November 2024 Flashcards

1
Q

Hyper Parameters

A

Parameters you set during the training. Not the parameters of the model that is generated.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Epoch (Machine Learning)

A

One full pass on the training data. There may be more, that would be multi epochs for training.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

ML training with multiple features (general knowledge)

A

When ML training using more than one feature. It is best to have the values of the metrics be similar in scale (because math).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

PBR (When it comes to 3d Assets)

A

Physically Based Rendering

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Albedo Map

A

True color without lighting effects

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Roughness Map

A

micro surface
shiny vs. matte

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Normal Map

A

Adds fine surface details w/o increasing poly count

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Ambient Occlusion (AO) Map

A

Simulates how light is occluded in crevices adding depth and realism

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Height Map

A

Adds actual displacement of geometry for more pronounced depth effects

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Metallic Map

A

Reflective (metal) vs. Diffuse (non-metal)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

MDL

A

Material Definition Language:
Standardized, high-level programming language designed to describe the appearance and behavior of materials in physically-based rendering systems.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

MaterialX

A

Open-source standard for defining and exchanging materials and their associated shading networks across different software tools and rendering engines. Developed by Industrial Light & Magic (ILM), it is widely used in industries like visual effects, gaming, and design to ensure consistency and interoperability of materials.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Wiki Q-codes semantic labels

A

To standardize semantic labels across different platforms and languages, the open-source database Wikidata is often utilized. Wikidata assigns unique alphanumeric identifiers, known as Q-codes or QIDs, to represent specific concepts or objects. For example, the QID Q42889 corresponds to the concept of a “car.” By embedding these Q-codes into 3D assets, creators can ensure consistent identification and interpretation across various systems and languages.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

DINO

A

DINO (DETR with Improved DeNoising Anchor Boxes)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Grounding DINO

A

Advanced Zero-Shot

How well did you know this?
1
Not at all
2
3
4
5
Perfectly