- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1(−1)
Read review
Paper summary
This paper externalizes the geometric memory of camera indices, and constructs a long-sequence teacher by combining sparse attention and global linear attention. It performs three steps of supervised training via joint distribution matching in approximately 30 seconds for a CFG-free student. The system supports continuous generation with a fixed memory budget, aiming to balance spatial revisit, long-term stability and segmented text control.
Critical strengths
- The external geometry library handles spatial recall, teachers provide long-term supervision, and students maintain bounded context; fixed storage and retrieval budgets impose clear computational constraints for continuous operation.
- In addition to the 2 hour qualitative example, statistical analysis of 8 65.5mins sequences is reported to conduct actual measurements of degradation issues during continuous operation.
Critical weaknesses
- Section 4.3 acknowledges that the content descriptor and event realization rate do not significantly distinguish short and long teachers, with clear benefits concentrated on photometric stability. Existing evidence is insufficient to attribute content drift suppression and dynamic interaction capability to long supervision windows, and the effects of teacher quality, training volume and window length need to be disentangled.
Questions for authors
- Varying only the supervision window (with fixed teacher, data, and budget), can achieve consistent gains in content metrics like returning individual identification and spatial layout?
- How does VBench perform when using multiple random seeds without prompt tuning for evaluation samples? Do all methods in WBench adopt consistent prompt processing and result selection rules?
- How to allow users update existing geometry while retaining the memory of unmodified areas? Please provide the success rate and consistency results of revisiting after editing.
3 comments
2 · Q1 · Supervision, Memory, and Camera Control
Long-horizon supervision suppresses accumulated color drift and maintains visual quality; retrieval and geometric reprojection provide cross-view scene memory and revisit consistency. Camera pose selects historical observations from the world state bank and defines their reprojection into the target view. This jointly supports scene recall and camera control (Fig. 4).
We compare revisit content similarity at step 500 to separate generation-quality and scene-memory effects.
Table 4. Revisit similarity (step 500; G: global, L: local; all ↑).
| Window | 12s G | 12s L | 24s G | 24s L | 48s G | 48s L | AVG |
|---|---|---|---|---|---|---|---|
| 2 chunks | 0.8346 | 0.5705 | 0.7834 | 0.4974 | 0.7171 | 0.4227 | 0.6376 |
| 8 chunks | 0.8346 | 0.5673 | 0.7802 | 0.4948 | 0.7258 | 0.4234 | 0.6377 |
| 16 chunks | 0.8373 | 0.5598 | 0.7796 | 0.4858 | 0.7117 | 0.4142 | 0.6314 |
Evaluation covers 12 scenes × 3 seeds. Global/local similarity uses cosine similarity of CLS/aligned patch features. AVG equally averages the six unrounded means.
Revisit similarity does not improve with longer supervision; global scores remain comparable. This agrees with Sec. 3.2 and Fig. 4: scene memory relies primarily on world-state retrieval and geometric reprojection.
3 · Q2 · Prompt Processing and Length Ablations
Thank you for your feedback; it will help clarify the final manuscript. Video models typically learn from detailed descriptions, while evaluation prompts may be brief. Expanding prompts while preserving intent follows existing practice: CogVideoX aligns inference text with training captions; VPO studies the detailed-caption/short-input gap; Wan2.1 provides prompt extension.[1][2][3]
GPT expands prompts under predefined guidelines: preserve subject, action, and scene intent, add details, and meet the required length. Outputs with incorrect lengths or altered intent are rewritten under the same rules.
Table 5 compares three versions of the same 200 prompts using the paper's model checkpoint, with duration, resolution, and sampling fixed. Semantic metrics use the original prompts throughout.
Table 5. Prompt length (mean words).
| Prompt | IQ | AQ | Subj | BG | MS | DD | Scene | Overall | AVG |
|---|---|---|---|---|---|---|---|---|---|
| Paper | 66.62 | 58.84 | 98.08 | 97.97 | 98.96 | 74.33 | 54.22 | 26.47 | 71.94 |
| Original | 60.76 | 56.60 | 98.17 | 97.23 | 98.73 | 50.45 | 28.59 | 25.23 | 64.47 |
| A (~30) | 61.59 | 56.96 | 97.53 | 97.05 | 98.76 | 80.50 | 42.18 | 25.22 | 69.97 |
| B (~120) | 67.40 | 59.16 | 98.02 | 97.93 | 98.98 | 73.23 | 56.56 | 27.01 | 72.29 |
Both expansions improve mean imaging quality, aesthetic quality, and dynamic degree. Overall scores stabilize beyond a certain prompt length; overly short prompts may deviate from the training-text distribution and yield larger score shifts.
References: [1] CogVideoX (2024), Appendix E; [2] VPO (ICCV 2025), Prompt optimization; [3] Wan2.1, Prompt extension.
4 · Q3 · Historical Geometry and Event Control
Historical geometry maintains world state and revisit memory; users trigger events through prompts without editing that geometry. Segment-specific prompts control auroras, meteor showers, crystal electrical storms, and stargate activation.
Using released Evoke weights, we test 50 prompts across these four event types. Event descriptions appear in chunks 5–7; the original prompt returns at chunk 8. Geometry retrieval, camera trajectories, and memory updates remain unchanged. Visual inspection of sampled frames counts an event as successful if it clearly appears during the specified interval and was absent beforehand.
Table 6. Event triggering.
| Event | Aurora | Meteor shower | Crystal storm | Stargate | Total |
|---|---|---|---|---|---|
| Success ↑ | 69.23% (9/13) | 84.62% (11/13) | 83.33% (10/12) | 100.00% (12/12) | 84.00% (42/50) |
42 of 50 prompts trigger the target event (84%). These events can thus be triggered without explicitly editing historical geometry.
Thank you for providing these additional results, which address most of my previous concerns. While I still have some reservations regarding the novelty of the proposed method, leading me to lower the soundness score to 1. But I recognize the extensive experiments provided by the authors. Accordingly, I will maintain my original overall rating.
Supervision-Length Ablations, Revisit Memory, Prompt Processing, and Event Control
Comment
Thank you for the careful review and recognition of our geometric memory, long-horizon supervision, and bounded computation. Your feedback helps clarify supervision, prompt processing, and event control.
1 · W1 · Benefits of Long-Horizon Supervision
Thank you for your feedback; it will help clarify benefit attribution. The redesigned teacher responds to timed prompts under both long and short supervision. Students learn this ability through distillation, so similar event-realization rates match our design. Longer supervision targets accumulated color and exposure drift, including excessive brightness, darkness, and overexposure. Length extrapolation enables joint scoring of 20 student-generated chunks (approximately 30 seconds); Appendix Fig. A5 shows extrapolation and instruction responses.
To isolate supervision length, we use the same fixed teacher checkpoint and Stage 2 student initialization, with training data and other settings fixed. We vary only 2, 8, and 16 chunks and compare at training step 500, holding teacher training state and training-data volume constant.
Given the full evaluation's scale and limited time and resources, we sample 200 VBench-Long prompts across categories and expand them under fixed guidelines. Generation and scoring settings are identical. Table 1 averages 3 fixed random seeds; semantic metrics use the original prompts.
Table 1. VBench-Long (step 500).
IQ/AQ: imaging/aesthetic quality; Subj/BG/Scene/Overall: consistency; MS: motion smoothness; DD: dynamic degree. All ↑.
Relative to 2 chunks, 16 chunks improve imaging quality, aesthetic quality, background consistency, and scene consistency; 8 chunks improve the latter three. Larger windows thus benefit several quality and consistency metrics.
Using 40 rewritten prompts spanning subjects, actions, and scenes, we generate approximately 30-second videos (20 inference chunks) with all three step-500 models under identical settings. Only the final 5 seconds are scored for imaging and aesthetic quality.
Table 2. Final 5-second quality (~30-second videos, step 500).
Both longer windows improve mean final-segment quality over 2 chunks, supporting long-horizon supervision for suppressing accumulated color and exposure deviations and improving visual stability.
We also compare event-triggering success rates:
Table 3. Event success by supervision window.
Success rates are similar across windows and exceed 44% (22/50) without Stage 3, supporting transfer of event responses through distillation.