<h1 id="z-anime-or-full-anime-fine-tune-on-z-image-base">🎌 Z-Anime | Full Anime Fine-Tune on Z-Image Base</h1>Full Fine-Tune • Rich Aesthetics • Strong Diversity • Full Negative Prompt SupportBF16 &amp; FP8 &amp; GGUF &amp; AIO • Natural Language Prompts • 8GB VRAM🤗 Now also on Hugging Face: <a target="_blank" rel="ugc" href="https://huggingface.co/SeeSee21/Z-Anime">huggingface.co/SeeSee21/Z-Anime</a> — including the full Diffusers folder for <code>ZImagePipeline.from_pretrained()</code> use.<hr /><h2 id="what-is-z-anime">✨ What is Z-Anime?</h2>Z-Anime is a full fine-tune of Alibaba's Z-Image (Base) architecture — not a LoRA merge, but a completely retrained model optimized for anime aesthetics from the ground up.Built on the S3-DiT (Single-Stream Diffusion Transformer) with 6 billion parameters, Z-Anime inherits everything that makes Z-Image Base special: rich diversity, strong controllability, full negative prompt support and a high ceiling for fine-tuning — now fully tuned for anime.This page contains the complete Z-Anime family:<ul><li>🎌 Z-Anime Base — Full quality, full control, full creativity</li><li>⚡ Z-Anime Distill-8-Step — Great results in 8 steps</li><li>🚀 Z-Anime Distill-4-Step — Maximum speed, 4 steps</li><li>📦 GGUF Variants — Q8_0 + Q4_K_S for low VRAM / CPU / AMD</li><li>📦 AIO Variants — All-in-one checkpoints (Base + 4-Step + 8-Step)</li></ul>Each main variant is available in BF16 (~12 GB) and FP8 (~6 GB).<hr /><h2 id="key-features">🎯 Key Features</h2><ul><li>✅ Full fine-tune on Z-Image Base — not a LoRA merge</li><li>✅ Rich anime aesthetics with strong style diversity</li><li>✅ Natural language prompts — detailed descriptions, not tag lists</li><li>✅ High diversity across characters, poses, compositions and layouts</li><li>✅ LoRA training ready — perfect base for further fine-tuning</li><li>✅ Partially NSFW capable</li><li>✅ 8 GB VRAM compatible</li><li>✅ All variants supported by the official Z-Anime ComfyUI Workflow</li></ul><hr /><h2 id="z-anime-roadmap">🗺️ Z-Anime Roadmap</h2><h3 id="released">✅ Released</h3>🎌 Z-Anime Base — Full fine-tune on Z-Image Base, BF16 &amp; FP8⚡ Z-Anime Distill-8-Step — fast anime generation in 8 steps, CFG 1.0, BF16 &amp; FP8🚀 Z-Anime Distill-4-Step — ultra-fast anime generation in 4 steps, CFG 1.0, BF16 &amp; FP8📦 GGUF Variants — for low VRAM and AMD GPUs. Since CivitAI currently has no dedicated GGUF category, here is what the files represent:<ul><li>Z-Anime-Base-Q8_0 = Pruned Model FP8 (6.73 GB)</li><li>Z-Anime-Base-Q4_K_S = Pruned Model NF4 (4.2 GB)</li></ul>📦 AIO Versions — All variants with VAE + Text Encoder integrated in a single file:<ul><li>z-anime-base-aio (BF16 + FP8)</li><li>z-anime-distill-8step-aio (BF16 + FP8)</li><li>z-anime-distill-4step-aio (BF16 + FP8)</li></ul>🔧 Z-Anime ComfyUI Workflow — Official workflow, supports all variants (auto-detects Diffusion / GGUF / AIO loaders, optional LoRA, optional 1.5× upscale)🤗 Hugging Face Repo — full mirror including the Diffusers folder for Python users: <a target="_blank" rel="ugc" href="https://huggingface.co/SeeSee21/Z-Anime">huggingface.co/SeeSee21/Z-Anime</a>More updates coming — follow to stay notified! 🎌<hr /><h2 id="versions-overview">📦 Versions Overview</h2><h3 id="bf16-(~12-gb)">🟢 BF16 (~12 GB)</h3>Maximum precision. BFloat16 format, no quality compromise. Best for professional or commercial work and LoRA training. Still runs on 8 GB VRAM.<h3 id="fp8-(~6-gb)">🟡 FP8 (~6 GB)</h3>Recommended for most users. Half the file size, much faster downloads. Excellent quality, barely distinguishable from BF16. Perfect for everyday use and testing.<h3 id="gguf">🔵 GGUF</h3>Optimized for lightweight inference setups, especially useful for low VRAM, CPU inference, or alternative backends.<h3 id="aio">🟣 AIO</h3>All-in-one checkpoints with image model + Text Encoder + VAE integrated into a single file. Single-file convenience, no extra loaders needed.<hr /><h2 id="z-anime-base">🎌 Z-Anime Base</h2>The foundation of the Z-Anime family. A full fine-tune with the highest quality ceiling, the widest creative range and full negative prompt support.Recommended Settings:<pre><code>Steps: 28–50
CFG: 3.0–5.0 (up to 9.0 possible)
Sampler: euler_ancestral
Scheduler: beta
Negative: strongly recommended — very responsive!
</code></pre>CFG Guide: 3.0–5.0 is the sweet spot for balanced quality and creativity. 5.0–7.0 gives tighter prompt adherence. 7.0–9.0 is for maximum control — watch for over-saturation. Above 9.0 is not recommended.Negative prompts have full effect on Z-Anime Base. The official workflow ships with an optimized negative prompt ready to use.<hr /><h2 id="z-anime-distill-8-step">⚡ Z-Anime Distill-8-Step</h2>The sweet spot of the family. Distilled from Z-Anime Base, delivering strong anime results in just 8 steps. Much faster than Base while keeping most of the quality intact.Recommended Settings:<pre><code>Steps: 8
CFG: 1.0 (max ~1.5)
Sampler: euler_ancestral
Scheduler: beta
Negative: limited effect
</code></pre>CFG Guide: Runs best at CFG 1.0 by design. Small nudges up to 1.3–1.5 are possible for slightly tighter prompt adherence. Do not go above 1.5 — artifacts may appear.Negative prompts have limited effect at this distillation level. Use ConditioningZeroOut (included in the workflow) instead of writing a full negative prompt.<hr /><h2 id="z-anime-distill-4-step">🚀 Z-Anime Distill-4-Step</h2>The fastest Z-Anime variant. Built for maximum throughput — rapid prototyping, batch generation and situations where speed matters most.Recommended Settings:<pre><code>Steps: 4
CFG: 1.0 (max ~1.5)
Sampler: euler_ancestral
Scheduler: beta
Negative: limited effect
</code></pre>CFG Guide: At 4 steps the model has very little correction room. Stay at CFG 1.0 for the most stable results. Nudging up to 1.3–1.5 is possible but increases instability. Do not go above 1.5.Tips for 4-Step: Be specific and front-load the most important details early in your prompt. The optional upscaler (hires fix or SeedVR2) in the workflow is especially useful here to recover fine detail.<hr /><h2 id="resolution-guide">📐 Resolution Guide</h2>| Use Case | Resolution | |---|---| | ⭐ Portrait / Character art | 832 × 1216 | | Landscape / Scenes / Backgrounds | 1216 × 832 | | Square / General purpose | 1024 × 1024 | | Tall / Full body / Phone wallpaper | 768 × 1344 | | Cinematic / Wide scenes | 1920 × 1088 | | High quality / Detailed portraits | 1024 × 1536 |Supported range: 512 × 512 to 2048 × 2048, any aspect ratio. All resolutions run on 8 GB VRAM.<hr /><h2 id="prompting-guide">💡 Prompting Guide</h2>Natural language — not tag lists!<h3 id="good">✅ Good</h3><pre><code>A young anime girl with long silver hair and golden eyes, wearing a
traditional shrine maiden outfit with white haori and red hakama.
She stands in a sunlit bamboo forest, cherry blossoms falling softly
around her. Warm afternoon light filtering through the trees,
detailed fabric shading, expressive face, calm serene expression.
High quality anime illustration with fine line work.
</code></pre><h3 id="avoid">❌ Avoid</h3><pre><code>anime girl, silver hair, shrine maiden, bamboo, cherry blossom, warm light
</code></pre><h3 id="character-portraits">Character portraits</h3><pre><code>Detailed anime portrait of [character], soft rim lighting,
expressive eyes with detailed reflections, fine hair strands,
clean linework, professional anime illustration quality.
</code></pre><h3 id="action-scenes">Action scenes</h3><pre><code>Dynamic anime [scene], dramatic angle, motion energy, speed lines,
particle effects, cinematic composition, detailed shading,
high quality anime art.
</code></pre><h3 id="backgrounds-and-landscapes">Backgrounds &amp; landscapes</h3><pre><code>Anime [location] at [time of day], [lighting], [atmosphere],
Studio Ghibli inspired detail level, beautiful background art,
wallpaper quality.
</code></pre><hr /><h2 id="installation">🔧 Installation</h2>Step 1 — Download your version (BF16, FP8, GGUF or AIO) for the variant you want.Step 2 — Place the files:Standard BF16 / FP8 models:<pre><code>ComfyUI/models/diffusion_models/
├── z-anime-base-bf16.safetensors
├── z-anime-base-fp8.safetensors
├── z-anime-distill-8step-bf16.safetensors
├── z-anime-distill-8step-fp8.safetensors
├── z-anime-distill-4step-bf16.safetensors
└── z-anime-distill-4step-fp8.safetensors
</code></pre>GGUF variants:<pre><code>ComfyUI/models/unet/
├── z-anime-base-q8_0.gguf
└── z-anime-base-q4_k_s.gguf
</code></pre>Text Encoder &amp; VAE (for the non-AIO variants):<pre><code>ComfyUI/models/clip/
└── qwen_3_4b.safetensors

ComfyUI/models/vae/
└── ae.safetensors
</code></pre>AIO variants — single file, no extras needed:<pre><code>ComfyUI/models/checkpoints/
├── z-anime-base-aio-bf16.safetensors
├── z-anime-base-aio-fp8.safetensors
├── z-anime-distill-8step-aio-bf16.safetensors
├── z-anime-distill-8step-aio-fp8.safetensors
├── z-anime-distill-4step-aio-bf16.safetensors
└── z-anime-distill-4step-aio-fp8.safetensors
</code></pre>Step 3 — Load in ComfyUI:<ul><li>Use the Load Diffusion Model node for the model file, a CLIPLoader for the text encoder and a VAELoader for the VAE.</li><li>For the GGUF versions: load the GGUF model from the <code>models/unet/</code> folder, use the same CLIP and VAE files as above.</li><li>For the AIO versions: just use a standard Checkpoint Loader — no extra CLIP or VAE loading required.</li><li>Or use the official Z-Anime ComfyUI Workflow — it handles all variants and precisions with a built-in model switch.</li></ul><hr /><h2 id="custom-nodes-(for-the-official-workflow)">📦 Custom Nodes (for the official workflow)</h2><ul><li>rgthree-comfy</li><li>ComfyUI-Lora-Manager</li><li>ComfyUI-GGUF (only for the GGUF variants)</li><li>ComfyUI-SeedVR2_VideoUpscaler (optional, only for SeedVR2 upscale)</li></ul><hr /><h2 id="hugging-face-repo">🤗 Hugging Face Repo</h2>The complete model family is also mirrored on Hugging Face:🔗 <a target="_blank" rel="ugc" href="https://huggingface.co/SeeSee21/Z-Anime">huggingface.co/SeeSee21/Z-Anime</a>The HF repo additionally contains:<ul><li>The full Diffusers-format folder (<code>diffusers/</code>) — drop-in compatible with <code>ZImagePipeline.from_pretrained()</code> for Python users</li><li>An alternative Text Encoder by <a target="_blank" rel="ugc" href="https://huggingface.co/BennyDaBall/Qwen3-4b-Z-Image-Engineer-V4">BennyDaBall</a> — Engineer V4 (full fine-tune of the Z-Image text encoder with SMART training, drop-in compatible — often produces more varied outputs from the same seed)</li></ul><hr /><h2 id="version-history">📈 Version History</h2><h3 id="v1.0-initial-release">v1.0 — Initial Release</h3><ul><li>Z-Anime Base in BF16 &amp; FP8</li><li>Z-Anime Distill-8-Step in BF16 &amp; FP8</li><li>Z-Anime Distill-4-Step in BF16 &amp; FP8</li><li>GGUF Variants added:<ul><li>Z-Anime-Base-Q8_0 = pruned FP8 model (6.73 GB)</li><li>Z-Anime-Base-Q4_K_S = pruned Q4_K_S / NF4-style model (4.2 GB)</li></ul></li><li>AIO Variants added (all 6):<ul><li>z-anime-base-aio-bf16 / -fp8</li><li>z-anime-distill-8step-aio-bf16 / -fp8</li><li>z-anime-distill-4step-aio-bf16 / -fp8</li></ul></li><li>Official ComfyUI Workflow included — supports all variants</li><li>Hugging Face mirror with full Diffusers folder for Python users</li><li>Optimized for euler_ancestral + beta, simple practical use across the family</li></ul><hr /><h2 id="credits">🙏 Credits</h2><ul><li>Base Architecture: Tongyi Lab (Alibaba) — Z-Image</li><li>Fine-Tune: SeeSee21</li><li>License: Apache 2.0</li><li>Architecture: S3-DiT (Single-Stream Diffusion Transformer, 6B parameters)</li><li>Base Model: <a target="_blank" rel="ugc" href="https://huggingface.co/Tongyi-MAI/Z-Image">Tongyi-MAI/Z-Image</a></li><li>GitHub: <a target="_blank" rel="ugc" href="https://github.com/Tongyi-MAI/Z-Image">Tongyi-MAI/Z-Image</a></li><li>Engineer V4 Text Encoder (HF only): <a target="_blank" rel="ugc" href="https://huggingface.co/BennyDaBall/Qwen3-4b-Z-Image-Engineer-V4">BennyDaBall/Qwen3-4b-Z-Image-Engineer-V4</a></li></ul><hr />Z-Anime — Anime at its finest, powered by Z-Image Base. 🎌

Z-Anime

🧠 Text EncoderZ-Anime is built for the standard Z-Image Base text encoder pipeline in ComfyUI.Please use the matching text encoder provided for Z-Anime / Z-Image Base, together with its correct tokenizer. Using the wrong text encoder or tokenizer can lead to:black imagespure noisebroken prompt understandingerrors during generationTo make setup easier, I am uploading matching Text Encoder files in BF16 and FP8, so you do not have to search for the correct one yourself.

zAnime_textEncoder.safetensors

Text Encoder

Distill-8-Step-AIO

Base-GGUF

Distill-4-Step

Distill-8-Step

Base

Z-Image-Turbo is live for Generation! We're monitoring performance, and we'll be tweaking functionality behind the scenes over the next few days!Z-Image is a powerful and highly efficient image generation model with 6B parameters. It is currently has three variants:<ul><li>🚀 Z-Image-Turbo (this model) – A distilled version of Z-Image that matches or exceeds leading competitors with only 8 NFEs (Number of Function Evaluations). It offers ⚡️sub-second inference latency⚡️ on enterprise-grade H800 GPUs and fits comfortably within 16G VRAM consumer devices. It excels in photorealistic image generation, bilingual text rendering (English &amp; Chinese), and robust instruction adherence.</li><li>🧱 Z-Image-Base – The non-distilled foundation model. By releasing this checkpoint, we aim to unlock the full potential for community-driven fine-tuning and custom development.</li><li>✍️ Z-Image-Edit – A variant fine-tuned on Z-Image specifically for image editing tasks. It supports creative image-to-image generation with impressive instruction-following capabilities, allowing for precise edits based on natural language prompts.</li></ul>Original ComfyUI Models: <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/z_image_turbo/tree/main/split_files">https://huggingface.co/Comfy-Org/z_image_turbo/tree/main/split_files</a>Original HF Repo: <a target="_blank" rel="ugc" href="https://huggingface.co/Tongyi-MAI/Z-Image-Turbo">https://huggingface.co/Tongyi-MAI/Z-Image-Turbo</a>

Files with SHA256: 6c671498573ac2f7a5501502ccce8d2b08ea6ca2f661c458e708f36b36edfc5a