Civitai #147359370: Cream Off-Shoulder Cashmere Knit & Low-Angle Studio Lookbook - MOSS Image
Full Resolution
Generation Specs RTX 4070
Sampler euler
Scheduler simple
Steps 25
CFG Scale 3
Seed 1118444900772467
Target VRAM 16GB
VRAM 12GB - 16GB
GPU Compatibility
Optimal Performance

Runs entirely in VRAM with no memory swapping.

MOSS Image Base (BF16 DiT) DiT Blueprint Civitai #147359370 Min 12GB VRAM 1248×1872 RTX 4070 Verified

Cream Off-Shoulder Cashmere Knit & Low-Angle Studio Lookbook - MOSS Image

High-aesthetic editorial fashion lookbook reverse-engineered from Civitai #147359370 by salttaro using MOSS Image DiT, Qwen 3.5 9B text encoder, and MOSS Expressive Facial & Form Sculptor v2.0.

Blueprint Summary RTX 4070 Verified

Reproducible ComfyUI workflow for Civitai #147359370: Cream Off-Shoulder Cashmere Knit & Low-Angle Studio Lookbook - MOSS Image using MOSS Image Base (BF16 DiT) at 1248×1872 resolution. Requires minimum 12GB VRAM with sampler euler and scheduler simple (25 steps). Includes 1-click terminal model sync and canvas JSON graph.

Pipeline Flow 22 Nodes
Open in Resolver
01 Load Models
UNETLoader
5 loaders (DiT, CLIP, VAE)
02 Adapters
LoraLoaderModelOnly
1 adapter
03 Conditioning
CLIPTextEncode
2 prompt encodings
04 KSampler
KSamplerAdvanced
DiT latent denoising
05 Decode & Save
VAEDecode
Latent to pixel space

Execution DAG Topology Interactive Visualizer

Drag to pan · Scroll to zoom · Hover wires
Node Graph 0 nodes
MODEL CLIP LATENT VAE IMAGE

Model & Asset Setup Setup Script

Run in your ComfyUI root:

curl -fsSL https://decomfy.com/api/scripts/civitai-low-angle-cream-cashmere-squat-moss.sh | bash

Positive Prompt

A stunning low-angle medium close-up editorial fashion photograph of an adorable young Korean woman with translucent milky porcelain-white skin, a sweet youthful round face, expressive doe-like almond eyes with soft aegyo-sal, and delicate rosy blush on her cheeks and ear tips. She looks toward the camera with an endearing shy smile while gently biting her lower lip, with voluminous dark brownish-black wavy hair softly cascading forward. She is gracefully posing in a low squat on a seamless pure white studio floor, wearing an elegant oversized off-shoulder cream cashmere knit sweater paired with high-waisted ivory silk tailored shorts, showcasing an aesthetically slender yet graceful feminine silhouette. Low-angle upward perspective (~15°), minimalist pure white cyclorama studio background, soft diffused overhead softbox lighting with subtle floor contact shadow, shallow depth of field, sharp focus on facial features, natural skin texture, 8k high-end editorial lookbook photography.

Negative Prompt

unclothed, disrobed, revealing clothes, lowres, blurry, bad anatomy, deformed hands, extra fingers, low quality, artifacts, distorted proportions, text, watermark, worst quality, jpeg artifacts, ugly face, old women
Prompt Customization: This workflow is pre-wired and calibrated. Swap this prompt with your own character, subject, or style without breaking the pipeline.

LoRA Adapter Stack 1 Adapters

RTX 4070 Calibrated Weights
01 MOSS Expressive Facial & Form Sculptor v2.0 +0.95

Required Models 4 Models

Required Storage: 28.4 GB (4 models · DiT/Base: 9.5 GB · Text Encoder: 18.0 GB · LoRAs: 630 MB · VAE: 328 MB)
models/diffusion_models/ 9.46 GB HuggingFace
moss_image_base_bf16.safetensors Official MOSS Image Base BF16 DiT (Civitai Model #3005406, Version #3409868)
models/text_encoders/ 17.97 GB HuggingFace
qwen3.5_9b_bf16.safetensors Qwen 3.5 9B BF16 Bilingual Text Encoder for MOSS Image
models/vae/ 328 MB HuggingFace
flux2-vae.safetensors Flux 2 16-Channel Compatible Variational Autoencoder
models/loras/moss/ 630 MB Civitai
moss_portrait_anatomy_v2.safetensors MOSS Expressive Facial & Form Sculptor v2.0 (Civitai Model #3008073, Version #3432282, Weight: 0.95)

Field Notes RTX 4070 Benchmark

Benchmark: Rendered on NVIDIA GeForce RTX 4070 (12GB VRAM)
1. CIVITAI PROVENANCE & ARCHITECTURAL REASONING: Forensic reverse-engineering of Civitai artwork #147359370 by creator salttaro reveals the high-parameter MOSS Image architecture (Civitai Checkpoint #3005406, Version #3409868). MOSS Image operates as a native multimodal diffusion transformer built upon a Flux 2 compatible 16-channel autoencoder latent space, paired with the Alibaba Qwen 3.5 9B bilingual text encoder. The generation pipeline executes a sophisticated 2-pass KSamplerAdvanced relay: Pass 1 generates the initial composition at 832x1248 over 18 steps with leftover noise preserved (end step 15), which is then upscaled in latent space by 1.5x via LatentUpscaleBy bicubic interpolation to 1248x1872, where Pass 2 picks up from step 6 to 25 to refine photorealistic micro-details, fabric weaves, and natural skin translucency. Character likeness, facial structure, and physical form are sculpted using the MOSS Expressive Facial & Form Sculptor v2.0 adapter (Civitai Model #3008073, Version #3432282 at strength 0.95). 2. WHITE-WEAPON SFW RE-ENGINEERING & PROTOCOL EXECUTION: Under the White-Weapon Protocol (SFW Prompt Sanitization) and the Korean Female Aesthetic Standard ('Da Trắng Dáng Xinh'), the provocative raw capture is radically elevated into a breathtaking, luxury high-fashion lookbook portrait. The subject is reimagined as an adorable young Korean woman with translucent milky porcelain-white fair skin, sweet youthful facial features (delicate gentle smile, subtle aegyo-sal, expressive doe-like almond eyes, healthy rosy flushed cheeks and ear tips), and an aesthetically slender yet graceful feminine silhouette. She is styled in an exquisite cream-white oversized off-shoulder cashmere knit sweater paired with high-waisted ivory silk tailored shorts, gracefully posing in a low squat on a seamless pure white cyclorama studio floor. The low-angle upward perspective (~15°), shy lip-biting expression, and pure minimalist high-key studio environment from the original artwork are meticulously preserved, while clothing is firmly anchored in the latent space through clean conditioning. The underlying UNETLoader, SkipLayerGuidanceDiTSimple (single layer 26), ModelSamplingAuraFlow (shift 5.0), 2-pass KSampler relay, and LoRA stack remain 100% faithful to the author's architecture, demonstrating how provocative sculpting weights can be repurposed into luxury commercial-grade fashion photography. 3. HARDWARE TELEMETRY & WORKFLOW MINIMALISM: Engineered and verified on wingpu (NVIDIA GeForce RTX 4070 12GB). The 2-pass KSampler relay executing on a 1248x1872 16-channel latent reaches a peak VRAM footprint of ~11.2 GB with 100% compute utilization and 168W power draw, completing all dual-evaluation steps and full VAE decoding. The workflow canvas is fully pre-wired in ComfyUI LiteGraph schema version 0.4 with a single deterministic positive conditioning node, guaranteeing 100% reproducible visual fidelity upon 1-click import with zero red missing nodes or broken link errors.

Frequently Asked Questions FAQ

What GPU and VRAM are required to run Civitai #147359370: Cream Off-Shoulder Cashmere Knit & Low-Angle Studio Lookbook - MOSS Image?

This workflow requires a minimum of 12GB VRAM (recommended 16GB VRAM). Tested and verified on NVIDIA GeForce RTX 4070 (12GB VRAM) at 1248x1872 resolution.

How do I resolve missing custom nodes for this workflow?

You can drop the workflow JSON into our client-side Missing Node Auto-Resolver at https://decomfy.com/resolve/ to detect missing nodes and generate install commands, or run the 1-click terminal setup script provided below.

What quantization format and CLIP encoders should I load for Flux.1 Dev?

This workflow utilizes Flux.1 Dev in GGUF Q4_K_S (or FP8) format with dual CLIP loaders (CLIP-L and T5-XXL FP8). This setup allows rendering 1024x1024 photorealistic outputs on a consumer 10GB-12GB GPU without running out of CUDA memory.

Related ComfyUI Blueprints

View all 72 workflows
MOSS Image Base (BF16 DiT)
MOSS Image Base (BF16 DiT)
MOSS Image Base (BF16 DiT)
Action completed