SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.08656 · 2024

TC-Bench: Benchmarking Temporal Compositionality in Text-to-Video and Image-to-Video Generation

arXiv · PDF · Open in the Atlas

Code that ran

5 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 0 have no contract verdict; 5 failures of our instrument; 0 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/2406.08656.json · badge: /api/ran/2406.08656.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
weixi-feng/tc-bench canonical (classifier label)
classifier basis: census
5 ran · 0 no recorded run
FunctionStatusWhere it lives
encode_image Ran
ran; our contract draft was wrong, not the code
weixi-feng/tc-bench/vlm_eval.py
code served (permissive licence) · get_code("3d39217a41c77766")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. 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 io
import numpy as np
import base64
from PIL import Image

# Create a valid PIL Image
pil_image = Image.new('RGB', (4, 3), color=(255, 0, 0))

ARGS = {
    'image': pil_image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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: none by the harness; any seed is in the setup block
encode_image_from_path Ran
ran; our contract draft was wrong, not the code
weixi-feng/tc-bench/vlm_eval.py
code served (permissive licence) · get_code("02caab74330abf6b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. 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 base64
import tempfile
import os

# Create a temporary image file for testing
temp_dir = tempfile.mkdtemp()
temp_image_path = os.path.join(temp_dir, "test_image.png")

# Write a minimal valid PNG file (1x1 pixel transparent PNG)
png_data = bytes([
    0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A,
    0x00, 0x00, 0x00, 0x0D, 0x49, 0x48, 0x44, 0x52,
    0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x01,
    0x08, 0x06, 0x00, 0x00, 0x00, 0x1F, 0x15, 0xC4,
    0x89, 0x00, 0x00, 0x00, 0x0A, 0x49, 0x44, 0x41,
    0x54, 0x78, 0x9C, 0x63, 0x00, 0x01, 0x00, 0x00,
    0x05, 0x00, 0x01, 0x0D, 0x0A, 0x2D, 0xB4, 0x00,
    0x00, 0x00, 0x00, 0x49, 0x45, 0x4E, 0x44, 0xAE,
    0x42, 0x60, 0x82
])

with open(temp_image_path, 'wb') as f:
    f.write(png_data)

ARGS = {
    "image_path": temp_image_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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: none by the harness; any seed is in the setup block
extract_checks Ran
ran; our contract draft was wrong, not the code
weixi-feng/tc-bench/generate_assertions.py
code served (permissive licence) · get_code("bd3d9e1031c8ee4d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 re

# The function extract_checks takes a single parameter: text (a string)
# It parses the text to extract check information from a formatted string

text = """- Check validity
Input: Frame 1, 2, 3
Q: Does the input match the expected pattern?
Input: Frame 5
Q: Is the output correct?

- Check format
Input: Frame 10, 15
Q: Are all values within bounds?
Input: Frame 8
Q: Does it handle edge cases?
"""

ARGS = {
    'text': text
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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: none by the harness; any seed is in the setup block
extract_video_id Ran
ran; our contract draft was wrong, not the code
weixi-feng/tc-bench/download_youtube.py
code served (permissive licence) · get_code("8cba1a977e907813")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. 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 re

def extract_video_id(youtube_url):
    match = re.search('(?<=v=)[^&#]+', youtube_url)
    return match.group() if match else None

ARGS = {
    "youtube_url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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: none by the harness; any seed is in the setup block
plot_frames Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
weixi-feng/tc-bench/vlm_eval.py
code served (permissive licence) · get_code("bc59900741312cc7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned JpegImageFile. 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 numpy as np
from PIL import Image
import tempfile

# Create a temporary directory for outputs
temp_dir = tempfile.gettempdir()

# Create sample frames as numpy arrays with shape (height, width, 3) for RGB images
# Using distinct sizes: height=5, width=7, channels=3
frames = []
for i in range(4):
    frame = np.random.randint(0, 256, size=(5, 7, 3), dtype=np.uint8)
    frames.append(frame)

# Define indices of frames to plot
indices = [0, 2]

# Define plot indices (labels for each frame)
plot_indices = [0, 2]

# Define output filename
fname = f'{temp_dir}/test_plot.jpg'

ARGS = {
    'frames': frames,
    'indices': indices,
    'plot_indices': plot_indices,
    'fname': fname
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, matplotlib, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type JpegImageFile — a JpegImageFile output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a JpegImageFile output the recorded ndim 3 and dtype family int describe that coercion, not the output, and are not 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: none by the harness; any seed is in the setup block

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

Video generation has many unique challenges beyond those of image generation. The temporal dimension introduces extensive possible variations across frames, over which consistency and continuity may be violated. In this study, we move beyond evaluating simple actions and argue that generated videos should incorporate the emergence of new concepts and their relation transitions like in real-world videos as time progresses. To assess the Temporal Compositionality of video generation models, we propose TC-Bench, a benchmark of meticulously crafted text prompts, corresponding ground truth videos, and robust evaluation metrics. The prompts articulate the initial and final states of scenes, effectively reducing ambiguities for frame development and simplifying the assessment of transition completion. In addition, by collecting aligned real-world videos corresponding to the prompts, we expand TC-Bench's applicability from text-conditional models to image-conditional ones that can perform generative frame interpolation. We also develop new metrics to measure the completeness of component transitions in generated videos, which demonstrate significantly higher correlations with human judgments than existing metrics. Our comprehensive experimental results reveal that most video generators achieve less than 20% of the compositional changes, highlighting enormous space for future improvement. Our analysis indicates that current video generation models struggle to interpret descriptions of compositional changes and synthesize various components across different time steps.

For agents

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

get_harvested_code_for_paper("2406.08656")
get_code_for_paper("2406.08656")
have("2406.08656")

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

curl https://syntology.ai/api/ran/2406.08656.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.08656.svg)](https://syntology.ai/paper/2406.08656)

Connect an agent — have() is free.