SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.14675 · ICCV · 2021

Synthesis of Compositional Animations from Textual Descriptions

Christian Theobalt, Philipp Slusallek, Cennet Oguz, Anindita Ghosh, Noshaba Cheema

arXiv · PDF · Open in the Atlas

Code that ran

2 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 0 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2103.14675.json · badge: /api/ran/2103.14675.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
anindita127/complextext2animation marked official in Papers with Code
Papers with Code archive
2 ran · 3 no recorded run
FunctionStatusWhere it lives
PoseEncoder Ran
ran on a synthesized input
anindita127/complextext2animation/src/model/model_hierarchical_twostream.py
code served (permissive licence) · get_code("2296e385cd876e4e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

class PoseEncoder(nn.Module):

    def __init__(self, h1, h2, h3, num_layers=1):
        super(PoseEncoder, self).__init__()
        self.layer1_rarm_enc = nn.Linear(9, h1)
        self.layer1_larm_enc = nn.Linear(9, h1)
        self.layer1_rleg_enc = nn.Linear(15, h1)
        self.layer1_lleg_enc = nn.Linear(15, h1)
        self.layer1_torso_enc = nn.Linear(16, h1)
        self.layer2_rarm_enc = nn.Linear(2 * h1, h2)
        self.layer2_larm_enc = nn.Linear(2 * h1, h2)
        self.layer2_rleg_enc = nn.Linear(2 * h1, h2)
        self.layer2_lleg_enc = nn.Linear(2 * h1, h2)
        self.batchnorm_up = nn.BatchNorm1d(2 * h2)
        self.batchnorm_lo = nn.BatchNorm1d(2 * h2)
        self.layer3_arm = nn.GRU(2 * h2, h3, num_layers=num_layers, batch_first=True)
        self.layer3_leg = nn.GRU(2 * h2, h3, num_layers=num_layers, batch_first=True)

    def forward(self, P_in):
        right_arm = P_in[..., 22:31]
        left_arm = P_in[..., 13:22]
        right_leg = P_in[..., 46:61]
        left_leg = P_in[..., 31:46]
        mid_body = torch.cat((P_in[..., :13], P_in[..., 61:64]), dim=-1)
        right_arm_layer1 = self.layer1_rarm_enc(right_arm)
        left_arm_layer1 = self.layer1_larm_enc(left_arm)
        mid_body_layer1 = self.layer1_torso_enc(mid_body)
        right_arm_layer2 = self.layer2_rarm_enc(torch.cat((right_arm_layer1, mid_body_layer1), dim=-1))
        left_arm_layer2 = self.layer2_larm_enc(torch.cat((left_arm_layer1, mid_body_layer1), dim=-1))
        upperbody = torch.cat((right_arm_layer2, left_arm_layer2), dim=-1)
        upperbody_bn = self.batchnorm_up(upperbody.permute(0, 2, 1)).permute(0, 2, 1)
        z_p_upper, h = self.layer3_arm(upperbody_bn.view(upperbody.shape[0], upperbody.shape[1], -1))
        right_leg_layer1 = self.layer1_rleg_enc(right_leg)
        left_leg_layer1 = self.layer1_lleg_enc(left_leg)
        right_leg_layer2 = self.layer2_rleg_enc(torch.cat((right_leg_layer1, mid_body_layer1), dim=-1))
        left_leg_layer2 = self.layer2_lleg_enc(torch.cat((left_leg_layer1, mid_body_layer1), dim=-1))
        lower_body = torch.cat((right_leg_layer2, left_leg_layer2), dim=-1)
        lower_body_bn = self.batchnorm_lo(lower_body.permute(0, 2, 1)).permute(0, 2, 1)
        z_p_lower, h = self.layer3_leg(lower_body_bn.view(lower_body.shape[0], lower_body.shape[1], -1))
        return (z_p_upper[:, -1, :], z_p_lower[:, -1, :])


INIT_ARGS = {
    'h1': 4,
    'h2': 5,
    'h3': 6,
    'num_layers': 2,
}

ARGS = {
    'P_in': torch.randn(2, 3, 64),
}

if __name__ == '__main__':
    model = PoseEncoder(**INIT_ARGS)
    model.eval()
    output = model(**ARGS)
    print("Output shapes:", (output[0].shape, output[1].shape))
    print("Success!")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
VelDecoderCell Ran
ran on a synthesized input
anindita127/complextext2animation/src/model/model_hierarchical_twostream.py
code served (permissive licence) · get_code("54465b3c341f286b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float, shape (2, 64) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

class VelDecoderCell(nn.Module):

    def __init__(self, h1, h2, h3):
        super(VelDecoderCell, self).__init__()
        self.layer1_rarm_dec = nn.Linear(h1, 9)
        self.layer1_larm_dec = nn.Linear(h1, 9)
        self.layer1_rleg_dec = nn.Linear(h1, 15)
        self.layer1_lleg_dec = nn.Linear(h1, 15)
        self.layer1_torso_dec = nn.Linear(h1, 16)
        self.layer2_rarm_dec = nn.Linear(h2, 2 * h1)
        self.layer2_larm_dec = nn.Linear(h2, 2 * h1)
        self.layer2_rleg_dec = nn.Linear(h2, 2 * h1)
        self.layer2_lleg_dec = nn.Linear(h2, 2 * h1)
        self.layer3_arm_dec = nn.Linear(h3, 2 * h2)
        self.layer3_leg_dec = nn.Linear(h3, 2 * h2)

    def forward(self, h_upper, h_lower):
        upper_body = self.layer3_arm_dec(h_upper)
        lower_body = self.layer3_leg_dec(h_lower)
        layer2_shape = int(upper_body.shape[-1] / 2)
        right_arm_layer2 = self.layer2_rarm_dec(upper_body[..., :layer2_shape])
        left_arm_layer2 = self.layer2_larm_dec(upper_body[..., -layer2_shape:])
        right_leg_layer2 = self.layer2_rleg_dec(lower_body[..., :layer2_shape])
        left_leg_layer2 = self.layer2_lleg_dec(lower_body[..., -layer2_shape:])
        layer1_shape = int(right_arm_layer2.shape[-1] / 2)
        right_arm_layer1 = self.layer1_rarm_dec(right_arm_layer2[..., :layer1_shape])
        torso_1_layer1 = self.layer1_torso_dec(right_arm_layer2[..., -layer1_shape:])
        left_arm_layer1 = self.layer1_larm_dec(left_arm_layer2[..., :layer1_shape])
        torso_2_layer1 = self.layer1_torso_dec(left_arm_layer2[..., -layer1_shape:])
        right_leg_layer1 = self.layer1_rleg_dec(right_leg_layer2[..., :layer1_shape])
        torso_3_layer1 = self.layer1_torso_dec(right_leg_layer2[..., -layer1_shape:])
        left_leg_layer1 = self.layer1_lleg_dec(left_leg_layer2[..., :layer1_shape])
        torso_4_layer1 = self.layer1_torso_dec(left_leg_layer2[..., -layer1_shape:])
        torso = (torso_1_layer1 + torso_2_layer1 + torso_3_layer1 + torso_4_layer1) / 4.0
        torso_trunk = torso[..., :13]
        torso_root = torso[..., -3:]
        pred = torch.cat((torso_trunk, left_arm_layer1, right_arm_layer1, left_leg_layer1, right_leg_layer1, torso_root), dim=-1)
        return pred


INIT_ARGS = {
    'h1': 4,
    'h2': 8,
    'h3': 16,
}

ARGS = {
    'h_upper': torch.randn(2, 16),
    'h_lower': torch.randn(2, 16),
}

if __name__ == '__main__':
    model = VelDecoderCell(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output: {output}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 64) — shape (2, 64) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
PoseGenerator Not yet run
unverified (harvested, no recorded run)
anindita127/complextext2animation/src/model/model_hierarchical_twostream.py
code served (permissive licence) · get_code("2b1c0c8de6b0bc47")
How it ran

no invocation recorded.

TeacherForcing Not yet run
unverified (harvested, no recorded run)
anindita127/complextext2animation/src/model/model_hierarchical_twostream.py
code served (permissive licence) · get_code("a64ba10b91371a0c")
How it ran

no invocation recorded.

VelDecoder Not yet run
unverified (harvested, no recorded run)
anindita127/complextext2animation/src/model/model_hierarchical_twostream.py
code served (permissive licence) · get_code("01b8c185a52eb16a")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 5 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

How can we animate 3D-characters from a movie script or move robots by simply telling them what we would like them to do?" "How unstructured and complex can we make a sentence and still generate plausible movements from it?" These are questions that need to be answered in the long-run, as the field is still in its infancy. Inspired by these problems, we present a new technique for generating compositional actions, which handles complex input sentences. Our output is a 3D pose sequence depicting the actions in the input sentence. We propose a hierarchical two-stream sequential model to explore a finer joint-level mapping between natural language sentences and 3D pose sequences corresponding to the given motion. We learn two manifold representations of the motion, one each for the upper body and the lower body movements. Our model can generate plausible pose sequences for short sentences describing single actions as well as long complex sentences describing multiple sequential and compositional actions. We evaluate our proposed model on the publicly available KIT Motion-Language Dataset containing 3D pose data with human-annotated sentences. Experimental results show that our model advances the state-of-the-art on textbased motion synthesis in objective evaluations by a margin of 50%. Qualitative evaluations based on a user study indicate that our synthesized motions are perceived to be the closest to the ground-truth motion captures for both short and compositional sentences.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2103.14675")
get_code_for_paper("2103.14675")
have("2103.14675")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2103.14675.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2103.14675.svg)](https://syntology.ai/paper/2103.14675)

Connect an agent — have() is free.