Викторина, связанная с руководством · Жесткий Уровень
Викторина по гауссовскому сплаттингу
Изучите явные примитивы и средство рендеринга оригинального 3D Gaussian Splatting, а также оптимизацию, инициализацию, направленный цвет и ковариацию.
Вопрос 1 из 8
Which representation defines a scene in the original 3D Gaussian Splatting method?
Предварительный просмотр каждого вопроса8 вопросы
Ответы и пояснения появляются по мере прохождения теста.
- Which representation defines a scene in the original 3D Gaussian Splatting method?
- A required single hidden MLP with no explicit primitives
- An explicit set of anisotropic 3D Gaussian primitives
- Only one flat reference photograph
- A text description with no spatial parameters
- Which rendering procedure is central to the original Gaussian Splatting method?
- Projection and visibility-aware tile-based alpha blending
- Only repeated MLP queries along every original NeRF ray
- A rule that all primitives bypass training
- Only a black-and-white image average
- Which property set is modeled for each primitive in the original method?
- Only an output filename
- Only a scalar volume-density field for the entire scene
- Only one mandatory class label for all photographs
- Position, covariance, opacity, and appearance
- Что используется для изменения цвета капли в зависимости от угла обзора?
- JPEG-сжатие
- Только фиксированное значение RGB
- Сферические гармоники
- Случайный шум
- Which initialization supplies sparse points for the stated original method?
- Structure from motion
- A guaranteed complete manually built mesh
- Only a text-to-image prompt
- A required globally identical random point set
- When was the cited original 3D Gaussian Splatting research presented?
- 2015 год
- 2023 год
- 2020 год
- 2026 год
- In the paper’s adaptive density control, which operation can add coverage from a small Gaussian in an under-reconstructed region?
- Deleting every observed image
- Forcing every primitive to be one isotropic scene-wide blob
- Cloning it into an additional primitive
- Replacing its appearance with a mandatory text label
- Why parameterize the original Gaussian covariance through scale and valid rotation factors?
- To guarantee that no scene artifact can occur
- To turn every Gaussian into a text token
- To remove all direction-dependent appearance parameters
- To preserve a valid positive-semidefinite covariance form