Muse by Stable Yogi · Krea2 · V2.5 NVFP4
Muse by Stable Yogi · Krea2 · V2.5 — NVFP4
The NVFP4 (4-bit) build of Realism V1.0 — made for RTX 50-series / Blackwell cards.
Smallest & fastest — ~8 GB, native FP4 tensor cores. Lowest VRAM, top speed on 50-series.
Same realism — real skin, natural light, true-to-life — no quality loss from 4-bit.
Runs in ComfyUI + the latest Forge (Blackwell only — on older cards use the INT8 / fp8 / GGUF versions instead).
⚙️ Settings: 8 steps · CFG 1 · Euler · Simple · native 1024²+
(Model only — pair with the Qwen3-VL text encoder + Qwen-Image VAE, same as base Krea 2.)
Made by Stable Yogi , get Pro release on my portal.
Muse by Stable Yogi · Krea2 · V2.5 EXTENDED
🟢 MUSE V2.5 EXTENDED
Your settings, your look — Muse V2.5 with room to move.
Most turbo models pin you to one setting: change the steps or the guidance and the image falls apart. Extended doesn't. It holds together across a real working range, so you can tune the result instead of accepting it.
⚙️ Settings
Recommended: 16 steps · CFG 2 · Euler · Simple · native 1024²+
Works across: 8–20 steps · CFG 1–3
⚠️ Don't go above CFG 3.
✨ Why you'd want it
Dial in your own look. Lower CFG for soft and natural, higher for punch and contrast. Fewer steps for speed, more for polish. Your call, not the model's.
Extra steps actually pay off. At 16 steps you get the richest version of Muse — deeper light, warmer tone, more presence.
It follows your prompt harder. More guidance headroom means when you ask for a specific scene, you get that scene.
Still quick when you're in a hurry. 8 steps works fine for drafts and batch runs.
🎯 Which one should I use?
Best for Muse V2.5 (standard)Speed. Sharpest at 8 steps — the everyday default.
Muse V2.5 Extended
Control and quality ceiling. Give it 16 steps, CFG 2.
📦 Download — INT8-ConvRot (fast INT8 tensor cores, loads natively in Forge + ComfyUI,
Made by Stable Yogi 🧘 — drop a ❤️ and show me what you tune it to. Looking for a Pro Version check here on my portal.
=========================================================
Muse by Stable Yogi · Krea 2 · V2.5
The full-spectrum edition of Muse on Krea 2 — same photoreal turbo, a wider, richer blend.
✨ What's NEW in V2.5
- A broader, more varied blend — more range across faces, skin, and lighting moods.
- True photorealism — real skin texture, pores, freckles, natural light (not AI-glossy).
- Believable faces, consistent — holds up shot to shot.
- Turbo — full images in 8 steps.
🔧 V2.0 vs V2.5
- Companion release to V2.0 — we put out both so you can pick your favourite. V2.0 is the tighter, more focused cut; V2.5 is the fuller blend with a touch more variety. Same engine, same speed.
📦 Downloads
- INT8-ConvRot — fast (INT8 tensor cores), loads natively in Forge + ComfyUI.
- Q8 GGUF — near-lossless. · Q4 GGUF — low VRAM / small download.
- (Model only — load the *Qwen3-VL text encoder + Qwen-Image VAE** alongside, same as base Krea 2; they auto-download.)*
⚙️ Settings
- 8 steps · CFG 1 · Euler · Simple · native 1024–2048.
⭐ Want more punch? Grab the PRO edition from my portal, dials detail and depth up further.
Made by Stable Yogi — drop a ❤️ and tell us: V2.0 or V2.5?
=========================================================
Muse by Stable Yogi — Krea2 · V2.0 EDIT (INT8)
For details of this release, please read the "About this model" section
==========================================================
Muse by Stable Yogi · Krea2 · V2.0 (INT8)
The next evolution of Muse on Krea 2 — more real, more detailed, same turbo speed.
✨ What's NEW in V2.0
A real jump in photorealism — skin now has genuine texture: pores, freckles, fine detail (less "AI-smooth").
Shot-on-camera look — believable faces, natural lived-in complexion.
Richer, more dimensional lighting — window light, soft shadows, real depth.
Holds up across prompts — lifelike skin and faces, consistent shot to shot.
🔧 What CHANGED (vs V1.5)
Built on the V1.5 engine, evolved with fresh photoreal character detail.
Cleaner, more natural rendering — over-polished feel gone.
Still turbo — full images in 8 steps.
📦 Downloads — INT8-ConvRot (Forge + ComfyUI native) · Q4 GGUF (low VRAM). (Load Qwen3-VL TE + Qwen-Image VAE alongside — auto-download.)
⚙️ Settings — 8 steps · CFG 1 · Euler · Simple · native 1024–2048.
⭐ PRO edition on my portal dials it up further.
===========================================================
Muse by Stable Yogi — Krea 2 · V1.5
Photoreal Krea 2 — fast, versatile, built to look genuinely real.
✨ What it is
Photorealistic model on Krea 2, tuned by Stable Yogi · turbo (8 steps) · plug-and-play with the Krea 2 TE + VAE.
🎯 Great for
Photoreal portraits & full-body
Any character
Fashion & editorial
Lifestyle & scenes
better lighting
Strong prompt-following
Natural skin & faces
📦 Downloads — Q8 GGUF (recommended) · Q4 GGUF (low VRAM) · fp8 (compat) (load TE + VAE alongside)
⚙️ 8 steps · CFG 1 · Euler · Simple · native res
⭐ Want more punch? The PRO edition on my portal is dialed up further.
===========================================================
Muse by Stable Yogi — Krea 2 · V1 Experimental
A photorealistic model on Krea 2, tuned for lifelike portraits, natural skin & light — at turbo speed. V1 Experimental: an early release, feedback welcome.
✨ Photoreal portraits & characters · 8-step turbo · native high-res · versatile (portrait / editorial / lifestyle / concept) · plug-and-play (TE + VAE built in).
⚙️ 8 steps · CFG 1 · Euler · Simple · native 1024–2048.
📦 Fp8 (full quality) · Q8_0 / Q4_0 GGUF (low-VRAM).
Looking for a Pro version get it free at my portal
Drop a ❤️ and share your creations.
To run this model on Forge UI, look here https://civarchive.com/articles/32010/run-krea-2-in-forge-free-extension-first-on-forge
Description
Muse by Stable Yogi — Krea 2 · V2.0 EDIT (INT8)
Instruction-based image editing baked into the Muse V2.0 checkpoint — no LoRA loader needed.
What's in it
Muse V2.0 (photoreal Krea 2 engine) + Krea 2 Identity Edit LoRA v1.1 merged in at full strength, tested, then re-quantized to INT8-ConvRot.
ComfyUI
Node pack:
https://github.com/lbouaraba/comfyui-krea2editSkip the LoRA loader — it's already baked in. Feed this checkpoint straight into
Krea2EditModelPatch.Free workflow JSON:
https://aistudynow.com/the-step-by-step-krea-2-edit-comfyui-workflow-with-free-json/Blur fix: Wan 2.1 VAE 2x upscaler (spacepxl) via ComfyUI-VAE-Utils.
TE + VAE: Qwen3-VL + Qwen-Image VAE (Comfy-Org/Krea-2).
Settings
8–12 steps · CFG 1.0 (3.0 for removals) · Euler · Simple
grounding_px384–768 — lower = stronger edit, higher = stronger likeness≤2 MP (~1–1.5 MP for two-person edits)
Credit — built on the Krea 2 Identity Edit LoRA v1.1 by conradlocke. Krea 2 Community License.
Made by Stable Yogi 🧘
FAQ
Comments (12)
Superb high quality model for real! Right off the bat the outputs are very and i mean VERY detailed and almost unreal to distinguish from reality. Man what a banging model!!
Ha, appreciate the energy, thank you. The detail was the thing I really leaned on with this one, so glad it's hitting for you. There's more builds on my portal if you wanna keep going. Cheers 🙏
this is a great finetune. thanks for making it!
Ah, thanks, that's kind of you. Really glad it's landing for you. There's more of these on my portal if you feel like digging in. Cheers 🙏
can we get an fp8 of v2.0 because for me int8 is literally 4x slower and uses more vram
Sure, will post it soon. Thanks
Fp8 is uploaded.
int8 is litterially the best there is on modern common GPUs, might be that yours is just ancient
@Lumiiii that is just factually not true lmao, on older cards that dont natively support fp8 then int8 is better, modern cards like the rtx50 series and rx9000 series ai accelerators excel in fp8 compute, the performance you will see between the two formats is between 0 and int8 being worse, i suspect why the int8 version in this case being 4x slower is due to it using significantly more vram for me causing it to overflow out of the 16gb vram and into ram whereas fp8 comfortably sits within 16gb, the only benefit to using int8 on modern gpu's is that its a bit more power efficient.
@Lumiiii The new GPU provides several times higher image quality with FP8 compared to INT8.
Wow, thanks for the quick update!! Looks fantastic, great color, lighting, cute faces, nice prompt adherence... Love it!!
Ah, thank you, really glad the update landed well for you. Prompt adherence was the thing I most wanted to tighten this round, so good to hear it's showing. There's more on my portal if you feel like exploring. Cheers 🙏