SDXL render

Statement: To prevent malicious theft by other platforms, the model will be available for download irregularly, but the images will continue to be processed.

Make friends, join AI mutual assistance group, QQ group number:482570609

The model has been merged into my large model, click here to download

→→→→→→Get more high-quality models←←←←←←

Comprehensive upgrade!

First of all, thank you for your support. In just a week, the online traffic exceeded 10K and the downloads exceeded 1K. This is probably the fastest-growing realistic optimization model so far, and this achievement would not have been possible without your support. Uncle is pleasantly surprised.

At the same time, this uncle also realizes the need to live up to the eager expectations of netizens, so some serious work needs to be done. So, what new breakthroughs are there in this update?

If you're not sure what SDXLrender is: It's a lora that enhances image quality. It learns from SDXL's rendering effects and applies them to SD1.5, allowing low-end computers to produce high-quality images.

SDXLrender_v2.0

1. Ultra-high generalization tailored for the Chinese: In response to feedback that SDXLrender's effects on some Asian models were not as good as on Western models, a special upgrade has been made. Now, models from any country, even 2.5D models, can render more delicate textures, with Chinese models being even more refined. Support for LCM acceleration, with some improvement in LCM image quality.

Test models: DreamShaper / majicMIX / epicphotogasm

2. Significantly improved clarity: This is the most significant upgrade in this version. Now you can generate clear images comparable to 1024 resolution at 512 resolution, producing higher-quality images with lower-end computer configurations, which is the original intention of SDXLrender.

Test models: majicMIX

3. Skin texture upgrade: Even in full-body shots, with Hires. fix applied, you can still feel the presence of skin texture, especially in close-up shots.

Test models: epicphotogasm

4. Contrast optimization: Some users requested an increase in contrast, so I immediately arranged it. I have made new optimizations to the highlights and shadows (after all, I started with low-lighting), click here if you're not sure what it's about).

Test models: Realistic Vision

5. Other optimizations: Removed the lens control feature, which was initially intended to enhance the controllability of SD1.5 scenes, but it seems to have caused great confusion for some people (maybe because they didn't understand how to use it). So I decided to remove this feature, even though I think it's very useful. Perhaps in the future I will release a separate lens control version, depending on whether there is a need.

Of course, removing the controller is not without benefits. Now, just by adding SDXLrender_v2.0 to the image you generate, you will immediately see a comprehensive improvement in image quality without worrying about disrupting the original composition and character design. Quite stable.

Test models: MagMix

For those using lora for the first time: Just add the lora to the forward prompt box, no need for trigger words or quality words, it's plug and play.

If you want to exchange and share, feel free to join the AI mutual assistance group, search QQ group number:482570609

Hope you have a great time playing!