from functools import partial import os import torch import numpy as np import gradio as gr import random import shutil print(f"Is CUDA available: {torch.cuda.is_available()}") print(f"CUDA device: {torch.cuda.get_device_name(torch.cuda.current_device())}") import os from os.path import join as pjoin import torch.nn.functional as F from models.mask_transformer.transformer import MaskTransformer, ResidualTransformer from models.vq.model import RVQVAE, LengthEstimator from options.hgdemo_option import EvalT2MOptions from utils.get_opt import get_opt from utils.fixseed import fixseed from visualization.joints2bvh import Joint2BVHConvertor from torch.distributions.categorical import Categorical from utils.motion_process import recover_from_ric from utils.plot_script import plot_3d_motion from utils.paramUtil import t2m_kinematic_chain from gen_t2m import load_vq_model, load_res_model, load_trans_model, load_len_estimator clip_version = 'ViT-B/32' WEBSITE = """

MoMask: Generative Masked Modeling of 3D Human Motions

Chuan Guo*Yuxuan Mu*Muhammad Gohar Javed*Sen WangLi Cheng

arXiv 2023

Description

πŸ”₯πŸ”₯πŸ”₯ This space presents an interactive demo for MoMask, a method for text-to-motion generation!!! It generates human motions (skeletal animations) based on your descriptions. To gain a better understanding of our work, you could try the provided examples first. πŸ”₯πŸ”₯πŸ”₯

πŸš€πŸš€πŸš€ In addition, we provide a link to download the generated human skeletal motion in BVH file format, compatible with CG software such as Blender!!! πŸš€πŸš€πŸš€

😁😁😁 If you find this demo interesting, we would appreciate your star on our github. More details could be found on our webpage. 🫢🫢🫢

If you have any issues on this space or feature requests, we warmly welcome you to contact us through our github repo or email.

""" WEBSITE_bottom = """

We thanks TMR for this cool space template.

""" EXAMPLES = [ "A person is running on a treadmill.", "The person takes 4 steps backwards.", "A person jumps up and then lands.", "The person was pushed but did not fall.", "The person does a salsa dance.", "A figure streches it hands and arms above its head.", "This person kicks with his right leg then jabs several times.", "A person stands for few seconds and picks up his arms and shakes them.", "A person walks in a clockwise circle and stops where he began.", "A man bends down and picks something up with his right hand.", "A person walks with a limp, their left leg gets injured.", "A person repeatedly blocks their face with their right arm.", "The person holds his left foot with his left hand, puts his right foot up and left hand up too.", "A person stands, crosses left leg in front of the right, lowering themselves until they are sitting, both hands on the floor before standing and uncrossing legs.", "The man walked forward, spun right on one foot and walked back to his original position.", "A man is walking forward then steps over an object then continues walking forward.", ] # Show closest text in the training # css to make videos look nice # var(--block-border-color); TODO CSS = """ .generate_video { position: relative; margin-left: auto; margin-right: auto; box-shadow: var(--block-shadow); border-width: var(--block-border-width); border-color: #000000; border-radius: var(--block-radius); background: var(--block-background-fill); width: 25%; line-height: var(--line-sm); } } """ DEFAULT_TEXT = "A person is " if not os.path.exists("checkpoints/t2m"): os.system("bash prepare/download_models_demo.sh") ########################## ######Preparing demo###### ########################## parser = EvalT2MOptions() opt = parser.parse() fixseed(opt.seed) opt.device = torch.device("cpu" if opt.gpu_id == -1 else "cuda:" + str(opt.gpu_id)) dim_pose = 263 root_dir = pjoin(opt.checkpoints_dir, opt.dataset_name, opt.name) model_dir = pjoin(root_dir, 'model') model_opt_path = pjoin(root_dir, 'opt.txt') model_opt = get_opt(model_opt_path, device=opt.device) ######Loading RVQ###### vq_opt_path = pjoin(opt.checkpoints_dir, opt.dataset_name, model_opt.vq_name, 'opt.txt') vq_opt = get_opt(vq_opt_path, device=opt.device) vq_opt.dim_pose = dim_pose vq_model, vq_opt = load_vq_model(vq_opt) model_opt.num_tokens = vq_opt.nb_code model_opt.num_quantizers = vq_opt.num_quantizers model_opt.code_dim = vq_opt.code_dim ######Loading R-Transformer###### res_opt_path = pjoin(opt.checkpoints_dir, opt.dataset_name, opt.res_name, 'opt.txt') res_opt = get_opt(res_opt_path, device=opt.device) res_model = load_res_model(res_opt, vq_opt, opt) assert res_opt.vq_name == model_opt.vq_name ######Loading M-Transformer###### t2m_transformer = load_trans_model(model_opt, opt, 'latest.tar') #####Loading Length Predictor##### length_estimator = load_len_estimator(model_opt) t2m_transformer.eval() vq_model.eval() res_model.eval() length_estimator.eval() res_model.to(opt.device) t2m_transformer.to(opt.device) vq_model.to(opt.device) length_estimator.to(opt.device) opt.nb_joints = 22 mean = np.load(pjoin(opt.checkpoints_dir, opt.dataset_name, model_opt.vq_name, 'meta', 'mean.npy')) std = np.load(pjoin(opt.checkpoints_dir, opt.dataset_name, model_opt.vq_name, 'meta', 'std.npy')) def inv_transform(data): return data * std + mean kinematic_chain = t2m_kinematic_chain converter = Joint2BVHConvertor() cached_dir = './cached' uid = 12138 animation_path = pjoin(cached_dir, f'{uid}') os.makedirs(animation_path, exist_ok=True) @torch.no_grad() def generate( text, uid, motion_length=0, use_ik=True, seed=10107, repeat_times=1, ): # fixseed(seed) print(text) prompt_list = [] length_list = [] est_length = False prompt_list.append(text) if motion_length == 0: est_length = True else: length_list.append(motion_length) if est_length: print("Since no motion length are specified, we will use estimated motion lengthes!!") text_embedding = t2m_transformer.encode_text(prompt_list) pred_dis = length_estimator(text_embedding) probs = F.softmax(pred_dis, dim=-1) # (b, ntoken) token_lens = Categorical(probs).sample() # (b, seqlen) else: token_lens = torch.LongTensor(length_list) // 4 token_lens = token_lens.to(opt.device).long() m_length = token_lens * 4 captions = prompt_list datas = [] for r in range(repeat_times): mids = t2m_transformer.generate(captions, token_lens, timesteps=opt.time_steps, cond_scale=opt.cond_scale, temperature=opt.temperature, topk_filter_thres=opt.topkr, gsample=opt.gumbel_sample) mids = res_model.generate(mids, captions, token_lens, temperature=1, cond_scale=5) pred_motions = vq_model.forward_decoder(mids) pred_motions = pred_motions.detach().cpu().numpy() data = inv_transform(pred_motions) ruid = random.randrange(99999) for k, (caption, joint_data) in enumerate(zip(captions, data)): animation_path = pjoin(cached_dir, f'{uid}') os.makedirs(animation_path, exist_ok=True) joint_data = joint_data[:m_length[k]] joint = recover_from_ric(torch.from_numpy(joint_data).float(), 22).numpy() bvh_path = pjoin(animation_path, "sample_repeat%d.bvh" % (r)) save_path = pjoin(animation_path, "sample_repeat%d_%d.mp4"%(r, ruid)) if use_ik: print("Using IK") _, joint = converter.convert(joint, filename=bvh_path, iterations=100) else: _, joint = converter.convert(joint, filename=bvh_path, iterations=100, foot_ik=False) plot_3d_motion(save_path, kinematic_chain, joint, title=caption, fps=20) np.save(pjoin(animation_path, "sample_repeat%d.npy"%(r)), joint) data_unit = { "url": pjoin(animation_path, "sample_repeat%d_%d.mp4"%(r, ruid)) } datas.append(data_unit) return datas # HTML component def get_video_html(data, video_id, width=700, height=700): url = data["url"] # class="wrap default svelte-gjihhp hide" #
# width="{width}" height="{height}" video_html = f"""

BVH Download

""" return video_html def generate_component(generate_function, text, motion_len='0', postprocess='IK'): if text == DEFAULT_TEXT or text == "" or text is None: return [None for _ in range(1)] # uid = random.randrange(99999) try: motion_len = max(0, min(int(float(motion_len) * 20), 196)) except: motion_len = 0 use_ik = postprocess == 'IK' datas = generate_function(text, uid, motion_len, use_ik) htmls = [get_video_html(data, idx) for idx, data in enumerate(datas)] return htmls # LOADING # DEMO theme = gr.themes.Default(primary_hue="blue", secondary_hue="gray") generate_and_show = partial(generate_component, generate) with gr.Blocks(css=CSS, theme=theme) as demo: gr.Markdown(WEBSITE) videos = [] with gr.Row(): with gr.Column(scale=3): text = gr.Textbox( show_label=True, label="Text prompt", value=DEFAULT_TEXT, ) with gr.Row(): with gr.Column(scale=1): motion_len = gr.Textbox( show_label=True, label="Motion length (<10s)", value=0, info="Specify the motion length; 0 to use the default auto-setting.", ) with gr.Column(scale=1): use_ik = gr.Radio( ["Raw", "IK"], label="Post-processing", value="IK", info="Use basic inverse kinematic (IK) for foot contact locking", ) gen_btn = gr.Button("Generate", variant="primary") clear = gr.Button("Clear", variant="secondary") gr.Markdown( f""" """ ) with gr.Column(scale=2): def generate_example(text): return generate_and_show(text) examples = gr.Examples( examples=[[x, None, None] for x in EXAMPLES], inputs=[text], examples_per_page=10, run_on_click=False, cache_examples=False, fn=generate_example, outputs=[], ) i = -1 # should indent for _ in range(1): with gr.Row(): for _ in range(1): i += 1 video = gr.HTML() videos.append(video) gr.Markdown(WEBSITE_bottom) # connect the examples to the output # a bit hacky examples.outputs = videos def load_example(example_id): processed_example = examples.non_none_processed_examples[example_id] return gr.utils.resolve_singleton(processed_example) examples.dataset.click( load_example, inputs=[examples.dataset], outputs=examples.inputs_with_examples, # type: ignore show_progress=False, postprocess=False, queue=False, ).then(fn=generate_example, inputs=examples.inputs, outputs=videos) gen_btn.click( fn=generate_and_show, inputs=[text, motion_len, use_ik], outputs=videos, ) text.submit( fn=generate_and_show, inputs=[text, motion_len, use_ik], outputs=videos, ) def clear_videos(): return [None for x in range(1)] + [DEFAULT_TEXT] clear.click(fn=clear_videos, outputs=videos + [text]) demo.launch(share=True)