Concurrent Image Understanding and Generation:
Self-Correcting Coupled Markov Jump Processes

1Google, 2Google DeepMind, 3Stony Brook University
Work done while being a student researcher at Google

Add a person walking along the dirt path, facing toward the ocean, wearing a backpack and casual hiking clothes.

Second research result visualization

CO2Jump in action: text and image co-author the answer. Three trajectories of CO2Jump on image editing, maze, and nonogram solving, showing the joint state at an intermediate step t and at the final step. The image-editing panel highlights the core mechanism: at step t the text branch has already begun committing a target-image bounding box in text for the new object person; by the final step the image branch has placed the hiker exactly inside the finalized box (we overlay the bounding boxes from generated text on edited image). The text branch plans where the edit should land, and the image branch executes that plan within the same denoising trajectory — no second forward pass, no external grounder. Maze and Nonogram show the same coupled-refinement pattern: partial-path and partial cell-fill commitments at step t converge with their text-side answers by the final step.

Abstract

Human cognition does not separate understanding and generation. A teacher at a whiteboard speaks and draws together, each modality reshapes the other. In this paper, we bring this coupled loop to artificial systems. Masked Diffusion Models (MDMs) are ideally suited to this task, yet existing samplers either decode text and image interleavedly or independently update them in parallel branches that share only previous-step history, but not the other modality's latest decisions within the same step; combined with MDMs' inability to remask, cross-modal contradictions are neither detected nor repaired. We introduce Self-Correcting Coupled Markov Jump Processes (SC-CMJP), a framework in which one modality's transition rates are functionals of the other modality's confidence score, as weighted by cross-modal attention. Furthermore, a remasking jump retracts commitments the moment cross-modal evidence turns against them. In conjunction with SC-CMJP, we introduce CO2Jump (Self-COrrecting COupled Jump), a novel training-free single-pass sampler for joint multimodal geneneration. For training and evaluation purposes, we have created and will release three large-scale joint multimodal generation corpora: JEdit-1M, JMaze-200K, JNono-200K, with matching in- and out-of-distribution benchmarks. CO2Jump achieves best joint performance for image understanding and editing as well as visual reasoning (maze and nonogram solving). The performance of the sampler scales monotonically with the number of denoising steps, evidence that the benefits of cross-modal coupling compound across the trajectory.

Self-Correcting Coupled Markov Jump Processes

SC-CMJP sampler step

CO2Jump sampler. A single denoising step from zt to zs. From one forward pass, the model produces per-token Self-Confidence for both modalities; cross-modal attention Aimage→textt propagates text confidence to image positions, and an entropy-based gate λ mixes self and cross signals into Coupled Confidence. The Death jump remasks the lowest-confidence committed tokens, and the Birth jump reveals the highest-confidence masked tokens under the noise schedule.

Datasets and Benchmarks for Joint Multimodal Generation

SC-CMJP sampler step

Dataset curation pipeline. For JEdit-1M, raw editing pairs are augmented by an oracle Qwen3-VL-235B that produces both the per-image scene-graph understanding and the thinking trace. For JMaze-200K and JNono-200K, source/target images and the structured understanding are produced algorithmically.

SC-CMJP sampler step

JEdit-1M training sample. A natural-language edit prompt (bottom) paired with the source and target images (right), the oracle Qwen3-VL-235B thinking trace, and the per-panel source/target scene-graph understanding. The bounding boxes and labels from MLLM's grounding solution are overlaid on the corresponding images.

Scaling Sampling Steps

Scaling sampling steps

Sweeping NFE from 8 to 512, CO2Jump is the only sampler that improves monotonically on both metrics: ImgEditBench rises from 1.72 to 1.93 and overall mAP from 0.074 to 0.369. Single-modality baselines plateau and regress at high NFE, and MMaDA-Parallel degrades from 1.52 to 1.44, its uncoupled schedule cannot productively use additional steps. The gap also widens: at 8 NFE the four methods sit within 0.009 mAP, but at 512 NFE CO2Jump is +0.015 / +0.016 / +0.034 ahead of MDM / ReMDM / MMaDA-Parallel. Coupling compounds across steps rather than saturating.

Qualitative Trajectories Expose the Uncoupling Failure Mode

Qualitative results

Every baseline violates at least one Nonogram clue (red ), while CO2Jump satisfies all clues; on the maze, baselines wander into wrong corridors or cut through walls and CO2Jump reproduces the ground-truth path. The right column exposes why: by step 160 of MDM's trajectory the text answer is correct but the image has drifted onto a different path, because under MDM's independence factorization the image branch never sees the latest committed text. CO2Jump's chain-rule decomposition propagates each text commitment into image scoring at the same step, so both modalities converge on the same solution.

Joint Image Editing and Understanding with CO2Jump

Maze Solving with CO2Jump

Nonogram Solving with CO2Jump

BibTeX

@article{le2026concurrent,
  title={Concurrent Image Understanding and Generation: Self-Correcting Coupled Markov Jump Processes},
  author={Minh-Quan Le and Armand Comas and Alexandros Lattas and Stylianos Moschoglou and Pedro Vélez and Amit Raj and Aaron Germuth and Thabo Beeler and Dimitris Samaras and Di Qiu},
  journal={preprint},
  year={2026},
  url={https://coupled-jump.github.io}
}