How to drive bones like the last part of the video(https://youtu.be/J90VETgWIDg)? about rignet HOT 8 CLOSED

------------------ 原始邮件 ------------------ 发件人: "zhan-xu/RigNet" <[email protected]&gt;; 发送时间:&nbsp;2020年10月12日(星期一) 凌晨3:26 收件人:&nbsp;"zhan-xu/RigNet"<[email protected]&gt;; 抄送:&nbsp;"945984093"<[email protected]&gt;;"State change"<[email protected]&gt;; 主题:&nbsp;Re: [zhan-xu/RigNet] How to drive bones like the last part of the video(https://youtu.be/J90VETgWIDg)? (#14) OK. As stated in our paper, the evaluation of skinning weights is performed with ground-truth skeleton, so that the error from skeleton prediction doesn't influence this part. To evaluate skinning weights, you first run python run_skinning -e --resume="checkpoints/skinnet/model_best.pth.tar" --checkpoint="checkpoints/skinnet" --train_folder='DATASET_DIR/train/' --val_folder='DATASET_DIR/val/' --test_folder='DATASET_DIR/test/' --info_folder='DATASET_DIR/rig_info_remesh/' This will output the skinning results in results/skinnet folder. Then you can evaluate it with the following script `import os import numpy as np import open3d as o3d from utils.rig_parser import Info def eval_our_model(): info_folder = "/home/zhanxu/ModelResource_RigNetv1_preproccessed/rig_info_remesh/" obj_remesh_folder = "/home/zhanxu/ModelResource_RigNetv1_preproccessed/obj_remesh/" obj_fixed_folder = "/home/zhanxu/ModelResource_RigNetv1_preproccessed/obj/" output_folders = ["results/test/"] test_model_list = np.loadtxt('/home/zhanxu/ModelResource_RigNetv1_preproccessed/test_final.txt', dtype=np.int) for output_folder in output_folders: prec_total = [] rec_total = [] l1_dist_total = [] for model_id in test_model_list: #model_id = 783 #print(model_id) info = Info(os.path.join(info_folder, "{:d}.txt".format(model_id))) mesh_remesh = o3d.io.read_triangle_mesh(os.path.join(obj_remesh_folder, "{:d}.obj".format(model_id))) mesh_remesh_vert = np.asarray(mesh_remesh.vertices) mesh_fixed = o3d.io.read_triangle_mesh(os.path.join(obj_fixed_folder, "{:d}.obj".format(model_id))) mesh_fixed_vert = np.asarray(mesh_fixed.vertices) vertice_remesh_distance = np.sqrt(np.sum((mesh_remesh_vert[np.newaxis, ...] - mesh_fixed_vert[:, np.newaxis, :]) ** 2, axis=2)) vertice_raw_id = np.argmin(vertice_remesh_distance, axis=1) with open(os.path.join(output_folder, '{:d}_bone_names.txt'.format(model_id)), 'r') as fin: lines = fin.readlines() bone_names = [] for li in lines: bone_names.append(li.strip().split()) skin_pred_full = np.load(os.path.join(output_folder, "{:d}_full_pred.npy".format(model_id))) joint_names = list(set([bone_name[0] for bone_name in bone_names])) skin_pred_full_new = np.zeros((len(skin_pred_full), len(joint_names))) for i in range(len(joint_names)): joint_name = joint_names[i] for j in range(len(bone_names)): if bone_names[j][0] == joint_name: skin_pred_full_new[:, i] += skin_pred_full[:, j] skin_pred_full = skin_pred_full_new skin_gt_full = np.zeros_like(skin_pred_full) for v in range(len(info.joint_skin)): skin = info.joint_skin[v] for w in np.arange(1, len(skin), 2): col_id = joint_names.index(skin[w]) skin_gt_full[v, col_id] = skin[w+1] skin_pred_full = skin_pred_full[vertice_raw_id] skin_gt_full = skin_gt_full[vertice_raw_id] skin_pred_full = skin_pred_full[np.abs(np.sum(skin_gt_full, axis=1)-1) < 1e-2] skin_gt_full = skin_gt_full[np.abs(np.sum(skin_gt_full, axis=1) - 1) < 1e-2] precision = np.sum(np.logical_and(skin_pred_full&gt;0, skin_gt_full&gt;0)) / (np.sum(skin_pred_full&gt;0) + 1e-10) recall = np.sum(np.logical_and(skin_pred_full&gt;0, skin_gt_full&gt;0)) / (np.sum(skin_gt_full&gt;0) + 1e-10) mean_l1_dist = np.sum(np.abs(skin_pred_full - skin_gt_full)) / len(skin_pred_full) prec_total.append(precision) rec_total.append(recall) l1_dist_total.append(mean_l1_dist) prec_total = np.array(prec_total) rec_total = np.array(rec_total) l1_dist_total = np.array(l1_dist_total) print('result name: ', output_folder, 'precision: ', np.mean(prec_total), 'recall: ', np.mean(rec_total), 'avg_l1_dist: ', np.mean(l1_dist_total)) if name == 'main': eval_our_model()` — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or unsubscribe.

------------------&nbsp;原始邮件&nbsp;------------------ 发件人: "zhan-xu/RigNet" <[email protected]&gt;; 发送时间:&nbsp;2020年10月12日(星期一) 中午12:10 收件人:&nbsp;"zhan-xu/RigNet"<[email protected]&gt;; 抄送:&nbsp;"945984093"<[email protected]&gt;;"State change"<[email protected]&gt;; 主题:&nbsp;Re: [zhan-xu/RigNet] How to drive bones like the last part of the video(https://youtu.be/J90VETgWIDg)? (#14) For each model loaded into Maya, I randomly picked three joints (excluding the root joint), added a random rotation to these three joints, and saved the deformed mesh to disk. After this I compared the vertices displacement in python. So the poses are not the same. They are all randomly generated. Some functions you may be interested (used in mayapy): Get all joints as a dictionary: def getJointDict(root): joint_pos = {} this_level = [root] while this_level: next_level = [] for p_node in this_level: jpos = cmds.xform( p_node, query=True, translation=True, worldSpace=True ) joint_pos[p_node] = {} joint_pos[p_node]['pos'] = jpos joint_pos[p_node]['ch'] = [] ch_list = cmds.listRelatives(p_node, children=True, type='joint') pa_list = cmds.listRelatives(p_node, parent=True, type='joint') if ch_list is not None: joint_pos[p_node]['ch'] += ch_list next_level += ch_list if pa_list is not None: joint_pos[p_node]['pa'] = pa_list[0] else: joint_pos[p_node]['pa'] = 'None' this_level = next_level return joint_pos Add a rotation to a joint: cmds.rotate(angles[0], angles[1], angles[2], rot_joint, ws=True, r=True) — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or unsubscribe.