GPU 3D Draw Call Isolation for Selective Upscaling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for secondary upscaling of 3D graphics content, such as Rendered-Super-Resolution (RSR), do not effectively handle rendered text or line-based UI elements with overlaid 2D content, leading to artifacts and inefficiencies, particularly in established applications and games that do not benefit from separating 2D and 3D rendering processes.

Innovation Solution

A GPU-based method that separates 2D and 3D draw calls by creating full-resolution and reduced-resolution frame buffers, determining the type of draw call based on application profiles, shader usage, and other criteria, and upscaling 3D draw calls while maintaining 2D content at full resolution, allowing for accurate combination and output.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If RSR is applied to upscale all rendered content, then frame rate is improved, but 2D UI elements become blurry and lose precision

Engineering Contradiction:
Improveframe rateVSAvoid2D UI element precision
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The rendering process is segmented into two separate paths: 2D draw calls are rendered at full resolution while 3D draw calls are rendered at reduced resolution and then upscaled. This segmentation allows each type of content to be processed according to its specific requirements, preventing UI blurriness while maintaining frame rate benefits for 3D content.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different quality levels are applied to different parts of the rendered output: 2D UI elements maintain full resolution for crispness, while 3D graphics content is rendered at lower resolution and upscaled. This local quality differentiation optimizes both UI precision and overall frame rate performance.

Inventive Principle:
Principle #3Local quality

2Productivity

If 2D and 3D rendering are separated, then RSR can be applied to 3D content, but device complexity increases

Engineering Contradiction:
Improverendering efficiencyVSAvoidrendering pipeline complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system performs preliminary classification of draw calls into 2D and 3D categories before rendering, using application profiles and detection mechanisms. This preliminary action enables the rendering pipeline to automatically route appropriate content to the correct resolution path without requiring complex real-time decision-making during rendering.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The GPU automatically detects and separates 2D UI content from 3D rendering content using built-in detection mechanisms and application profiles, eliminating the need for manual developer implementation. This self-service approach reduces the burden on developers while maintaining rendering efficiency benefits.

Inventive Principle:
Principle #25Self-service

3Adaptability or versatility

If automatic 2D/3D detection is implemented, then established games can benefit from RSR, but detection accuracy may be compromised

Engineering Contradiction:
Improvecompatibility with established applicationsVSAvoid2D/3D content detection accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

Application profiles are created in advance that contain pre-defined characteristics and signatures of 2D UI rendering patterns. During runtime, the GPU compares actual draw calls against these pre-established profiles to accurately detect 2D content, improving detection reliability without requiring complex real-time analysis.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The detection system uses multiple detection criteria and heuristics to identify 2D UI content, applying a series of checks that may be more comprehensive than strictly necessary. This excessive detection approach ensures high accuracy by cross-validating multiple indicators rather than relying on a single detection method.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentUS12008675B2Method and apparatus for isolating 3D draw calls for selective upscaling
Publication Date: 2024.06.11 SAMSUNG ELECTRONICS CO LTD
  • US12008675B2 patent drawing
  • US12008675B2 patent drawing
  • US12008675B2 patent drawing

AI summary

A system and a method are disclosed for separating 3D content from 2D UI content for selectively upscaling 3D draw calls. A controller, coupled to a graphics pipeline, determines whether an application supports upscaling and, if so, creates a full-resolution framebuffer for rendering 2D drawcalls bound to the graphics pipeline and a reduced-resolution framebuffer for rendering 3D drawcalls bound to the graphics pipeline. A drawcall is then determined to be a 2D or a 3D drawcall. The controller stores the drawcall in the full-resolution framebuffer if the draw call is a 2D drawcall and stores the drawcall in the reduced-resolution framebuffer if the draw call is a 3D drawcall. The draw stored in the reduced-resolution framebuffer is upscaled to be a full-resolution drawcall, and the 2D draw in the full-resolution framebuffer and the upscaled 3D draw are combined to form a final output.