general {

    base_exp_dir = exp/CASE_NAME/wmask

    tag = "train_retargeted_shadow_hand_seq_102_diffhand_model_"

    recording = [
        ./,
        ./models
    ]
}


dataset {
    data_dir = public_data/CASE_NAME/
    render_cameras_name = cameras_sphere.npz
    object_cameras_name = cameras_sphere.npz

    obj_idx = 102
}

train {
    learning_rate = 5e-4
    # learning_rate = 5e-6
    learning_rate_actions = 5e-6
    learning_rate_alpha = 0.05
    end_iter = 300000

    # batch_size = 128 #  64
    # batch_size = 4000
    # batch_size = 3096 #  64
    batch_size = 1024
    validate_resolution_level = 4
    warm_up_end = 5000
    anneal_end = 0
    use_white_bkgd = False

    # save_freq = 10000
    save_freq = 10000
    val_freq = 20 #  2500
    val_mesh_freq = 20 #  5000
    report_freq = 10
    ### igr weight ###
    igr_weight = 0.1
    mask_weight = 0.1
}
 
model {
    

    load_redmax_robot_actions = ""
    penetrating_depth_penalty = 0.0
    train_states = True

    minn_dist_threshold = 0.000
    obj_mass = 100.0
    obj_mass = 30.0

    # use_mano_hand_for_test = False
    use_mano_hand_for_test = True

    # train_residual_friction = False
    train_residual_friction = True

    # use_LBFGS = True
    use_LBFGS = False


    use_mano_hand_for_test = False
    train_residual_friction = True 
    
    extract_delta_mesh = False
    freeze_weights = True
    # gt_act_xs_def = True
    gt_act_xs_def = False
    use_bending_network =  True
    use_delta_bending = True

    use_passive_nets = True
    use_split_network = True

    n_timesteps = 60


    # using_delta_glb_trans = True
    using_delta_glb_trans = False
    # train_multi_seqs = True


    # optimize_with_intermediates = False
    optimize_with_intermediates = True


    loss_tangential_diff_coef = 1000
    loss_tangential_diff_coef = 0

    

    optimize_active_object = True

    no_friction_constraint = False

    optimize_glb_transformations = True


    sim_model_path = "DiffHand/assets/hand_sphere_only_hand_testt.xml"
    mano_sim_model_path = "rsc/mano/mano_mean_wcollision_scaled_scaled_0_9507_nroot.urdf"
    mano_mult_const_after_cent = 1.0
    sim_num_steps = 1000000

    bending_net_type = "active_force_field_v18"    


    ### try to train the residual friction ? ###
    train_residual_friction = True
    optimize_rules = True

    load_optimized_init_actions = ""


    use_optimizable_params = True


    ### grab train seq 224 ###
    penetration_determining = "sdf_of_canon"
    train_with_forces_to_active = False
    loss_scale_coef = 1000.0
    # penetration_proj_k_to_robot_friction = 40000000.0
    # penetration_proj_k_to_robot_friction = 100000000.0 # as friction coefs here #
    use_same_contact_spring_k = False


    # sim_model_path = "/home/xueyi/diffsim/DiffHand/assets/hand_sphere_only_hand_testt.xml"
    sim_model_path = "rsc/shadow_hand_description/shadowhand_new.urdf"


    optimize_rules = False

    penetration_determining = "sdf_of_canon"

    optimize_rules = False

    optim_sim_model_params_from_mano = False
    optimize_rules = False    


    penetration_proj_k_to_robot_friction = 100000000.0 # as friction coefs here # ## confs ##
    penetration_proj_k_to_robot = 40000000.0 #


    penetrating_depth_penalty = 1

    minn_dist_threshold_robot_to_obj = 0.0


    minn_dist_threshold_robot_to_obj = 0.1

    optim_sim_model_params_from_mano = False
    optimize_rules = False    

    optim_sim_model_params_from_mano = True
    optimize_rules = True    
    minn_dist_threshold_robot_to_obj = 0.0

    optim_sim_model_params_from_mano = False
    optimize_rules = False    
    minn_dist_threshold_robot_to_obj = 0.1


    ### kinematics confgs ###
    obj_sdf_fn = "data/grab/102/102_obj.npy" 
    kinematic_mano_gt_sv_fn = "data/grab/102/102_sv_dict.npy"
    scaled_obj_mesh_fn = "data/grab/102/102_obj.obj"
    # ckpt_fn = ""
    # load_optimized_init_transformations = ""
    optim_sim_model_params_from_mano = True
    optimize_rules = True    
    minn_dist_threshold_robot_to_obj = 0.0

    optim_sim_model_params_from_mano = False
    optimize_rules = False    
    optimize_rules = True

    ckpt_fn = "ckpts/grab/102/retargeted_shadow.pth"
    load_optimized_init_transformations = "ckpts/grab/102/retargeted_shadow.pth"
    optimize_rules = False

    optimize_rules = True
    
    ## opt roboto ##
    opt_robo_glb_trans = True
    opt_robo_glb_rot = False # opt rot # ## opt rot ##
    opt_robo_states = True
    

    use_multi_stages = False

    minn_dist_threshold_robot_to_obj = 0.1

    penetration_proj_k_to_robot = 40000
    penetration_proj_k_to_robot_friction = 100000

    drive_robot = "actions"
    opt_robo_glb_trans = False
    opt_robo_states = True
    opt_robo_glb_rot = False 

    train_with_forces_to_active = True



    load_redmax_robot_actions_fn  = ""


    optimize_rules = False
    


    loss_scale_coef = 1.0



    use_opt_rigid_translations=True

    train_def = True


    optimizable_rigid_translations=True
    
    nerf {
        D = 8,
        d_in = 4,
        d_in_view = 3,
        W = 256,
        multires = 10,
        multires_view = 4,
        output_ch = 4,
        skips=[4],
        use_viewdirs=True
    }

    sdf_network {
        d_out = 257,
        d_in = 3,
        d_hidden = 256,
        n_layers = 8,
        skip_in = [4],
        multires = 6,
        bias = 0.5,
        scale = 1.0,
        geometric_init = True,
        weight_norm = True,
    } 

    variance_network {
        init_val = 0.3
    }

    rendering_network {
        d_feature = 256,
        mode = idr,
        d_in = 9,
        d_out = 3,
        d_hidden = 256,
        n_layers = 4,
        weight_norm = True,
        multires_view = 4,
        squeeze_out = True,
    }

    neus_renderer {
        n_samples = 64,
        n_importance = 64,
        n_outside = 0,
        up_sample_steps = 4 ,
        perturb = 1.0,
    }

    bending_network {
        multires = 6,
        bending_latent_size = 32,
        d_in = 3,
        rigidity_hidden_dimensions = 64,
        rigidity_network_depth = 5,
        use_rigidity_network = False,
        bending_n_timesteps = 10,
    }
}