ON V2
You need to use low CFG and 30 to 40 steps.
Some samplers will not give proper quality since the model has a lot of texture in the line work, some models inject unnecessary noise giving out a gritty image.
I am currently testing samplers since most don't have Res_Multistep, but it's highly recommended to use it with BETA scheduler. Karras, SGM_Uniform and others tend to place too much.
I will make a new version that is more flexible.

LCM CFG 1 and 10-15 steps is also a great choice




"Confound Atelier"

---

🎨 Introducing "Confound Atelier" – A professional-grade comic art model that merges the elegance of anime with the bold dynamism of Western comic artistry. Perfect for generating stunning, publication-ready character art with delicate faces and powerful presence.

---

## 📋 Model Details:

- Model Name: Confound Atelier

- Base Model: Illustrious

- Style: Comic Art + Anime Fusion

- Specialization: Curvy female characters, expressive faces, clean linework

- Best For: Character design, comic panels, illustrative art, and dynamic portraits

---

## ⚙️ Recommended Settings:

Fast & Efficient Workflow:

- CFG Scale: 1.0 - 2.5

- Steps: 10 - 15 (LCM compatible!)

- Sampler: DPM++ 2M Karras / Euler A

- Scheduler: Any (Karras recommended)

- Resolution: 1024x1024 or 832x1216 (for full-body shots)

Pro Tip: Pair with DMD LoRA for even faster generations and enhanced detail!

---

## 🛠️ Advanced Workflow (Optional):

For those who want to dive deeper, here’s my personal workflow:

1. First Pass KSampler (Base generation)

2. Latent Upscale (2x)

3. Second Pass KSampler (Refinement)

4. Face Detailing Node (For perfect expressions)

5. Final Upscale (For print-ready quality)

Note: This is optional – the model performs beautifully even with a basic workflow!

---

## 🎭 Prompting Tips:

- Use quality tags like *masterpiece, best quality, detailed lineart, dynamic pose, comic book style**

- Add specific artist inspirations (e.g., J. Scott Campbell, Artgerm, Adam Hughes) for stylistic nuance

- For delicate faces: *soft features, expressive eyes, elegant lips**

- For curvy figures: *voluptuous, athletic build, balanced proportions**

📌 My personal quality tags (paste your own here):

(Your curated tags go here – e.g., “detailed background, dramatic lighting, vibrant colors, sharp focus”)

---

## 🌟 Why Choose Confound Atelier?

- ✅ Optimized for speed (works great with LCM + low steps)

- ✅ Blends comic boldness with anime elegance

- ✅ High consistency across varied prompts

- ✅ Perfect for character artists and comic creators

---

## 🔗 Resources & Inspirations:

- Character Prompt Guide: [Link to my civitai article with image examples used with this model]

- DMD LoRA Recommendation: [DMD2 | 1 CFG SCALE | Fewer Steps]

- Example Workflow: Included in download (my workflow is usually saved in my metadata so downloading any of my images and pasting it in Comfyui will open up my workflow)

---

---

🎉 Final Notes:

This model is designed for artists who crave both speed and quality. Whether you’re generating concepts for a comic, designing original characters, or just creating beautiful art – Confound Atelier delivers.

Let’s see what you create! Post your generations with #ConfoundAtelier I will be gifting buzz to images I enjoy as usual! 🚀

I have been using a pattern on the names I use (most not all), if any one finds where I get the names (inspiration) with some context to it, I will gift 200buzz. A tip is it has to do with an online game from before 2010, it currently has a private server and is still being played.

---

Model crafted with precision. Generations powered by passion.

— Jetlistic