Discussion on: NFT images generator using Python Jupyter Notebook

[Armen Minassian]

Hey I have a question, I'm trying to run this portion of code

TOTAL_IMAGES = 30 # Number of random unique images we want to generate

all_images = []

def create_new_image():

new_image = {} #

# For each trait category, select a random trait based on the weightings 
new_image ["Background"] = random.choices(background, background_weights)[0]
new_image ["Circle"] = random.choices(circle, circle_weights)[0]
new_image ["Square"] = random.choices(square, square_weights)[0]

if new_image in all_images:
    return create_new_image()
else:
    return new_image

Generate the unique combinations based on trait weightings

for i in range(TOTAL_IMAGES):

new_trait_image = create_new_image()

all_images.append(new_trait_image)

Except whenever this portion runs I get
an error saying RecursionError: maximum recursion depth exceeded in comparison. Any suggestions?

I hope you still check these forums.

[Victor Quan Lam]

May I know how many images you put into your background, circle and square folder? thanks

[Armen Minassian]

I just have two of each.

backgrounds (folder name)
-> blue.png
-> orange.png
circles
-> blue-circle.png
-> orange-circle.png
squares
-> blue-square.png
-> orange-square.png

[Victor Quan Lam]

Ok. Can you try to reduce the total images down to 8?

[Armen Minassian]

Reduce to 8? I only have 6 total images. blue.png, orange.png (background pics), two circles and two squares. Six total images.

[Armen Minassian]

Oh I see what you meant. I reduced it to 8 and It worked. What can I do to generate more images though? I mean the goal is generate a couple thousand images.

[Armen Minassian]

Could you explain the way the weights work?

[Victor Quan Lam]

You can add different colours and shapes (rectangle, star) to generate more images to get familiar. A good practice would be having different characters with a set of variation. For instance:

Furthermore, you can add more colours to create more images according to the following example.

[shape]_weights should be added up to 100 (optional) and the less the shape weights the less likely it will be populated.

\\ random.choices(your list, weights) is the method that returns a list with the randomly selected element from the specified sequence. [weights] is the possibility for each value.
new_image ["Circle"] = random.choices(circle, circle_weights)[0]

`

[Armen Minassian]

background = ["Blue", "Orange"]
background_weights = [30, 40]

circle = ["Blue", "Orange"]
circle_weights = [30, 15]

square = ["Blue","Orange"]
square_weights = [30, 15]

turtle = ["Blue","Orange"]
turtle_weights = [30,15]

For example I added a new attribute called turtle above ^. But when the composite is formed the result images are only of the turtle and it doesn't include any of the orange or blue backgrounds or squares and circles.

[Victor Quan Lam]

Can you try to add a new composite? It will let the processor know which combination of shapes you want to generate.

for item in all_images:

    im1 = Image.open(f'./layers/backgrounds/{background_files[item["Background"]]}.jpg').convert('RGBA')
    im2 = Image.open(f'./layers/circles/{circle_files[item["Circle"]]}.png').convert('RGBA')
    im3 = Image.open(f'./layers/squares/{square_files[item["Square"]]}.png').convert('RGBA')

// add this line of code
    im4 = Image.open(f'./layers/squares/{square_files[item["Turtle"]]}.png').convert('RGBA')

    #Create each composite
    com1 = Image.alpha_composite(im1, im2)
    com2 = Image.alpha_composite(com1, im3)

// add new combination 3
com3 = Image.alpha_composite(com2, im4 )

    #Convert to RGB
    rgb_im = com3.convert('RGB')

`

You can use for statement to handle this process better. something like this
`

for shape in shapeList:
    // generate rgb_img....

`