Text to image (using google collab)

"The challenge is to live freely as though we will live forever, and gratefully, as though life could anytime." freely, gratefully = clouds, sky. Find this on twitter

A lot of people have been asking how I make these images. So I compiled this post so that you can make these images yourself. All you need is a Google account and some imagination. If you don't have a Gmail account, you will have to wait for me to compile a desktop version of this notebook. And if you don't have imagination there are always accounts like - Conceptbot 

About Collab

A great place to start is to go through this documentation about google collab.  For new users, I have one and only one tip, press the play button on the left side of the cell in sequence. 

This is my collab notebook, for the sake of this post I will be using this notebook for reference. 

** this notebook is converted from Spanish, so some of the variables are in Spanish. To avoid errors do not change them. 

"I can only experience the thinking mind clearly when I'm resting in awareness" This quote seems to have naturally evoked an image of Buddha in the image. Find this on twitter

this is a variation of the same quote.

Lets Diffuse 

'Diffuse' is a term that you will be using a lot. To summarize why,  the dataset we use is often trained with 1000s of images from the internet. So when we give a prompt to this notebook it will diffuse an image from the library(dataset) of images it was trained on. Hence the images ofter look like mushed up the goo. For now, we'll just call it art. 

The first cell mentions the credits and licenses. you don't really need to run this cell. 
Run the second cell, this will show you the computer that google allocated to you. You can see it on the top right of the notebook. 

On the left, if you click the folder icon you can access the drive you have been allocated. All your models will be set up here. 

By now you must have figured out how google collab works. Run all cells and let them execute till you reach Params. 

Textos  - add your text prompt here
ancho and alto - width and height 
modelo - leave it to default settings 
imagen_inicial - add the path of the image you want to wanna start your training from. I'd suggest something in high contrast.  
You can also specify stuff like, 'draw a cat anime style' or 'an elephant in the garden, as on art station'
Follow the instructions on the cells and watch the Ai paint.
When I practice being kinder to my self, I'm naturally kinder to others.

So these are some of my examples, you can find me on Twitter to see more of my AI experiments. 

Today, I'm just gonna be present with a kind and open mind. Anything else will just be a bonus. Find this on twitter

"I don't know why and how the mind does what it does." I like this image. mind = pink, brainish image
on twitter
"When I'm present, the mind is naturally calm and we experience quieter clarity and perspective." calm gives that impression of window/ gradient. Find this on twitter

A girl on the bus

Face in the flower

Concept bot's tweet - you can eat a banana with jellyfish in the ocean

Conversatiyou can eat a banana with jellyfish in the ocean

another wise wisdom by the concept bot. okay, drink tea on a mountain

cat in the house

another one by the concept bot - hmm, find a lonely bot on a plane

"My mind thinks its the best in the world, coincidently not so different from the mind that thinks it's the worst in the world." Sometimes it feels like the computer is being lazy, but it uses all of the ram to come up with this green scratched image. Find this on twitter

Hope this post was hopeful and it opens doors for you to explore AiArt. I'm looking forward to seeing what people come up with. @shashrvacai me to share your work. 


Popular Posts