Model z własnych danych / zdjęć

Wykonaj następujące kroki, aby wygenerować swoje zdjęcia:

1. Znajdź kilka swoich zdjęć. Im wiecej tym lepiej.
2. Przejdź do https://tryleap.ai i uzyskaj KLUCZ API.
3. Uruchom kod w notatniku poniżej (najpierw prześlij swoje zdjęcia).

import requests
import json
import time

API_KEY = "YOUR API KEY GOES HERE"

HEADERS = {
    "accept": "application/json",
    "content-type": "application/json",
    "authorization": f"Bearer {API_KEY}"
}

IMAGES = [
    "URL to photo 1",
    "URL to photo 2",
    "URL to photo 3",
    "URL to photo 4",
    "URL to photo 5",
]

def create_model(title):
    url = "https://api.tryleap.ai/api/v1/images/models"

    payload = {
        "title": title,
        "subjectKeyword": "@me"
    }

    response = requests.post(url, json=payload, headers=HEADERS)

    model_id = json.loads(response.text)["id"]
    return model_id


def upload_image_samples(model_id):
    url = f"https://api.tryleap.ai/api/v1/images/models/{model_id}/samples/url"

    payload = {"images": IMAGES}
    response = requests.post(url, json=payload, headers=HEADERS)


def queue_training_job(model_id):
    url = f"https://api.tryleap.ai/api/v1/images/models/{model_id}/queue"
    response = requests.post(url, headers=HEADERS)
    data = json.loads(response.text)

    print(response.text)

    version_id = data["id"]
    status = data["status"]

    print(f"Version ID: {version_id}. Status: {status}")

    return version_id, status


def get_model_version(model_id, version_id):
    url = f"https://api.tryleap.ai/api/v1/images/models/{model_id}/versions/{version_id}"
    response = requests.get(url, headers=HEADERS)
    data = json.loads(response.text)

    version_id = data["id"]
    status = data["status"]

    print(f"Version ID: {version_id}. Status: {status}")

    return version_id, status


def generate_image(model_id, prompt):
    url = f"https://api.tryleap.ai/api/v1/images/models/{model_id}/inferences"

    payload = {
        "prompt": prompt,
        "steps": 50,
        "width": 512,
        "height": 512,
        "numberOfImages": 1,
        "seed": 4523184
    }

    response = requests.post(url, json=payload, headers=HEADERS)
    data = json.loads(response.text)

    inference_id = data["id"]
    status = data["status"]

    print(f"Inference ID: {inference_id}. Status: {status}")

    return inference_id, status


def get_inference_job(model_id, inference_id):
    url = f"https://api.tryleap.ai/api/v1/images/models/{model_id}/inferences/{inference_id}"

    response = requests.get(url, headers=HEADERS)
    data = json.loads(response.text)

    inference_id = data["id"]
    state = data["state"]
    image = None

    if len(data["images"]):
        image = data["images"][0]["uri"]

    print(f"Inference ID: {inference_id}. State: {state}")

    return inference_id, state, image

# Let's create a custom model so we can fine tune it.
model_id = create_model("Sample")

# We now upload the images to fine tune this model.
upload_image_samples(model_id)

# Now it's time to fine tune the model. Notice how I'm continuously 
# getting the status of the training job and waiting until it's
# finished before moving on.
version_id, status = queue_training_job(model_id)
while status != "finished":
    time.sleep(10)
    version_id, status = get_model_version(model_id, version_id)

# Now that we have a fine-tuned version of a model, we can
# generate images using it. Notice how I'm using '@me' to 
# indicate I want pictures similar to the ones we upload to 
# fine tune this model.
inference_id, status = generate_image(
    model_id, 
    prompt="A closeup photo of @me wearing a whool hat in front of a building"
)
while status != "finished":
    time.sleep(10)
    inference_id, status, image = get_inference_job(model_id, inference_id)

print(image)