Skip to content
WorldAtlas WorldAtlas WorldAtlas
Home

Temporal Flickering

Temporal Flickering quantifies inter-frame pixel stability in a generated video. High-frequency brightness or color shifts between consecutive frames produce visible flicker; this metric penalizes such instability using mean absolute error (MAE) on RGB pixels.

Given NN decoded frames {xi}i=1N\{x_i\}_{i=1}^{N}, the per-pair pixel MAE is

MAE(xi,xi+1)=1HWC∑h,w,c∣xi(h,w,c)−xi+1(h,w,c)∣(1)\mathrm{MAE}(x_i, x_{i+1}) = \frac{1}{HWC}\sum_{h,w,c}\left|x_i(h,w,c)-x_{i+1}(h,w,c)\right| \tag{1}

where HH, WW, and CC are height, width, and channel count. WorldAtlas evaluates MAE on uint8 RGB values in [0,255][0,255]. When resolutions differ, the later frame is resized to match the earlier frame before differencing.

The temporal flickering score aggregates adjacent-frame MAE:

Sflick=255−1N−1∑i=1N−1MAE(xi,xi+1)255(2)S_{\mathrm{flick}} = \frac{255 - \frac{1}{N-1}\sum_{i=1}^{N-1}\mathrm{MAE}(x_i, x_{i+1})}{255} \tag{2}

Higher SflickS_{\mathrm{flick}} indicates less flicker. The metric is prediction-only and requires at least two frames.

Verified Same-Input Examples

Section titled “Verified Same-Input Examples”

Audit manifest: docs/assets/metric_examples/missing_pages_examples_20260824.json.

Input image: image_static_stylized_passageways_0134
Input image
Input verified · same benchmark case · gap 0.3521

image_static_stylized_passageways_0134

Suite: image_staticScene: stylized / indoor / passageways
Benchmark prompt target

Camera pan right. An office hallway with office doors, bulletin board, and people. Camera move right. An office lounge area with couches, coffee table, and plants. Camera pull out. An office lounge area with entrance, reception desk, and waiting area.

prompt source: prompt_target

Both videos belong to the same benchmark sample; labels compare this metric only.

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 0.9907

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 0.6386
Input image: image_static_stylized_city_0123
Input image
Input verified · same benchmark case · gap 0.3323

image_static_stylized_city_0123

Suite: image_staticScene: stylized / outdoor / city
Benchmark prompt target

Camera pull out. A park path with benches, trees, and wider path. Camera move right. A city street with buildings, parked cars, and benches. Camera pan right. An urban square with fountain, people, and shops.

prompt source: prompt_target

Both videos belong to the same benchmark sample; labels compare this metric only.

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 0.9766

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 0.6443
Input image: image_static_photorealistic_verdant_landscape_0028
Input image
Input verified · same benchmark case · gap 0.2363

image_static_photorealistic_verdant_landscape_0028

Suite: image_staticScene: photorealistic / outdoor / verdant_landscape
Benchmark prompt target

Camera pan left. A building facade with trees, pine needles, and car. Camera pull out. A street view with parked cars, pedestrians, and building facade. Camera move left. A street view with city park, benches, and people.

prompt source: prompt_target

Both videos belong to the same benchmark sample; labels compare this metric only.

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 0.9873

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 0.7510

References

Section titled “References”