://github.com/Tongyi-MAI/Z-Image
#HackerNews #ZImage #Powerful #ImageGeneration #6BParameters #AIModel #GitHub
Stable-diffusion.cpp, phiên bản tối ưu của Stable Diffusion, nay đã hỗ trợ "Z-image". Cập nhật này có thể mang lại cải tiến về hiệu suất hoặc tính năng cho việc tạo ảnh AI cục bộ.
#StableDiffusion #ZImage #AIArt #LocalAI #AI_TạoẢnh #TríTuệNhânTạo #HỗTrợZImage
https://www.reddit.com/r/LocalLLaMA/comments/1pbfisv/stablediffusioncpp_now_supports_zimage/