Neural Frame Extrapolation With Confidence-Guided Ray Tracing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Cloud gaming experiences are hindered by the high computational demands of rendering high-resolution, high-frame-rate visuals with ray tracing, leading to frame rendering delays and drops that degrade user experience.

Innovation Solution

Implement neural frame extrapolation to predict the next frame and generate a per-pixel confidence map, focusing compute resources on regions below an acceptable threshold, thereby optimizing ray tracing within limited budgets.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If ray tracing is used to increase photo realism, then visual quality is improved, but compute resource demand increases

Engineering Contradiction:

Improvevisual qualityVSAvoidcompute resource demand

Core Design Contradiction:

Manufacturing precisionVSUse of energy by moving object

Solution Approach 1:

The system applies different processing qualities to different regions of the image. High-confidence regions use extrapolated pixels without ray tracing, while low-confidence regions use full ray tracing. This local differentiation allows photo realism to be maintained in critical areas while reducing overall compute resource demand.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

Instead of applying ray tracing to the entire image, the system applies it partially only to regions where the confidence map indicates low confidence in the extrapolated pixels. This partial action reduces compute resource demand while maintaining visual quality where it matters most.

Inventive Principle:

Principle #16Partial or excessive action

2Manufacturing precision

If high resolution/high frame rate rendering is used, then visual realism is improved, but frame rendering time increases

Engineering Contradiction:

Improvevisual realismVSAvoidframe rendering time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The system performs preliminary action by using neural frame extrapolation to predict future frame pixels before ray tracing is needed. This allows the system to prepare high-resolution data in advance, reducing the time required for final rendering while maintaining visual realism.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system applies high-resolution rendering selectively to regions identified by the confidence map as low-confidence areas, while using lower-resolution extrapolated pixels for high-confidence regions. This local quality approach maintains visual realism in critical areas while reducing overall frame rendering time.

Inventive Principle:

Principle #3Local quality

3Productivity

If neural frame extrapolation is used to predict next frame, then frame rendering speed is improved, but precision of pixel prediction decreases

Engineering Contradiction:

Improveframe rendering speedVSAvoidpixel prediction precision

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system uses feedback from the confidence map to identify regions where neural frame extrapolation predictions are unreliable. This feedback mechanism allows the system to detect low-precision areas and apply corrective ray tracing only where needed, maintaining overall rendering speed while improving prediction precision in critical regions.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20260057561A1Neural frame extrapolation rendering mechanism

Publication Date: 2026.02.26 INTEL CORP

AI summary

A mechanism is described for image frame rendering. An apparatus of embodiments, as described herein, includes one or more processors to receive a plurality of past image frames including a plurality of pixels, receive a predicted optical flow, generate a predicted frame and a confidence map associated with the predicted frame based on the plurality of past image frames and the predicted optical flow, render a first set of the plurality of pixels in the predicted frame based on the confidence map and adding the rendered pixels to the predicted frame to generate a final frame.