AiARTiST-ImmersiveXL-ImmersionXL-LoRA

AiARTiST-Immersive XL - immersionXL

Local Filename: AiARTiST-Illusion of Immersion-0020.safetensors

----------------------------------------------------------------------

【This model is customized to match the UNIT-lib basic unit/CADS2.0 XL commercial advertising assistance system】

Dimension 64 / Training Set 400 / Epoch 20 / Pruning No / LR 2e-04 / STEP 8000

The base model used for training is: AiARTiST-CADS2.0 XL Commercial Advertising Assistance System(Enterprise Customization)

Direct Link: https://www.liblib.art/modelinfo/a56ebacdba7d4e30b97bb124bc3fc28f

Model List: https://www.liblib.art/search?keyword=AiARTiST

Open-source model UNIT-lib XL:

AiARTiST XL Basic Unit CADS2 LoRA Compatible Version

https://www.liblib.art/modelinfo/b10dfccc06f34dfa9031f1d070d846ee <<Direct link

UNIT-lib XL is open for member download, supporting fusion and online accelerated generation for members.

----------------------------------------------------------------------

Immersion

The experience of immersion is a deep emotional and sensory involvement that allows the audience to completely immerse themselves in the atmosphere of the music, as if surrounded by it, resonating strongly with the melody, rhythm, harmony, and emotions conveyed by the song.

On a quiet night, moonlight drapes like a veil over the sleeping city. The moment I put on my headphones, the world seems to mute itself, leaving only the melody flowing in my ears, leading me into a new dimension—the realm of music immersion.

Notes dance, like the brightest stars in the night sky, lighting up my heart one by one. The symphony unfolds gently, with strings delicate like silk, caressing every inch of my thoughts; brass instruments, with their grandeur, open the door of imagination, making me feel as though I have traveled to an ancient and mysterious palace, with splendid lights and endless corridors before me.

As the rhythm shifts, electronic synthesizer effects blend in, transporting me to a futuristic city, with neon lights flickering, skyscrapers rising high, each beat stepping to the rhythm of my heartbeat, my body swaying involuntarily with the music. At this moment, I am not just an observer but a participant in this auditory feast, resonating in sync with every note.

Then, a gentle piano solo begins, like a pair of gentle hands smoothing out all the noise and restlessness. Closing my eyes, I feel like standing by the seaside, with waves gently caressing the shore, bringing a refreshing breeze that calms my soul, providing an unprecedented sense of peace and release.

Music immersion is not just an auditory pleasure; it's a journey of the soul. It makes me forget about time and space, leaving only pure emotions and endless imagination. At the end of this journey, I return to reality with abundant inspiration and contentment, with the wonderful immersion planting a precious memory in my heart for countless reflections in the future.

(The above text is generated by AI)

----------------------------------------------------------------------

Trigger Words: AiARTiST, AIGCExperience

Common Parameters

AIGCExperience, Playing guitar, One man one baby, European, Warm indoor environment, Happy Experience Scene

Post-processing

Postprocess upscale by: 2, Postprocess upscaler: Lanczos, Postprocess upscaler 2: R-ESRGAN 4x+

----------------------------------------------------------------------

"Sugges* ***ng sample keywords directly, then add modifying detail words as needed"

Recommended Lora Weight: 0.5~0.8, Suggested Value 0.6

Recommended Sampling Steps: 25~30   CFG Scale: 5~8, Suggested Value 7

Recommended Sampling Method: DPM++ 2M Karras

High-definition resizing or image-to-image redraw: DPM++ 3M SDE Exponential

Hyper Sampling Steps: 8~12   CFG Scale: 1, Clip Skip 1

Hyper Sampling Method: Euler A

Recommended Resolution: Any resolution

Final piece lower than 1080x1920 or 960x1920 cannot fully reflect the base model's material accuracy

Can be divided into two steps, text-generated image long side 1024-1280, sending to image-to-image 0.4-0.5 for redraw and enlarge 1.5 times

Hyper Suggested Large Output to avoid structural errors can install kohya_hrfix plugin (Built-in in Forge)

FORGE Image-to-image interface supports secondary high-definition redraw, increasing visual information and enhancing art style

For other front ends, can send to image-to-image 0.4-0.5 for redraw and enlarge 1.2~1.5 times

If the image size is used for publication, can send to "Tiled Diffusion" or "Post-processing" for enlargement

Image post-processing high-definition setting: 8x_NMKD-Superscale_150000_G can be layered

                                       4x-UltraSharp or R-ESRGAN 4x+ Anime6B

----------------------------------------------------------------------

Hyper-SD is one of the latest diffusion model acceleration technologies, lossless high-definition acceleration, CN well adapted

Through testing with multiple base models, Hyper-SD can derive more image information than before acceleration!

Direct Link: https://www.liblib.art/modelinfo/a56ebacdba7d4e30b97bb124bc3fc28f

Strongly recommend ad design users to use the Hyper version, CADS2.0-Hyper XL supports enterprise customization

Be a tool for artists, contact:https://afdian.net/a/AiARTiST-lib +v zyuan980

----------------------------------------------------------------------

Parameter Configuration Applicable for CADS-HyperRAY XL:(Very Important)

Clip Skip 1 / CFG 1 / Steps 10 / Sampling Method (Sampler) Euler a / VAE Auto

Size: 1280x720 / 1536x768 / 960x1920 / 1920x960 / 2048x1024 / 1024x2048

For the output size exceeding 1536, it is recommended to open Kohya HRFix plugin to correct the composition error and enhance

Avoid negative words, no need to provide negative words, well adapted to LoRA and ControlNet.

----------------------------------------------------------------------

Afterword

As of 2024, there are still many people who do not fully understand XL, thinking that online image generation is slow and requires tagging for output.

In fact, these are misconceptions from outdated tutorials. Today, I tested online image generation at Lib, and the speed for XL and 1.5 outputs is almost the same.

Furthermore, thanks to the larger parameter quantity and dual text layers design, XL can already generate the majority of artistic styles based on the base model characteristics.

Even if some concepts are locked through LoRA, you can still modify the required visual elements using sample keywords, and Lib online also has an automatic translation tool.

SDXL is a form between commercial models and community models, requiring less LoRA combination, no complex sample keywords,

negative words, only needing your creativity and aesthetic taste. Even with image-to-image redraw, you can achieve very good results.

Additionally, I often hear voices saying that XL's ControlNet model is not user-friendly.

As of April 1, 2024, there are over 20 mature CN models in the SDXL ecosystem, with more constantly being added.

Soft and hard edges, Openpose, depth and normal maps, sketches, lines, doodles, various face swapping plugins, high-definition redraws, and more are all readily available.

For the pseudoshadow and image quality degradation issues common with XL's CN controller models, they can be optimized by slightly decreasing CFG and increasing steps,

If facing difficulty in strict control, my optimization plan is to initially output with a model similar to the base model plus CN for strict control,

then send to image-to-image for high-definition enlargement redraw with a factor of 0.4~0.6,

A major advantage of Lib online image generation is seamless model switching without frequent uninstallations from the GPU memory.

Lib has recently introduced XL Layer Diffusion transparent background images, Tile Diffusion redraw and enlargement, image to animation, and a series of popular tools

Avoid noise, reduce disk usage, member acceleration—this is why we always encourage everyone to become a member, to explore online creation.

----------------------------------------------------------------------

Model Usage Disclaimer:

1. You may not host this model and its derivative versions (such as fusion model versions) on websites/apps intended to earn revenue or donations.

2. You may not directly sell this model and its derivative versions (such as fusion model versions) unless you have sufficiently manually modified the model to be legally considered your personal work. Any legal consequences resulting from a violation of this clause are solely your responsibility, and I bear no liability.

3. You may not use this model to intentionally create or share illegal or harmful content for dissemination and output. Please uphold moral principles and use this model for positive and beneficial purposes.

----------------------------------------------------------------------

The base model used for training is: AiARTiST-CADS2.0 XL Commercial Advertising Assistance System(Enterprise Customization)

Direct Link: