- Jul 23, 2020
- 877
- 2,651
Install a Stable Diffusion environment (I like ComfyUI:
You must be registered to see the links
)Download some models from civitai or huggingface and put them in your model folder.
Type in what you want to make in the text box and click generate.
- Jun 24, 2022
- 2,888
- 23,057
- Jul 6, 2021
- 414
- 6,415
- Jul 6, 2021
- 414
- 6,415
- Jun 24, 2022
- 2,888
- 23,057
- Jul 6, 2021
- 414
- 6,415
- Jul 6, 2021
- 414
- 6,415
- Oct 21, 2020
- 31
- 66
anybody can do some ai stuff with becca? It seems that I am too stupid to get some useful stuff out of it 
- Jan 5, 2022
- 167
- 3,301
- Jan 25, 2020
- 76
- 141
- Jan 25, 2020
- 76
- 141
- Jun 24, 2022
- 2,888
- 23,057
- Jan 25, 2020
- 76
- 141
- Jul 6, 2021
- 414
- 6,415
You don't have permission to view the spoiler content.
Log in or register now.
Been generating some ai art but I cant get the eyes to not look weird. My stuff looks nothing close to what you guys are doing
There are a few things to consider. There's a positive prompt for things you want to see and a negative prompt for things you don't want. In the positive prompt, you can add keywords to embellish your image, like best quality, 4k, sharp lineart, etc. In the negative prompt, you can add things like ugly, blurry, grainy, photographic, etc.
You can also download and apply negative embeddings, which is similar to a lora.
In the settings, you don't want to put the cfg scale too high. I put it at 4-5.
For the sampler, you can use a simple one like euler a. The DPM++ 2M Karras sampler is better for photography, but for this style of add it adds unneccessary texture detail that makes the image muddy.
You can also play with the weigh of the lora. A lower lora weight will have a better image quality, but the result will be farther from the lora's style.
If you have a local installation, you can do an image to image pass over a higher resolution version of your image. It'll make it much better.
- Jul 6, 2021
- 414
- 6,415
yeah i hate when people post things i much prefer 0 activity