Various Tools
In this chapter we're going to explore the tools and techniques that will take your creations to the next level.
There are many different tools we can use out there and i’ll try to go through the most important ones here!
Clip Interrogator:
Clip Interrogator is a tool that allows users to extract text from an image. Simply upload an image or screenshot, and Clip Interrogator will analyze it and provide a prompt with the extracted text.
You can access it here: https://replicate.com/methexis-inc/img2prompt
As an example, I put this image I made of a Darth Vader X-ray.
Here’s what I got:
a close up of a person wearing a darth vader helmet, digital art, shutterstock, digital art, anatomically accurate skeleton, hacking into the mainframe, hq 4k phone wallpaper, family photo
As you can’t see, it doesn’t work so well.
But, that’s not the point!
This tool is really useful to get some extra words / ideas / inspiration on how you can start building a prompt that might work.
You should use it as a “starting point tool”.
For example you can ask your client to send you a reference for what he wants, then run that reference through Interrogator and get a good idea on what it is exactly that he wants - in text form.
Have I Been Trained:
“Have I been Trained” is a website that allows you to see through the eyes of the algorithm.
https://huggingface.co/tasks/image-classification
When you type a word into the website's search bar, the search engine shows images that the algorithm has in it’s database.
It's like the website is showing you all the pictures the algorithm knows that match the word you used.
It's really useful to use different words and see what pictures come up.
You can even compare words to see which one makes the computer show more or better results.
And you can see if the computer knows about something specific by searching for it.
For example, try searching Van Gogh, and then try searching Vincent Van Gogh and look a the difference.
Real ESRGAN
Real ESRGAN is a tool that uses Generative Adversarial Networks (GANs) to upscale images. It is an external image upscaler, meaning it can be used to increase the resolution of an image beyond its original resolution. For example, a 512x512 image can be upscaled to 4k resolution. ESRGAN stands for Enhanced Super-Resolution Generative Adversarial Networks, and "Real" refers to the use of real-world images during the training process, which is said to produce more visually pleasing results compared to traditional methods.
Or basically:
“Real ESRGAN make image big.”
https://replicate.com/xinntao/realesrgan
For example, for one of the chapters in this tutorial I wanted to put the Logo of ChatGPT.
I only found a low resolution image online, but it didn’t matter to me.
Because all I had to do is throw that image into Real ESRGAN and upscale to 4K.
Now it looks amazing!
The List:
And here’s a constantly updating list of the most recent AI tools:
https://pharmapsychotic.com/tools.html
Van Gogh
Vincent Van Gogh
BEFORE
AFTER
