Post

crispz-moss is a Fooocus-style web studio for MOSS-Image, the 5B open-weights model OpenMOSS (Fudan) released in October under Apache 2.0. It is the text-in-image and design studio of the crispz family: posters, logos, packaging, slides, covers, dense typography. Everything runs on your own GPU, nothing leaves the machine.

What makes it different from a prompt box:

• Layouts, not luck. You draw the text boxes on a canvas, type the exact words, and the model renders those words and nothing else. Nine design presets at their native sizes (A-series poster, logo, business card, book cover, 16:9 slide, packaging, thumbnail, 9:16 story, banner).
• Magic prompt and Describe, both local. A plain idea becomes a layout through Ollama in seconds, or through the model's own Qwen3.5-9B encoder in about a minute. Describe reads an existing image back into a layout, texts included.
• Drafts in 9 seconds. The 8-step DMD LoRA preset gives a clean poster with exact texts on an RTX 5090; the 50-step render takes 43 s at 1024.
• The rest of a studio. img2img, inpaint, outpaint and reframe (emulated), ESRGAN + refine upscale, LoRAs in every published format, editing with references, local vectorization to SVG, Remove BG, CivitAI search, X/Y/Z grids, a job queue, a CLI and the family JSON protocol.

Measured against Ideogram 4 on the same 24 posters: every requested text exact on both, and MOSS-Image did it faster, with no licence strings attached. The other crispz studios stay better at photos and editing: a hand-over sends your image to crispz-flux3 for native multi-reference edits.

Hardware: an NVIDIA GPU. 32 GB of VRAM is comfortable with the default auto offload; the nf4 encoder keeps everything resident at 21 GB. About 27 GB of weights download on the first render, from public Hugging Face repos: no licence to accept, no token.

Pinokio launcher: https://github.com/mikecastrodemaria/crispz-moss.pinokio
Source: https://github.com/mikecastrodemaria/crispz-moss

Install (Pinokio)

1. In Pinokio, open Discover and paste https://github.com/mikecastrodemaria/crispz-moss.pinokio.
2. Click Install: the launcher clones the app, creates a Python 3.11 venv, installs the requirements and PyTorch (CUDA 12.8 on NVIDIA), and checks that the Qwen3.5 encoder and the FLUX.2 VAE really load.
3. Click Start, then Open Web UI. Pick a design preset, write your texts in the layout editor, Generate.