使用介紹
A LoRA-style model created by Akira Toriyama, the author of “Dragon Ball”.
Prompt: For Akira Toriyama, I recommend a weight of 0.5 or above.
I have uploaded both the 60-epoch and 40-epoch versions. Both can work according to the
model and the desired intensity.
I have also found that using CLIP skip 2 usually gives better results (except for the
first image, perhaps that was just lucky).
This was trained on Anything v4.5, so if you want accurate results, please use that
version or AbyssOrangeMix2. Other models may still work, but these are recommended.
One major difference between this and my other LoRAs is that this one basically requires
at least 1.5-2 times the size for high-resolution restoration.
Please remember that this is a LoRA, so you need to use it in this way. It is not a
standalone model or a TI.
How to use LoRA in auto1111:
1. Update the webui (use git pull or redownload it as shown here).
2. Copy the files to stable-diffusion-webui/models/lora.
3. Select your LoRA as shown in this video.
4. Make sure to change the weight (the default is :1, which is usually too high).
由《龍珠》的作者鳥山明創作的勞拉風格的模型。
提示:對于鳥山明,我建議權重爲0.5或以上。
60紀元和40紀元的版本我都上傳了。兩者都可以根據
型号和所需的強度。
我還發現使用CLIP skip 2通常會得到更好的結果(除了
第一張圖片,也許隻是運氣好)。
這是在任何4.5版本上訓練的,所以如果你想要準确的結果,請使用它
版本或AbyssOrangeMix2。其他型号可能仍然有效,但推薦使用這些型号。
這個和我其他的LoRAs的一個主要區别是這個基本上需要
至少是高分辨率恢複大小的1.5-2倍。
請記住,這是一個勞拉,所以你需要這樣使用它。它不是一個
獨立模型或層。
如何在auto1111中使用LoRA:
1.更新webui(使用git pull或重新下載它,如下所示)。
2.将文件複制到stable-diffusion-webui/models/lora。
3.如本視頻所示選擇您的LoRA。
4.确保改變權重(默認爲:1,通常過高)。
觸發詞 Trigger Words: | toriyama akira |
使用提示: | CLIP SKIP: 1 |
作品參照
tag:動漫,數字藝術,漫畫,LORA,勇者鬥惡龍,時辰觸發,龍珠,鳥山明,風格,藝術風格,插圖
Warning: A non-numeric value encountered in /var/www/html/wp-content/plugins/erphpdown/addon/tuan/index.php on line 209
評論0