nerdyrodent / VQGAN-CLIP

Just playing with getting VQGAN+CLIP running locally, rather than having to use colab.
Other
2.59k stars 428 forks source link

About code implementation Feedback example #2

Closed zhanghongyong123456 closed 2 years ago

zhanghongyong123456 commented 2 years ago
  1. I particularly like this example, which is a great discovery. Can you use code to realize this example? I'm running under WIN, but I can't realize zoom.sh
  2. Is there any text prompt that can be generated automatically? I wonder if I can generate it myself , replace random.sh
nerdyrodent commented 2 years ago

Hi,

  1. One day I may indeed implement the image rotation in python... just as soon as I figure out how :) Ideas and tips welcome!
  2. Another way to generate random text would be to use something like GPT-2 or GPT-J-6B.
zhanghongyong123456 commented 2 years ago

Thank you very much for your prompt reply. I am wondering if it is possible to randomly generate some descriptive scenes.Thanks again for your code implementation. I have tried to achieve a similar effect to that achieved in Eleuther AI. My computer is 8G, so the image I get is about 400*400,I was wondering if the generated image needs to be super-resolution or 3D painting later

zhanghongyong123456 commented 2 years ago

Hi,

  1. One day I may indeed implement the image rotation in python... just as soon as I figure out how :) Ideas and tips welcome!
  2. Another way to generate random text would be to use something like GPT-2 or GPT-J-6B. Can the text prompt be typed in English only? Can other languages work, such as Chinese,How do I do it if I want to enter Chinese
nerdyrodent commented 2 years ago

Image post processing is entirely up to the artist! Feel free to upscale or do novel view synthesis as you feel. I've seen some good ones out there!

For other languages, I'd start with something like google translate. You could "pip install deep_translator" and generate English a bit like this: deep_translator --translator "google" --source "chinese" --target "english" --text "我现在可以用英文写文字了"

zhanghongyong123456 commented 2 years ago

Image post processing is entirely up to the artist! Feel free to upscale or do novel view synthesis as you feel. I've seen some good ones out there!

For other languages, I'd start with something like google translate. You could "pip install deep_translator" and generate English a bit like this: deep_translator --translator "google" --source "chinese" --target "english" --text "我现在可以用英文写文字了"

Wow, that's amazing,Suddenly feel your knowledge level must be very high,Know a lot of

ecmjohnson commented 2 years ago

@zhanghongyong123456 If you're still looking to run the zoom.sh example app on Windows, you can have a look at my .bat translation with the same functionality at https://gist.github.com/ecmjohnson/d41ada89c22ed84631ecdeb753ec9f04. Though I definitely think a Python implementation would be a better and more portable approach

zhanghongyong123456 commented 2 years ago

@zhanghongyong123456 If you're still looking to run the zoom.sh example app on Windows, you can have a look at my .bat translation with the same functionality at https://gist.github.com/ecmjohnson/d41ada89c22ed84631ecdeb753ec9f04. Though I definitely think a Python implementation would be a better and more portable approach

Thank you for providing the link, I think it will be a good choice

DrJKL commented 2 years ago
  1. One day I may indeed implement the image rotation in python... just as soon as I figure out how :) Ideas and tips welcome! Maybe something like this?
nerdyrodent commented 2 years ago

Pillow is an option, but I've been playing with a lot of these - https://imagemagick.org/script/convert.php Implode/explode is fun, as are quite a few others. Hence me considering Wand - https://docs.wand-py.org/en/0.6.6/

zhanghongyong123456 commented 2 years ago

Pillow is an option, but I've been playing with a lot of these - https://imagemagick.org/script/convert.php Implode/explode is fun, as are quite a few others. Hence me considering Wand - https://docs.wand-py.org/en/0.6.6/

That should be a good idea

nerdyrodent commented 2 years ago

There is a basic version in python now, e.g. python generate.py -p "a zooming movie" -zvid -i 7200 -zse 15 -se 7200 -vl 16

zhanghongyong123456 commented 2 years ago

There is a basic version in python now, e.g. python generate.py -p "a zooming movie" -zvid -i 7200 -zse 15 -se 7200 -vl 16

Thank you for the great project, I will try ,it should be very magical