AI Paper Atlas

Research map / Video / Video generation

PISCES: Annotation-free Text-to-Video Post-Training via Optimal Transport-Aligned Rewards

Minh-Quan Le, Gaurav Mittal, Cheng Zhao, David Gu, Dimitris Samaras, Mei Chen

ICML 2026 · Video › Video generation · approach: Compositional text-to-video diffusion · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add OT alignment of text-video embeddings (ablation)PISCES · VBench VideoCrafter2Semantic score: 75.82 -> 77.63Improved
add OT alignment of text-video embeddings (ablation)PISCES · VBench VideoCrafter2Quality score: 83.44 -> 83.73Improved
add OT-aligned semantic reward to quality-only (ablation)PISCES · VBench VideoCrafter2Semantic score: 75.97 -> 77.63Improved
add OT-aligned quality reward to semantic-only (ablation)PISCES · VBench VideoCrafter2Quality score: 82.87 -> 83.73Improved
add OT-aligned quality reward (ablation)PISCES · VBench VideoCrafter2Quality score: 82.20 -> 83.77Improved
add OT-aligned semantic reward (ablation)PISCES · VBench VideoCrafter2Semantic score: 73.42 -> 76.99Improved
replace L2 mapping with OT alignment (ablation)PISCES · VBench HunyuanVideoSpearman correlation: 0.4873 -> 0.9018Improved
replace contrastive alignment with OT (ablation)PISCES · VBench HunyuanVideoSemantic score: 77.00 -> 80.33Improved
use partial OT m=0.9 instead of full OT (ablation)PISCES semantic reward · VTM accuracyVTM accuracy (%): 86.87 -> 87.54Improved
lower partial OT mass to m=0.5 (ablation)PISCES semantic reward · VTM accuracyVTM accuracy (%): 81.25 -> 78.92Hurt
add spatio-temporal cost (gamma=eta=0.2) to partial OT (ablation)PISCES semantic reward · VTM accuracyVTM accuracy (%): 87.54 -> 89.36Improved
lower spatio-temporal weights to gamma=eta=0.1 (ablation)PISCES semantic reward · VTM accuracyVTM accuracy (%): 87.54 -> 87.06Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy