PaNDeMoNiuM XL版本v1.03 (ID: 602746)

If you like what I do, leave me a ❤️ to receive notifications of future updates and follow me here on Civitai. Remember that it is a hobby for me and it costs a lot of time and everything is shared entirely for free, I appreciate that you make a review and share your creations on this wall.

---

PaNDeMoNiuM V2

- Understands and reproduces any prompt very well, no matter how complex and long it is, adding every detail.

- He excels in creating spectacular landscapes.

- Better hands and facial expressions.

- Better lights and shadows.

- Fewer steps needed to get great images.

- Reduced error rate.

Recommended Settings:

• Width/Height: 896x1152

• Sampler: DPM++ 2M SDE Karras (or the one you prefer)

• Steps: around 20

• CFG: around 5

• VAE: Already baked in.

• Negative Prompt: No negative prompt needed, add afterwards the things you don´t want to see in the image.

• HiRes: 4x_NMKD-Siax_200k with half the regular steps and 0.15 Denoise + x2 Upscale.

PaNDeMoNiuM V1.03

Pandemonium is born from a LoRa intended to be an update for Realism from HaDeS XL. The problem is that it made too many changes to the checkpoint, so I scrapped it as part of a future update. However, that LoRa had such quality and was so well trained, that I decided to train a new checkpoint from scratch, and then apply this LoRa to it. After several not very successful mixes, this is the result.

Despite being a first version, this checkpoint has a lot of quality. It is different from everything, completely far from Realism from HaDeS XL and any of my previous checkpoints.

It is very powerful and obedient, its generation rate of good images is very high. It makes very few mistakes without a negative prompt. It recognizes many styles very well and gets along excellently with the vast majority of LoRas. It also accepts simple and elaborate prompts.

Recommended Settings:

• Width/Height: 1024x1408

• Sampler: DPM++ 2M SDE Karras, DPM++ 3M SDE Karras, DPM++ SDE Karras, etc...

• Steps: around 30

• CFG: 3-10

• VAE: Already baked in.

• Negative Prompt: No negative prompt needed, add afterwards the things you don´t want to see in the image.

• HiRes: 4x_NMKD-Siax_200k with half the regular steps and 0.25 Denoise + 1.5 Upscale.

In short and to finish: I hope you like this model, that you try it, that you give it a lot of love and fill this wall with your images. It is very important to me, so I can improve it in the future if I see that it has your affection.

See you in the next update.

描述:

v1.03

训练词语: