### 💬 Questions or setup issues?

Join the [Artfat AI Discord](https://discord.gg/r9tepyNSw) — I help with installation, workflows and prompting there, and it's where the community hangs out.

---

> ## ⚡ THE EASY WAY INTO SELFORA

> Selfism Simple is the stripped-down, beginner-friendly version of my AIO workflow. Load it, type your idea, hit Queue — a clean realistic photo in one pass. No Edit / ControlNet / Colorgrade clutter, just generation.

> 👉 Model: Selfora v.2 — Krea 2 Realistic Amateur Checkpoint

> Want the full toolkit (Edit, ControlNet, detailers, colorgrade)? Grab Selfism AIO

## v1.1 — update

Fix: reference image was wired into both prompter inputs.

Reference image 1 was connected to both image inputs on the Artfat LLM Prompter node. The

prompter was reading the same picture twice instead of treating the second slot as a separate

reference. Now wired correctly — one image per input.

Two-stage sampling.

The single sampler was replaced by a split schedule across two KSampler (Advanced) nodes:

- Stage 1 — res_2s / beta, steps 0-8, leftover noise passed on

- Stage 2 — deis_3m / bong_tangent, steps 8-10

CFG stays at 1.0. Nothing else in the workflow changed — the upscale toggle, the LoRA loader

and the GGUF switch all work exactly as they did.

Reminder — the upscale has two settings. scale_by 0.38 is the default and gives roughly

1.5x. Set it to 0.5 for a full 2x when the face is small or far from the camera, or when

you need the image for print. Leave the upscale off entirely if the base image already looks

the way you want — that stays the rawest and fastest option.

## Selfism Simple v.1 — Krea 2

The fast, no-fuss version of the Selfism pipeline for realistic amateur / phone-photo generation. Built around Selfora v.2, works with any Krea 2 checkpoint. Everything you need to make a great image, nothing you don't.

Solo project — the free stuff stays free ☕ Want new models early + a Discord role? Patreon's open, totally optional → https://www.patreon.com/c/ArtfatAI

## What's inside

- Artfat LLM Prompter — turns a short idea (or a reference image) into a full photographic prompt, on-device, no API. Dozens of ready selfie / amateur style presets.

- Clean single-pass generation — pre-dialed sampler and settings, just hit Queue.

- Optional pixel upscale (4xPurePhoto-RealPLSKR) on a simple ON/OFF toggle — default ~1.5x (scale_by 0.38), push to 2x (0.5) for small/far faces or print. Off = rawest and fastest; on = clean hi-res. If the base image already looks great, leave it off.

- Your own character LoRA — drop it in the LoRA loader and start your prompt with its trigger word to generate a consistent persona.

- Checkpoint or GGUF — a toggle lets you run the full safetensors model or a GGUF version for low VRAM.

## How to use

1. Load Selfora v.2 (or any Krea 2 checkpoint) — or toggle the GGUF loader for low VRAM.

2. (Optional) load your character LoRA and put its trigger word at the start of the prompt.

3. Type your idea in the LLM Prompter, or drop a reference image.

4. Hit Queue. Toggle the upscale on if you want extra resolution.

No trigger words needed for the base model. Sampler + settings pre-dialed.

💡 Quick tips: keep the shorter side around 1024 for best results, dimensions multiple of 16. For open eyes, write "relaxed wide-open eyes looking at the camera" — a hard laugh tends to squint the eyes. Negatives only work at CFG > 1 (ignored at CFG 1 turbo). Full setup + download links are in the "Model Links" note inside the workflow.

---

👉 Model: Selfora v.2 — Krea 2 Realistic Amateur Checkpoint

Want more? The full Selfism AIO adds Identity Edit, ControlNet, eye/hair detailers, Projector Tuner and Colorgrade.

### Required custom nodes

See the "Model Links" note inside the workflow — everything is listed with direct download links. Install missing ones via ComfyUI-Manager.

### Support

Solo project — the free stuff stays free ☕ Want new models early + a Discord role? Patreon's open, totally optional → https://www.patreon.com/c/ArtfatAI