gistfile1.txt

import rp
from google import genai
from google.genai.types import (
    GenerateContentConfig,
    Modality,
    ThinkingConfig,
    ImageConfig,
)
import os

os.environ["GOOGLE_CLOUD_PROJECT"] = "400060882616"
os.environ["GOOGLE_CLOUD_LOCATION"] = "global"
os.environ["GOOGLE_GENAI_USE_VERTEXAI"] = "True"

def nanobanana_i2i(input_image, prompt):
    client = genai.Client()
    response = client.models.generate_content(
        # model="gemini-2.5-flash-image-preview",
        # model="gemini-3-pro-preview",
        model="gemini-3-pro-image-preview",
        contents=[
            prompt,
            rp.as_pil_image(input_image),
        ],
        config=GenerateContentConfig(
            response_modalities=[Modality.TEXT, Modality.IMAGE],
            candidate_count=1,
            thinking_config=ThinkingConfig(
                # thinking_budget=0,
            ),
            safety_settings=[
                {"category": "HARM_CATEGORY_HARASSMENT", "threshold": "BLOCK_NONE"},
                {"category": "HARM_CATEGORY_HATE_SPEECH", "threshold": "BLOCK_NONE"},
                {"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT", "threshold": "BLOCK_NONE"},
                {"category": "HARM_CATEGORY_DANGEROUS_CONTENT", "threshold": "BLOCK_NONE"},
            ],
            image_config=ImageConfig(
                # aspect_ratio="16:9",
                image_size="4K",
            ),
        ),
    )

    [candidate] = response.candidates
    images = [
        rp.decode_image_from_bytes(part.as_image().image_bytes)
        for part in candidate.content.parts
        if part.inline_data
    ]
    [image]=images
    
    return image

if __name__ == "__main__":
    input_image_url = "https://people.com/thmb/oIWvqTiZiSegeOjfASchPsiuaMQ=/4000x0/filters:no_upscale():max_bytes(150000):strip_icc():focal(704x0:706x2)/kim-petras-3-1-2000-b7a11e2a0e534ed6b6f458c1d42f8025.jpg"
    input_image = rp.load_image(input_image_url, use_cache=True)
    prompt = "Give her a leather jacket and put her in the sunlight with the same pose"
    rp.tic()
    print("Starting nanobanana request...")
    output_image = nanobanana_i2i(input_image, prompt)
    print("Request took %.2f seconds" % rp.toc())
    output_image_path = rp.save_image(output_image)
    output_image_path = rp.get_absolute_path(output_image_path)
    print("Saved image:", output_image_path)
    rp.display_image(output_image)
    rp.web_copy_path(output_image_path)

video_path='/root/CleanCode/Github/ChronoEdit/outputs/mtfi2i_nongrid/prof/smoothwater.mp4'
video_path='/root/CleanCode/Experiments/Nanobananna_Edits/ENV_OCN_0010_FL-2dOcean-CrstMist-CamA_proxyMOV_v0008.mov'
video=load_video_via_decord(video_path,indices=resize_list(range(64),16))
video=load_video_via_decord(video_path,indices=16)
video=resize_images_to_fit(video,allow_growth=False,height=1024,width=1024)
grid_image=tiled_images(video,border_thickness=0)

prompt='Add tons of seafoam and spray, making the seas rougher - to each of the 16 panels, which are in chronological order from top left going right. Also make it daytime'
input_image=grid_image

rp.tic()
print("Starting nanobanana request...")
output_image = nanobanana_i2i(input_image, prompt)
print("Request took %.2f seconds" % rp.toc())
output_image_path = rp.save_image(output_image)
output_image_path = rp.get_absolute_path(output_image_path)
print("Saved image:", output_image_path)
rp.display_image(output_image)
rp.web_copy_path(output_image_path)