animation_demo.py

import matplotlib
matplotlib.use('Agg')
import os, sys
import yaml
from argparse import ArgumentParser
from tqdm import tqdm
import copy

import imageio
import numpy as np
from skimage.transform import resize
from skimage import img_as_ubyte
import torch
import torch.nn.functional as F

from modules.generator import OcclusionAwareGenerator
from modules.keypoint_detector import KPDetector
from modules.tdmm_estimator import TDMMEstimator

from logger import Logger, Visualizer

from animate import normalize_kp
from scipy.spatial import ConvexHull


if sys.version_info[0] < 3:
    raise Exception("You must use Python 3 or higher. Recommended version is Python 3.7")

def load_checkpoints(config_path, checkpoint_path, cpu=False):

    with open(config_path) as f:
        config = yaml.load(f)

    generator = OcclusionAwareGenerator(**config['model_params']['generator_params'],
                                        **config['model_params']['common_params'])
    kp_detector = KPDetector(**config['model_params']['kp_detector_params'],
                             **config['model_params']['common_params'])
    tdmm = TDMMEstimator()

    if cpu:
        checkpoint = torch.load(checkpoint_path, map_location=torch.device('cpu'))
    else:
        checkpoint = torch.load(checkpoint_path)
        generator.cuda()
        kp_detector.cuda()
        tdmm.cuda()
 
    generator.load_state_dict(checkpoint['generator'])
    kp_detector.load_state_dict(checkpoint['kp_detector'])
    tdmm.load_state_dict(checkpoint['tdmm'])

    generator.eval()
    kp_detector.eval()
    tdmm.eval()

    return generator, kp_detector, tdmm


def make_animation(source_image, driving_video, 
                   generator, kp_detector, tdmm, with_eye=False,
                   relative=True, adapt_movement_scale=True, cpu=False):

    def batch_orth_proj(X, camera):
        camera = camera.clone().view(-1, 1, 3)
        X_trans = X[:, :, :2] + camera[:, :, 1:]
        X_trans = torch.cat([X_trans, X[:,:,2:]], 2)
        shape = X_trans.shape
        Xn = (camera[:, :, 0:1] * X_trans)
        return Xn

    with torch.no_grad():
        predictions = []
        visualizations = []
        source = torch.tensor(source_image[np.newaxis].astype(np.float32)).permute(0, 3, 1, 2)
        if not cpu:
            source = source.cuda()
        kp_source = kp_detector(source)
        source_codedict = tdmm.encode(source)
        source_verts, source_transformed_verts, _ = tdmm.decode_flame(source_codedict)
        source_albedo = tdmm.extract_texture(source, source_transformed_verts, with_eye=with_eye)

        driving = torch.tensor(np.array(driving_video)[np.newaxis].astype(np.float32)).permute(0, 4, 1, 2, 3)
        driving_initial = driving[:, :, 0].cuda()
        kp_driving_initial = kp_detector(driving[:, :, 0].cuda())
        driving_init_codedict = tdmm.encode(driving_initial)
        driving_init_verts, driving_init_transformed_verts, _ = tdmm.decode_flame(driving_init_codedict)

        for frame_idx in tqdm(range(driving.shape[2])):
            driving_frame = driving[:, :, frame_idx]
            if not cpu:
                driving_frame = driving_frame.cuda()

            kp_driving = kp_detector(driving_frame)
            driving_codedict = tdmm.encode(driving_frame)

            # calculate relative 3D motion in the code space
            if relative:
                delta_shape = source_codedict['shape'] + driving_codedict['shape'] - driving_init_codedict['shape']
                delta_exp = source_codedict['exp'] + driving_codedict['exp'] - driving_init_codedict['exp']
                delta_pose = source_codedict['pose'] + driving_codedict['pose'] - driving_init_codedict['pose']
            else:
                delta_shape = source_codedict['shape']
                delta_exp = driving_codedict['exp']
                delta_pose = driving_codedict['pose']

            delta_source_verts, _, _ = tdmm.flame(shape_params=delta_shape,
                                           expression_params=delta_exp,
                                           pose_params=delta_pose)

            if relative:
                delta_scale = source_codedict['cam'][:, 0:1] * driving_codedict['cam'][:, 0:1] / driving_init_codedict['cam'][:, 0:1]
                delta_trans = source_codedict['cam'][:, 1:] + driving_codedict['cam'][:, 1:] - driving_init_codedict['cam'][:, 1:]
            else:
                delta_scale = driving_codedict['cam'][:, 0:1]
                delta_trans = driving_codedict['cam'][:, 1:]

            delta_cam = torch.cat([delta_scale, delta_trans], dim=1)
            delta_source_transformed_verts = batch_orth_proj(delta_source_verts, delta_cam)
            delta_source_transformed_verts[:, :, 1:] = - delta_source_transformed_verts[:, :, 1:]

            render_ops = tdmm.render(source_transformed_verts, delta_source_transformed_verts, source_albedo)

            # calculate relative kp
            kp_norm = normalize_kp(kp_source=kp_source, kp_driving=kp_driving,
                                   kp_driving_initial=kp_driving_initial, use_relative_movement=relative,
                                   use_relative_jacobian=relative, adapt_movement_scale=adapt_movement_scale)
            out = generator(source, kp_source=kp_source, kp_driving=kp_norm, render_ops=render_ops,
                                        driving_features=driving_codedict)
            del out['sparse_deformed']
            out['kp_source'] = kp_source
            out['kp_driving'] = kp_driving
            visualization = Visualizer(kp_size=5, draw_border=True, colormap='gist_rainbow').visualize(source=source,
                                                                                driving=driving_frame, out=out)
            visualizations.append(visualization)

            predictions.append(np.transpose(out['prediction'].data.cpu().numpy(), [0, 2, 3, 1])[0])

    return predictions, visualizations

def find_best_frame(source, driving, cpu=False):
    import face_alignment

    def normalize_kp(kp):
        kp = kp - kp.mean(axis=0, keepdims=True)
        area = ConvexHull(kp[:, :2]).volume
        area = np.sqrt(area)
        kp[:, :2] = kp[:, :2] / area
        return kp

    fa = face_alignment.FaceAlignment(face_alignment.LandmarksType._2D, flip_input=True,
                                      device='cpu' if cpu else 'cuda')
    kp_source = fa.get_landmarks(255 * source)[0]
    kp_source = normalize_kp(kp_source)
    norm  = float('inf')
    frame_num = 0
    for i, image in tqdm(enumerate(driving)):
        kp_driving = fa.get_landmarks(255 * image)[0]
        kp_driving = normalize_kp(kp_driving)
        new_norm = (np.abs(kp_source - kp_driving) ** 2).sum()
        if new_norm < norm:
            norm = new_norm
            frame_num = i
    return frame_num


if __name__ == "__main__":
    parser = ArgumentParser()
    parser.add_argument("--config", required=True, help="path to config")
    parser.add_argument("--checkpoint", default=None, help="path to checkpoint to restore")

    parser.add_argument("--source_image_pth", default='', help="path to source image")
    parser.add_argument("--driving_video_pth", default='', help="path to driving video")
    parser.add_argument("--result_video_pth", default='result.mp4', help="path to output")
    parser.add_argument("--result_vis_video_pth", default='result_vis.mp4', help="path to output vis")
 
    parser.add_argument("--with_eye", action="store_true", help="use eye part for extracting texture")
    parser.add_argument("--relative", dest="relative", action="store_true", help="use relative or absolute keypoint coordinates")
    parser.add_argument("--adapt_scale", dest="adapt_scale", action="store_true", help="adapt movement scale based on convex hull of keypoints")

    parser.add_argument("--find_best_frame", dest="find_best_frame", action="store_true", 
                        help="Generate from the frame that is the most alligned with source. (Only for faces, requires face_aligment lib)")

    parser.add_argument("--best_frame", dest="best_frame", type=int, default=None,  
                        help="Set frame to start from.")
 
    parser.add_argument("--cpu", dest="cpu", action="store_true", help="cpu mode.")
 

    parser.set_defaults(relative=False)
    parser.set_defaults(adapt_scale=False)

    opt = parser.parse_args()

    source_image = imageio.imread(opt.source_image_pth)
    reader = imageio.get_reader(opt.driving_video_pth)
    fps = reader.get_meta_data()['fps']
    driving_video = []
    try:
        for im in reader:
            driving_video.append(im)
    except RuntimeError:
        pass
    reader.close()

    source_image = resize(source_image, (256, 256))[..., :3]
    driving_video = [resize(frame, (256, 256))[..., :3] for frame in driving_video]
    generator, kp_detector, tdmm = load_checkpoints(config_path=opt.config, checkpoint_path=opt.checkpoint, cpu=opt.cpu)

    if opt.find_best_frame or opt.best_frame is not None:
        i = opt.best_frame if opt.best_frame is not None else find_best_frame(source_image, driving_video, cpu=opt.cpu)
        print ("Best frame: " + str(i))
        driving_forward = driving_video[i:]
        driving_backward = driving_video[:(i+1)][::-1]
        predictions_forward, visualizations_forward = make_animation(source_image, driving_forward, 
                                                                     generator, kp_detector, tdmm, with_eye=opt.with_eye,
                                                                     relative=opt.relative, adapt_movement_scale=opt.adapt_scale, cpu=opt.cpu)
        predictions_backward, visualizations_backward = make_animation(source_image, driving_backward, 
                                                                       generator, kp_detector, tdmm, with_eye=opt.with_eye,
                                                                       relative=opt.relative, adapt_movement_scale=opt.adapt_scale, cpu=opt.cpu)
        predictions = predictions_backward[::-1] + predictions_forward[1:]
        visualizations = visualizations_backward[::-1] + visualizations_forward[1:]
    else:
        predictions, visualizations = make_animation(source_image, driving_video, 
                                    generator, kp_detector, tdmm, with_eye=opt.with_eye,
                                    relative=opt.relative, adapt_movement_scale=opt.adapt_scale, cpu=opt.cpu)

    out_name = os.path.basename(opt.source_image_pth).split('.')[0] + "_" + os.path.basename(opt.driving_video_pth).split('.')[0]
    imageio.mimsave(opt.result_video_pth, [img_as_ubyte(frame) for frame in predictions], fps=fps)
    imageio.mimsave(opt.result_vis_video_pth, visualizations)