High quality CGMIX – V1
| Run Model | 使用GRAVITI Diffus 关于GRAVITI Diffus |
| Model Name | High quality CGMIX |
| Model Type | Checkpoint |
| Base Model | SD 1.5 |
| Version Name | V1 |
| File Name | highQualityCGMIX_v1.safetensors |
| SHA256 | 2a0b0c04d8 |
| NSFW | False |
| Trigger Words | |
| Tags | girl, 3d, photorealistic, realistic, video game |
| Metadata | fp = fp16 size = full format = SafeTensor |
| Size | 2.24 GB |
| Version Description | None |
| Source | https://civitai.com/models/22294 |
| Author | gfareast |
I want to create a model that can produce super-high quality CG, and I have this model after mixing some models and lora. This model is good at making cg characters that are very close to real people, and its texture is similar to the effect of high-quality character modeling in 3A games.
Unfortunately, its effect on making Asian faces is not very good, probably because most CG have no Asian faces.
The illustration uses a small number of lora models, all of which can be found in Station civitai, and you can search directly in the website.
In addition, VAE uses stable-780000.
You can get better results by using HD redrawing. The parameters are
Size: 512×640 Denoising strength: 0.6
Hires upscale: 1.5, Hires upscaler: ESRGAN_4x
我想创造一个能制作超高质量CG的模型，混合了一些模型和lora后有了这个模型。
这个模型擅长制作非常接近真人的CG人物，质感类似3A游戏中高质量人物建模的效果，遗憾的是它对亚裔面孔的制作效果不是很好，可能是由于大部分cg都没亚裔面孔。
例图用到了少量的lora模型，都在C站可以找到，你可以直接在网站内进行搜索。
另外VAE使用的是stable-780000。
使用高清重绘可以获得更好的效果，参数是
Size: 512×640 Denoising strength: 0.6
Hires upscale: 1.5, Hires upscaler: ESRGAN_4x
相关文章
