[ AI Art ] - stable diffusion 討論 分享 01

有加，我試下加下simple background個weighting睇下得唔得先

駛撚咩，你而家落尖沙咀，D舖都執到十足廢墟咁XD

a room

midjourney?

SD

what is your workflow?

counterfeit v2.5

an underwater palace with a lot of lights and people might include a variety of brightly lit coral formations, a large underwater pavilion, and a variety of people swimming and floating in the water. The underwater palace may be illuminated by a variety of colored lights that add to its majesty and beauty. The people swimming and floating in the water may be having a great time, enjoying the spectacular scenery and the vastness of the ocean.

Negative prompt: EasyNegative
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 1508798110, Size: 768x1024, Model hash: a074b8864e

再inpaint到2046

thanks!

Steampunk style

有無人對TemporalNet有研究？搞好耐都唔明，但功能好勁！
https://twitter.com/Artoid_XYZ/status/1642103402326007808

想問吓用自己Train嘅LoRA gen 圖特別慢係咪正常
用down落嚟嘅LoRA 頂晒窿都係20秒
用自己嗰個差唔多要1分鐘

用咗41張圖加 invert 每張train 100 步左右
GPU 3060Ti

futuristic scifi

冇咁既野，同平時差唔多

civitai啲lora開始要俾錢先有正常版免費版專登俾啲垃圾你

邊到睇有收錢

有冇知點解我單獨gen一張相會好怪
但係xyz plot gen出來反而會靚D...

RuntimeError: Detected that PyTorch and torchvision were compiled with different CUDA versions. PyTorch has CUDA Version=11.7 and torchvision has CUDA Version=11.8. Please reinstall the torchvision that matches your PyTorch install.

我用colab係咁出依行野 要點解決

我local 開都有呢種問題

Colab值唔值得比錢用pro

己搵到解決方法
run sd 之前禁 左上角+code
入以下依行野

%pip install torch==1.13.1+cu117 torchvision==0.14.1+cu117 torchtext==0.14.1 torchaudio==0.13.1 torchdata==0.5.1 --extra-index-url https://download.pytorch.org/whl/cu117

然後run一次 之後先再run sd 就會update左個torchvision

有啲lora依家淨係會俾垃圾版你 (不過Preview張相就係正常版) 然後Description會話上面張圖唔係你down個version 你要撳條link課金先有得down

So far未見過咁樣
多數叫你請佢飲咖啡