🎨 Chroma-Anime-AIO

The plug-and-play anime checkpoint for Chroma — no external VAE or CLIP needed

Tired of juggling multiple files? This checkpoint combines Chroma-2K-QC, the Chroma-Anime-V3 LoRA, the FLUX VAE and the FLAN-T5-XXL text encoder in one file. Load it with the standard Load Checkpoint node and generate.

🚀 Quick start: put the file in ComfyUI/models/checkpoints/, load it with Load Checkpoint, use 16 – 26 steps and CFG 3.6 – 5, and add anime, to your prompt for a stronger anime look.

✨ At a glance

Property	Details
🧩 Type	All-in-one checkpoint (merge and packaging, no new base-model training)
🏗️ Base model	lodestones/Chroma (Chroma-2K-QC)
📄 File	Chroma-Anime-AIO-v1-FP8.safetensors (about 14.2 GB) → ComfyUI/models/checkpoints/
🎛️ Recommended	16 – 26 steps · CFG 3.6 – 5 · Euler Normal (any FLUX-compatible sampler works)
📐 Resolutions	512×512 up to 2K and beyond, depending on your VRAM
✍️ Prompting	Natural language, danbooru tags, or a mix of both
🧰 Custom nodes	None required

🖼️ Preview gallery

All images were posted to the v1-FP8 version on Civitai.

⭐ Why this checkpoint?

🎯 All-in-one	🚀 Fast	🎨 Flexible
One file, no separate VAE or CLIP	Based on Chroma-2K-QC: faster convergence, fewer steps	Natural language and tags both work
Works with the default ComfyUI workflow	Good results in 16 – 32 steps	Any FLUX-compatible sampler
Chroma-Anime-V3 style is pre-merged	512×512 up to 2K and more	Anime first, other styles with the right prompt

🧩 What is inside

Component	Details
Base model	Chroma-2K-QC by Lodestone Rock (faster convergence)
Style	Chroma-Anime-V3 LoRA, pre-merged at the optimal strength
VAE	FLUX VAE (integrated)
Text encoder	FLAN-T5-XXL (integrated)

❓ Why FLAN-T5-XXL?

FLAN-T5-XXL was fine-tuned on more than 1,000 additional tasks. This gives you better understanding of natural-language descriptions, more accurate reading of complex prompts and better detail compared with the standard T5-XXL v1.1.

🚀 Getting started

1. Download the checkpoint.
2. Place it in ComfyUI/models/checkpoints/.
3. Open ComfyUI and use the default workflow, or the companion workflow in workflows/.
4. Select the checkpoint in Load Checkpoint, enter your prompt and generate.

Traditional Chroma setup 😓	This checkpoint 🎉
Load Checkpoint → Load VAE → Load CLIP → Load LoRA	Load Checkpoint → Generate
Several nodes and files to manage	One node, done

A simple companion workflow is included in workflows/; see the Workflow section below.

🎛️ Recommended settings

Setting	Recommendation
Steps	16 – 26 (22 – 32 for more detail)
CFG scale	3.6 – 5 (try 4.2 – 5.0 for a stronger anime style)
Sampler	Euler Normal, Euler Beta or Simple; any FLUX-compatible sampler works
Clip skip	1
Trigger word	anime, (optional, for a stronger anime look)

💡 Pro tips

• More anime look? Add anime, to your prompt.
• More detail? Increase the steps to 22 – 32.
• Stronger style? Push the CFG above 4.0.
• Less anime? Leave out the trigger word; the base is versatile.

💬 Prompting guide

Natural language (recommended with FLAN-T5-XXL)

anime, 1girl, long flowing silver hair, detailed emerald eyes,
gentle smile, cherry blossoms falling, soft spring lighting,
elegant white dress, peaceful atmosphere

Tag based

anime, 1girl, silver hair, green eyes, smile, cherry blossoms,
spring, white dress, peaceful

Mixed (best of both)

anime, 1girl with long silver hair and emerald eyes, smile,
cherry blossoms, spring lighting, white dress, detailed, peaceful

🧩 Workflow

🎨 Chroma-Anime-AIO Simple Workflow

Civitai page · Download: simpel_workflow.zip · Chroma-Anime-AIO-Simple-Workflow.json

What it does: plug-and-play text-to-image with only the checkpoint loader. The VAE and text encoder come from the AIO checkpoint. Note nodes contain tips, settings and resolution presets.

Preset	Steps	CFG	Resolution
Fast / test	16	3.6	512×512 or 1024×1024
Balanced	16 – 20	4.2	1024×1024
High detail	22 – 26	4.2 – 5.0	1024×1024 or higher

The preset saved in the JSON is 16 steps · CFG 4.2 · euler / normal · 1024×1024. Tested on an RTX 4060 Ti with 8 GB VRAM.

Model file	Required?	Download	ComfyUI folder
Checkpoint (CheckpointLoaderSimple)	required	Chroma-Anime-AIO-v1-FP8.safetensors	models/checkpoints/

The JSON preselects Chroma-Anime-AIO-v1.safetensors. Select the downloaded Chroma-Anime-AIO-v1-FP8.safetensors in the loader.

🔌 Custom nodes: none. The workflow uses only ComfyUI core nodes (CheckpointLoaderSimple, CLIPTextEncode, EmptyLatentImage, KSampler, VAEDecode, SaveImage, Note).

🖼️ More results from the workflow page (9)

📄 Full workflow description from Civitai

My super simple plug-and-play workflow for Chroma-Anime-AIO that just works. It is perfect for beginners and for anyone who wants to get started quickly. No extra nodes, no VAE to load, no CLIP to configure: just load the checkpoint and start creating.

What is included

• Standard ComfyUI nodes only, no custom nodes required
• Helpful note nodes with tips and settings guides
• Pre-configured settings (16 steps, CFG 4.2, 1024×1024)
• Example prompts to get you started
• Resolution presets and tweaking tips

Quick start

1. Download Chroma-Anime-AIO-v1-FP8.safetensors and place it in ComfyUI/models/checkpoints/.
2. Load the workflow in ComfyUI.
3. Enter your prompt (tip: start with anime, for a stronger anime style).
4. Click Queue Prompt.

Prompting tips

• Add anime, at the start for a stronger anime look.
• Natural language works great: a girl with silver hair in a spring garden.
• Tags work too: 1girl, silver hair, green eyes, smile.
• Mix both: anime, 1girl with flowing silver hair, detailed emerald eyes, cherry blossoms.

Tested on an RTX 4060 Ti (8 GB VRAM). All resolutions up to 2K work smoothly, and it should also work on cards with 6 GB or more.

📊 Benchmarks and test results

All tests were run on 8 GB VRAM with the Euler Normal sampler.

Test	Resolution	Steps	CFG	Time
1 · Quick iteration	512×512	16	3.6	28.20 s
2 · Portrait	512×768	16	3.6	34.57 s
3 · Standard HD	1024×1024	16	4	68.21 s
4 · Llama mode 😅	832×1216	24	3.6	100.50 s
5 · Tall format	1024×1536	22	4.2	135.65 s
6 · Cinematic 2K	2048×1152	16	4.2	154.62 s

🔬 Test 1 · Quick iteration (512×512)

dark anime close-up portrait, elven woman warrior with glowing crimson eyes and black hair, pale skin illuminated by red firelight, expression fierce yet sorrowful, detailed molten cracks reflecting in her armor collar, gothic fantasy atmosphere, background of writhing shadow and ember sparks, inspired by Berserk and Claymore, high-contrast chiaroscuro lighting, ultra-realistic anime face, dramatic composition, detailed eyelashes and reflections, cinematic 4K gothic portrait.

Use case: rapid prototyping and style testing.

🔬 Test 2 · Portrait (512×768)

anime close-up portrait, dynamic lighting, elven swordswoman mid-battle, face glowing from flaming sword reflection, sharp cel shading like Fate/Stay Night or Demon Slayer, intense determination in her eyes, strands of black hair whipping across her face, glowing red-orange highlights from fire, cinematic anime realism, shallow depth of field, bokeh sparks behind, ultra-detailed 8K anime face, heroic expression, modern fantasy action style. Rendered in ultra high definition 8K resolution.

Use case: character portraits and vertical compositions.

🔬 Test 3 · Standard HD (1024×1024)

1024×1024 · 16 steps · CFG 4 · 68.21 s — prompt omitted on this page.

Use case: balanced quality and speed for general use.

🔬 Test 4 · Llama mode (832×1216)

anime, In the fiery depths of an erupting volcano, a solitary llama stands amidst smoldering ruin, silhouetted against molten lava and towering ash clouds. Its plush coat is immaculate despite the surroundings—jagged metal, cracked concrete, glowing embers. The llama's calm demeanor contrasts sharply with the chaos around it. Bright hues of flames and smoke frame the scene. Studio anime style emphasizes detailed backgrounds and expressions; vibrant colors enhance the dramatic setting.

Tip: for more detail, increase the steps. For a stronger anime look, use the trigger word anime, or push the CFG above 4.0.

🔬 Test 5 · Tall format (1024×1536)

anime, masterpiece, ultra-HD, high detail, depth of field, best quality, 8k, ergonomic composition, (anime coloring, anime screencap:1.5), official art, 1girl, Astrid Frost Valkyrie, white-blue flowing hair, ice-blue radiant eyes, majestic pose in a grand crystalline ice hall, intricate silver-blue armor adorned with frost patterns and glowing runes, snow particles suspended in the air, elegant and powerful expression, cinematic lighting, reflective icy floor, ethereal mist, glacial magic energy swirling around her, add microdetails, ultra-realistic anime rendering, masterpiece illustration, epic fantasy atmosphere.

Note: resolutions above 1 megapixel are fully supported.

🔬 Test 6 · Cinematic 2K (2048×1152)

anime, masterpiece, 2k resolution, ultra-detailed, depth of field, best quality, (anime coloring, anime screencap:1.5), official art style, 1girl, Princess Zelda from The Legend of Zelda, standing on an ancient stone balcony overlooking Hyrule at sunrise, long golden hair flowing in the wind, wearing elegant white and gold royal robes with light blue accents, glowing runes on fabric edges, soft light reflecting from the Master Sword resting nearby, ancient ruins and floating islands in the background, golden morning mist, warm cinematic lighting, anime fantasy atmosphere, calm expression full of determination, official art aesthetics, add microdetails, masterpiece illustration, ultra-realistic anime rendering, inspired by Tears of the Kingdom and Breath of the Wild.

Note: even 2K works smoothly; you are only limited by your VRAM.

❓ FAQ

Do I need separate VAE or CLIP files?

No. Everything is built in.

Can I use my own LoRAs on top?

Yes. This is a full checkpoint, so you can layer any LoRA on it.

Can I adjust the merged LoRA strength?

It is pre-merged at the optimal strength, but you can add further LoRAs with custom strengths.

Does it work with SDXL workflows?

No. This checkpoint uses the FLUX/Chroma architecture only.

Do I need custom nodes?

No. It works with the default ComfyUI nodes.

📦 Repository contents

Path	Contents
Chroma-Anime-AIO-v1-FP8.safetensors	All-in-one FP8 checkpoint (14,225,590,118 bytes)
workflows/	Simple ComfyUI workflow (JSON and original zip)
images/	Preview images

🔞 Content notice

The original Civitai page states that this checkpoint can generate both SFW and NSFW content, that the training data includes artistic nudity and adult themes, and that it is intended for users aged 18 and over. Please use it responsibly.

🙏 Credits

• Chroma-2K-QC by Lodestone Rock — optimized merge for faster convergence
• Chroma-Anime-V3 (my own LoRA) — 427 MB, trained on 154 diverse images
• FLUX VAE — industry-standard quality
• FLAN-T5-XXL by Google — fine-tuned on more than 1,000 tasks

Special thanks to Lodestone Rock for the Chroma base models, Fictional.ai for supporting open-source AI development, and the whole ComfyUI community.

📜 License & attribution

• This checkpoint is a merge and packaging by SeeSee (Hugging Face: SeeSee21, Civitai: SeeSeeLP) of the components listed above. No new base-model training was done.
• Original Chroma project: lodestones/Chroma, Apache-2.0. This repository uses Apache-2.0, consistent with the Chroma base.
• Integrated components keep their own upstream terms: google/flan-t5-xxl (Apache-2.0); FLUX VAE — see the original Black Forest Labs release terms.
• base_model in the metadata points to the original Chroma project. The direct base, Chroma-2K-QC, is linked on Civitai (models/1966367) and has no confirmed Hugging Face repository ID.

🔗 Related links

Chroma-2K-QC · Chroma-Anime-V3 LoRA · FLAN-T5-XXL · Original Chroma project · Workflow page

---

Enjoying this checkpoint? Please leave a review and share your work. Found an issue or have a suggestion? Let me know in the comments.

_{Version 1.0 · October 2025}

---

🧭 More from SeeSee — all models

🧩 Checkpoints & AIO

Model	Hugging Face	Civitai
AniSee	SeeSee21/AniSee	Model · Workflows
Z-Anime	SeeSee21/Z-Anime	Model
Z-Image-Turbo-AIO	SeeSee21/Z-Image-Turbo-AIO	Model · Workflow
ERNIE-AIO	SeeSee21/ERNIE-AIO	Model · Workflow
Chroma-Anime-AIO	SeeSee21/Chroma-Anime-AIO	Model · Workflow
FLUX.2-klein-4B-AIO	SeeSee21/FLUX.2-klein-4B-AIO	Model · Workflow
QWEN-Anime	SeeSee21/QWEN-Anime	Model · Workflow
SeeMoss-Genesis	SeeSee21/SeeMoss-Genesis	Model · Workflow
See-Krea2	SeeSee21/Krea-2-See-Krea2-Genesis	Model · Workflow

🎀 LoRAs

Model	Hugging Face	Civitai
Illustrious Anime Collection	SeeSee21/Illustrious-Anime-Collection	Model
Chroma Semi-Anime	SeeSee21/chroma-Semi-Anime	Model
Kitsune Furry Girl	SeeSee21/Kitsune-Furry-Girl-Z-Image	Model
Ayase Momo (Dandadan)	SeeSee21/Ayase-Momo-Z-Image	Model

🧩 Checkpoints & AIO  ·  🎀 LoRAs  ·  🌐 Civitai  ·  ▶️ YouTube

_{Made with ❤️ by SeeSee}