<h1>🌙 LunaKrea</h1>
Warm light, clean lines, and a soft spot for the moon.
Lunakrea is a Krea 2 merge that leans toward luminous, storybook illustration with a semi-real finish. v2.0 keeps what made v1 work, the bright, warm palette, the clean anime detail and the baked-in TextFusion Refusal-Reduction LoRA, and blends in MoonMaster Narrative Drift, a Krea 2 Turbo checkpoint built to paint rather than photograph. The result leans further into digital painting: richer colour, stronger shapes, deeper shadows and more expressive light. It follows long, natural-language prompts closely without any extra LoRAs.
It loves a glowing light source (lanterns, neon, moonlight, oven fire), and it renders quoted text cleanly, so posters, shop signs and chalkboards are fair game.
No trigger words. Just describe the scene.
<hr />
<h2>🌒 The recipe</h2>
v2.0
<ol>
<li>Base (A): LunaKrea v1 at full precision (v1.0 and v1.1 share the same weights)</li>
<li>Blended in (B): <a rel="ugc" href="https://civarchive.com/models/2961766">🐻 MoonMaster Narrative Drift v1.0</a> by Moonbear_AIArt, with SLERP at t = 0.35 (roughly 65 % A / 35 % B)</li>
</ol>
LunaKrea v1, the A side:
<ol>
<li>Base (A): <a rel="ugc" href="https://civarchive.com/models/2813776">Krea CinnamonSaphire v2.0</a> by Doughboy645</li>
<li>Blended in (B): <a rel="ugc" href="https://civarchive.com/models/2938537">VXP Krea2 Beta 4 (Anime)</a> by VXP, via <a rel="ugc" href="https://civarchive.com/models/2968817">mikmod's INT8 release</a>, with SLERP at t = 0.35</li>
<li>Baked in: <a rel="ugc" href="https://civarchive.com/models/2775340">Krea2 TextFusion Refusal-Reduction LoRA v2.0 Full Rank</a> by Capitan01R, at strength 1.0</li>
</ol>
SLERP blends along the arc between two models instead of averaging in a straight line, so neither parent washes the other out. The refusal-reduction LoRA rides along inside LunaKrea v1, so in v2.0 the blend dilutes it to roughly two-thirds of its strength; the look comes from the merges, while the LoRA changes how the prompt is read. Merged and converted locally with SevenHarness and stable-diffusion.cpp.
<h2>🌓 Settings</h2>
<ul>
<li>Steps: 10 (8–12 works)</li>
<li>CFG: 1. It is a turbo-style model, so negative prompts have no effect at CFG 1.</li>
<li>Sampler / scheduler: ER SDE + simple</li>
<li>Resolution: 832×1216 portrait, 1216×832 landscape</li>
<li>Text encoder: the Krea 2 Qwen3-VL 4B encoder (the showcase used Ektome-Qwen3-VL-4Bi-PristinelyUncensored Q4_K_M)</li>
<li>VAE: <code>qwen_image_vae.safetensors</code></li>
</ul>
The text encoder and VAE are attached as optional components. Skip them if you already have Krea 2's.
<h2>🌔 Prompting</h2>
Write it the way you would describe a painting to a friend: the subject first, then what they are doing, where they are, how the scene is lit, and finally the style.
<ul>
<li>Full sentences beat tag lists. Weight syntax like <code>(word:1.2)</code> is read as literal text.</li>
<li>Put any text you want in the picture in double quotes and say where it goes: a chalkboard on the wall reads "LUNA BAKERY".</li>
<li>Lunakrea defaults to warm and bright. If you want dark and gritty, ask for it explicitly.</li>
</ul>
<h2>🌕 Which file?</h2>
<ul>
<li>BF16 · <code>lunakrea-v2.0-bf16.safetensors</code> · 25.6 GB: the full-precision master, for merging, training or making your own quants.</li>
<li>FP8 (scaled) · <code>lunakrea-v2.0-fp8_scaled.safetensors</code> · 13.5 GB: a lighter safetensors file for ComfyUI.</li>
<li>Q8_0 · <code>lunakrea-v2.0-q8_0.gguf</code> · 13.6 GB: near-lossless GGUF. Every v2.0 showcase image was rendered with this exact file.</li>
<li>Q5_K · <code>lunakrea-v2.0-q5_K.gguf</code> · 8.8 GB: for tighter VRAM, at a small quality cost.</li>
<li>Q4_K · <code>lunakrea-v2.0-q4_K.gguf</code> · 7.2 GB: the smallest file; some fine detail is lost.</li>
<li>INT8 · <code>lunakrea-v2.0-int8.safetensors</code> · 13.5 GB: for ComfyUI INT8 (convrot) workflows. It also runs natively in recent stable-diffusion.cpp builds.</li>
<li>W4A4 · <code>lunakrea-v2.0-int4_convrot.safetensors</code> · 9.4 GB: ComfyUI INT4 W4A4. Experimental, and it does not run in stable-diffusion.cpp yet.</li>
</ul>
Every format is converted directly from the BF16 master, never from another quant, with the same ComfyUI-safe converter as v1.1. GGUF files need ComfyUI-GGUF with Krea 2 support (such as molbal/ComfyUI-GGUF), Forge or stable-diffusion.cpp.
<h2>🌖 Versions</h2>
<ul>
<li>v2.0: LunaKrea v1 blended with MoonMaster Narrative Drift. More painterly, same warmth.</li>
<li>v1.1: the v1.0 weights, with every file rebuilt so it loads in ComfyUI.</li>
<li>v1.0 (broken): the first release. Its files stop in ComfyUI with "tuple index out of range"; use v1.1 or v2.0 instead.</li>
</ul>
<h2>🌗 Content note</h2>
LunaKrea's v1 parents are NSFW-capable and the refusal-reduction LoRA rides along, so LunaKrea handles mature content when you prompt for it. One v2.0 showcase image (the hot spring) contains nudity; everything else is SFW. All characters are adults. Please follow CivitAI's rules and the laws where you live.
<h2>⚖️ Permissions</h2>
v2.0 contains MoonMaster Narrative Drift, whose license asks that shared merges keep its permissions: generate with LunaKrea locally, on hardware you control, or on CivitAI, but please do not host it on third-party generation platforms such as TensorArt or SeaArt. Credit the creators when you share a merge of it.
<h2>🌘 Cover prompt</h2>
<blockquote>
A young woman with long silver hair and pale violet eyes stands on the tiled rooftop of an old Japanese inn at night, holding a glowing paper lantern in one hand. She wears an indigo kimono patterned with small white crescent moons and a loose obi that flutters in the wind. Behind her an enormous full moon fills half the sky, soft clouds drifting across it, and the warm orange lantern light mixes with the cool blue moonlight on her face. Full-body shot from a slightly low angle, detailed anime illustration with soft painterly shading and fine linework.
</blockquote>
10 steps · CFG 1 · ER SDE / simple · 832×1216 · seed 1818456205
Every showcase image keeps its full prompt and settings, so you can remix any of them. The v2.0 showcase reuses the v1.1 prompts on the same seeds (one prompt is replaced by a new LunaKrea poster), so you can compare the two versions image by image.
<h2>💜 Credits</h2>
Thank you to Moonbear_AIArt (MoonMaster Narrative Drift), Doughboy645 (CinnamonSaphire), VXP (VXP Krea2) and mikmod (its INT8 release), Capitan01R (TextFusion Refusal-Reduction), and the Krea team for Krea 2. Lunakrea would not exist without their work. 💜

LunaKrea

v2.0: Narrative Drift 🌙
<ul>
<li>A new blend: LunaKrea v1 (A) SLERP-merged with 🐻 MoonMaster Narrative Drift v1.0 by Moonbear_AIArt (B), t = 0.35, roughly 65 % LunaKrea / 35 % MoonMaster</li>
<li>What changes: MoonMaster is a Krea 2 Turbo checkpoint built to paint rather than photograph. LunaKrea keeps its warm, luminous palette and clean anime lines, and picks up richer colour, stronger shapes, deeper shadows and more expressive light</li>
<li>The TextFusion Refusal-Reduction LoRA comes along inside LunaKrea v1, diluted by the blend to roughly two-thirds of its strength</li>
<li>Seven formats, all converted from the BF16 master: BF16, FP8 (scaled), Q8_0, Q5_K, Q4_K, INT8 and W4A4 (INT4 convrot). They load in ComfyUI and stable-diffusion.cpp; GGUFs in ComfyUI need a ComfyUI-GGUF build that knows Krea 2, such as molbal/ComfyUI-GGUF</li>
<li>Optional components: the Qwen3-VL 4B text encoder (Q4_K_M) and the Qwen Image VAE</li>
<li>Showcase: the v1.1 prompts on the same seeds, so you can compare the versions image by image (one prompt replaced by a new LunaKrea poster), all rendered with the Q8_0 file</li>
<li>Permissions: MoonMaster's terms apply. Generate locally, on hardware you control or on CivitAI, not on third-party generation platforms</li>
<li>Recommended: 10 steps · CFG 1 · ER SDE / simple · 832×1216</li>
</ul>

Files with SHA256: 8bef8632f38c733a628e8faf244588995e75c30415b00524b3cb66dea4100e9a