I have been trying to create smooth interpolated images between two images (they might be a bit rotated, translated and/or zoomed) and came up with this function

def interpolate_images(image1, image2, n_interpolations):
    # Convert to grayscale for feature detection
    gray1 = cv2.cvtColor(image1, cv2.COLOR_BGR2GRAY)
    gray2 = cv2.cvtColor(image2, cv2.COLOR_BGR2GRAY)

    # Detect ORB keypoints and descriptors
    orb = cv2.ORB_create()
    kp1, des1 = orb.detectAndCompute(gray1, None)
    kp2, des2 = orb.detectAndCompute(gray2, None)

    # Match features using brute-force matcher
    bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)
    matches = bf.match(des1, des2)
    matches = sorted(matches, key=lambda x: x.distance)

    # Extract location of good matches
    src_pts = np.float32([kp1[m.queryIdx].pt for m in matches]).reshape(-1, 1, 2)
    dst_pts = np.float32([kp2[m.trainIdx].pt for m in matches]).reshape(-1, 1, 2)

    # Find homography matrix
    M, mask = cv2.findHomography(dst_pts, src_pts, cv2.RANSAC, 5.0)

    # Identity matrix for the first image
    identity_matrix = np.eye(3)

    h, w = image1.shape[:2]

    # Interpolation between identity matrix and homography
    interpolated_images = []
    for i in range(n_interpolations + 1):
        alpha = i / n_interpolations
        # Interpolated transformation matrix
        interpolated_matrix = (1 - alpha) * identity_matrix + alpha * M

        # Warp the first image with the interpolated matrix
        warped_image1 = cv2.warpPerspective(image1, interpolated_matrix, (w, h))

        interpolated_images.append(warped_image1)

    return interpolated_images

However, when I run it with two very similar images like

and with the code

img1 = cv2.imread("image1.jpg")
img2 = cv2.imread("image2.jpg")

# Interpolate frames
n = 4
interpolated_frames = interpolate_images(img1, img2, n)

for index, image in enumerate(interpolated_frames):
    cv2.imwrite(f"interp_{index}.jpg", image)

cv2.imwrite(f"interp_{n+1}.jpg", img2)

there is still a rough cut between the last interpolated image (interp_4.jpg) and the actual second image (image2.jpg=interp_6.jpg).

How can this be fixed, so I get a smooth transition interpolation between the two images image1 and image2?

Ideally the interpolated images only show a combination of rotation, translation and zooming between the images image1 and image2 …

Also must not be OpenCV, could also be a different library …