Skip to content
WorldAtlas WorldAtlas WorldAtlas
Home

Long Sequence Imaging Quality

long_sequence_imaging_quality measures mean no-reference technical image quality (sharpness, noise, compression artifacts) over prediction frames using MUSIQ. Native scores are divided by 100100 and averaged to [0,1][0,1].

Simg=1T∑t=1Tstimg(1)S_{\mathrm{img}} = \frac{1}{T}\sum_{t=1}^{T} s_t^{\mathrm{img}} \tag{1}

Higher is better. Long-rollout quality signal computed prediction-only via the long_sequence backend (imaging_quality dimension; default MUSIQ-SPAQ weights, longer side resized to 512512). See Long-Sequence Metrics.

60-Second Same-Input Examples

Section titled “60-Second Same-Input Examples”

Audit manifest: docs/assets/metric_examples/longseq_memory_60s_examples_20260909.json.

Input image: memory-dolly_loop-memory_pool_371
Input image
60s memory_loop · same benchmark case · gap 0.6354

memory-dolly_loop-memory_pool_371

Suite: memory_loopScene: photorealistic / outdoor / city
Benchmark prompt target

Move the camera forward through the scene, then back to where it started.

prompt source: memory_loop_prompt

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Higher metric scoreScore 0.8007 · official

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Lower metric scoreScore 0.1652 · official
Input image: memory-yaw_loop_180-memory_pool_040
Input image
60s memory_loop · same benchmark case · gap 0.2667

memory-yaw_loop_180-memory_pool_040

Suite: memory_loopScene: photorealistic / outdoor / terrestrial_landscape
Benchmark prompt target

Turn the camera to the right, then turn back to the starting view.

prompt source: memory_loop_prompt

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Higher metric scoreScore 0.7690 · official

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Lower metric scoreScore 0.5023 · official
Input image: memory-dolly_loop-memory_pool_238
Input image
60s memory_loop · same benchmark case · gap 0.5673

memory-dolly_loop-memory_pool_238

Suite: memory_loopScene: stylized / outdoor / suburb
Benchmark prompt target

Move the camera forward through the scene, then back to where it started.

prompt source: memory_loop_prompt

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Higher metric scoreScore 0.8018 · official

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Lower metric scoreScore 0.2345 · official
Input image: memory-yaw_loop_180-memory_pool_427
Input image
60s memory_loop · same benchmark case · gap 0.4539

memory-yaw_loop_180-memory_pool_427

Suite: memory_loopScene: stylized / indoor / workspaces
Benchmark prompt target

Turn the camera to the right, then turn back to the starting view.

prompt source: memory_loop_prompt

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Higher metric scoreScore 0.7870 · official

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Lower metric scoreScore 0.3331 · official
Input image: memory-double_yaw_loop-memory_pool_320
Input image
60s memory_loop · same benchmark case · gap 0.4422

memory-double_yaw_loop-memory_pool_320

Suite: memory_loopScene: photorealistic / indoor / living_spaces
Benchmark prompt target

Turn away and back to the start twice in a row.

prompt source: memory_loop_prompt

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Higher metric scoreScore 0.7634 · official

Model result videos will be available with the ModelScope release. / 模型推理视频将在 ModelScope 方案上线后提供。

Lower metric scoreScore 0.3212 · official

References

Section titled “References”