AdSpark: A Large-Scale Dataset and Benchmark for Product-Centric Advertisement Video Generation

Hugging Face Daily Papers Papers
advertisement-video-generation dataset benchmark video-generation multi-shot-storytelling product-identity ecommerce

Summary

AdSpark introduces a large-scale dataset (AdSpark-300K) and diagnostic benchmark (AdSpark-Bench) for product-centric advertisement video generation, evaluating six dimensions including product fidelity, temporal coherence, and advertisement effectiveness using data from a major e-commerce platform.

Product-centric advertisement video generation aims to create promotional videos that preserve fine-grained product identity while presenting selling points through coherent multi-shot narratives. However, this emerging task remains underexplored due to the lack of large-scale advertisement-specific datasets and comprehensive evaluation frameworks. To address this gap, we introduce AdSpark, a large-scale dataset and benchmark for product-centric advertisement video generation, based on data from a major e-commerce platform. AdSpark-300K contains approximately 300K reference image--prompt--video triplets, comprising a real-world subset and a synthetic subset. Each sample provides structured advertisement annotations, including product identity annotations, selling-point descriptions, creative plans, and aligned audio scripts, enabling models to learn product preservation and advertisement-oriented visual storytelling. We further propose AdSpark-Bench, a diagnostic benchmark that evaluates generated advertisements across six dimensions, including visual quality, product fidelity, instruction adherence, temporal coherence, audio alignment, and advertisement effectiveness. Based on AdSpark-Bench, we evaluate representative models, revealing key challenges in product preservation, multi-shot storytelling, and selling-point visualization. Experiments with AdSpark-300K-finetuned models further validate the effectiveness of our dataset. AdSpark provides a unified dataset and benchmark for future research, and we will release the dataset upon acceptance.
Original Article
View Cached Full Text

Cached at: 10/08/26, 12:28 PM

Paper page - AdSpark: A Large-Scale Dataset and Benchmark for Product-Centric Advertisement Video Generation

Source: https://huggingface.co/papers/2610.10047

Abstract

Product-centricadvertisementvideogenerationaimstocreatepromotionalvideosthatpreservefine-grainedproductidentitywhilepresentingsellingpointsthroughcoherentmulti-shotnarratives.However,thisemergingtaskremainsunderexploredduetothelackoflarge-scaleadvertisement-specificdatasetsandcomprehensiveevaluationframeworks.Toaddressthisgap,weintroduceAdSpark,alarge-scaledatasetandbenchmarkforproduct-centricadvertisementvideogeneration,basedondatafromamajore-commerceplatform.AdSpark-300Kcontainsapproximately300Kreferenceimage--prompt--videotriplets,comprisingareal-worldsubsetandasyntheticsubset.Eachsampleprovidesstructuredadvertisementannotations,includingproductidentityannotations,selling-pointdescriptions,creativeplans,andalignedaudioscripts,enablingmodelstolearnproductpreservationandadvertisement-orientedvisualstorytelling.WefurtherproposeAdSpark-Bench,adiagnosticbenchmarkthatevaluatesgeneratedadvertisementsacrosssixdimensions,includingvisualquality,productfidelity,instructionadherence,temporalcoherence,audioalignment,andadvertisementeffectiveness.BasedonAdSpark-Bench,weevaluaterepresentativemodels,revealingkeychallengesinproductpreservation,multi-shotstorytelling,andselling-pointvisualization.ExperimentswithAdSpark-300K-finetunedmodelsfurthervalidatetheeffectivenessofourdataset.AdSparkprovidesaunifieddatasetandbenchmarkforfutureresearch,andwewillreleasethedatasetuponacceptance.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2610\.10047

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2610.10047 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2610.10047 in a dataset README.md to link it from this page.

Spaces citing this paper1

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

AgenticGen: Reward-Guided Agentic Video Generation for Advertising

Hugging Face Daily Papers

AgenticGen proposes a reward-guided agentic framework for advertising video generation, decomposing the task into strategy selection and draft generation supervised by online business feedback and human quality rewards, with A/B tests showing improvements in CTR, CVR, and Advv on TikTok.

AdScope

Product Hunt

AdScope is a product that transforms confusing ad data into clear visual dashboards for agencies and business owners, displaying ad creatives alongside key metrics and adapting to business types during onboarding.

Adomate

Product Hunt

Adomate is a product that helps turn data into winning ads at scale.

AdAnt AI

Product Hunt

AdAnt AI is a new tool that leverages Claude to create viral, high-converting social ads, featured on Product Hunt.