Image rendering method and apparatus

1. An image rendering method for rendering a pixel at a viewpoint, comprising the steps of:

for a first element of a virtual scene, having a predetermined surface at a position within that scene,

evaluating whether to render a pixel corresponding to the first element using a machine learning system having been trained to output a value representative of the lighting of the predetermined surface at the position, or using an alternative rendering approach, and

rendering the pixel according to which of the machine learning system and the alternative rendering approach are chosen in the evaluating step, wherein at least one of:

(i) when the rendering step comprises rendering the pixel using the machine learning system, the rendering step comprises: (a) combining the output of the machine learning system with a distribution function that characterises an interaction of light with the predetermined surface to generate a pixel value corresponding to the first element of the virtual scene as illuminated at the position; and (b) incorporating the pixel value into a rendered image for display,

(ii) the evaluating step comprises evaluating whether the computational cost of using the machine learning system is less than that of the alternative rendering approach for a render of the first element, and selecting the machine learning system if it is,

(iii) the evaluating step comprises evaluating whether the virtual material corresponding to the first element is of a type predetermined as for being rendered using the alternative approach,

(iv) the evaluating step comprises evaluating whether a distribution function that characterises an interaction of light with the predetermined surface provides an indication that the first element is predetermined as for being rendered using the alternative approach, and

(v) the evaluating step comprises evaluating whether the error value or convergence performance of the machine learning system for the first element meets a predetermined criterion indicating that the alternative approach should be used.

2. The image rendering method of claim 1 , in which the rendering step comprises:

using respective machine learning systems that have each been trained for one of a plurality of contributing components of the image;

using a respective distribution function for each of the plurality of contributing components of the image; and

combining the respective generated pixel values to create a final combined pixel value incorporated into the rendered image for display.

3. The image rendering method of claim 1 , in which the alternative rendering approach is ray tracing.

4. The image rendering method of claim 1 in which the evaluating step comprises evaluating the image accuracy of the machine learning system for a render of the first element, and selecting the machine learning system if the accuracy meets a predetermined threshold.

5. The method of claim 1 , in which the evaluating step comprises performing a test render of a subset of pixels in a rendered image using both the machine learning system and the alternative rendering approach to enable a comparison.

6. The method of claim 1 , in which the evaluating step comprises referring to a predetermined criterion for the first element, based upon one or more selected from the list consisting of:

a material type for the first element associated with a below-threshold performance of a machine learning system;

an ID associated with for the first element and associated with a below-threshold performance of a machine learning system;

a distribution function associated with for the first element and associated with a below-threshold performance of a machine learning system;

a training performance of the machine learning system; and

a comparative render of one or more pixels using the machine learning system and the alternative rendering approach.

7. The image method of claim 1 , in which the virtual material of the first element has a material ID with which a value or flag can be obtained indicating whether the virtual material is of a type predetermined as for being rendered using the alternative approach.

8. A non-transitory, computer readable storage medium containing a computer program comprising computer executable instructions, which when executed by a computer system, causes the computer system to perform an image rendering method for rendering a pixel at a viewpoint by carrying out actions, comprising:

for a first element of a virtual scene, having a predetermined surface at a position within that scene,

evaluating whether to render a pixel corresponding to the first element using a machine learning system having been trained to output a value representative of the lighting of the predetermined surface at the position, or using an alternative rendering approach, and

rendering the pixel according to which of the machine learning system and the alternative rendering approach are chosen in the evaluating step, wherein at least one of:

(i) when the rendering step comprises rendering the pixel using the machine learning system, the rendering step comprises: (a) combining the output of the machine learning system with a distribution function that characterises an interaction of light with the predetermined surface to generate a pixel value corresponding to the first element of the virtual scene as illuminated at the position; and (b) incorporating the pixel value into a rendered image for display,

(ii) the evaluating step comprises evaluating whether the computational cost of using the machine learning system is less than that of the alternative rendering approach for a render of the first element, and selecting the machine learning system if it is,

(iii) the evaluating step comprises evaluating whether the virtual material corresponding to the first element is of a type predetermined as for being rendered using the alternative approach,

(iv) the evaluating step comprises evaluating whether a distribution function that characterises an interaction of light with the predetermined surface provides an indication that the first element is predetermined as for being rendered using the alternative approach, and

(v) the evaluating step comprises evaluating whether the error value or convergence performance of the machine learning system for the first element meets a predetermined criterion indicating that the alternative approach should be used.

9. An entertainment device, comprising

a graphics processing unit configured to render a pixel at a viewpoint within an image of a virtual scene comprising a first element having a predetermined surface at a position within that scene;

an evaluation processor configured to evaluate whether to render a pixel corresponding to the first element using a machine learning system having been trained to output a value representative of the lighting of the predetermined surface at the position, or using an alternative rendering approach, and

the graphics processing unit being configured to render the pixel according to which of the machine learning system and the alternative rendering approach are chosen, wherein at least one of:

(i) when the graphics processing unit is configured to render the pixel using the machine learning system by: (a) combining the output of the machine learning system with a distribution function that characterises an interaction of light with the predetermined surface to generate a pixel value corresponding to the first element of the virtual scene as illuminated at the position; and (b) incorporating the pixel value into a rendered image for display,

(ii) the evaluation processor configured to evaluate whether the computational cost of using the machine learning system is less than that of the alternative rendering approach for a render of the first element, and selecting the machine learning system if it is,

(iii) the evaluation processor configured to evaluate whether the virtual material corresponding to the first element is of a type predetermined as for being rendered using the alternative approach,

(iv) the evaluation processor configured to evaluate whether a distribution function that characterises an interaction of light with the predetermined surface provides an indication that the first element is predetermined as for being rendered using the alternative approach, and

(v) the evaluation processor configured to evaluate whether the error value or convergence performance of the machine learning system for the first element meets a predetermined criterion indicating that the alternative approach should be used.

10. The entertainment device of claim 9 , comprising

a machine learning processor configured to provide the position and a direction based on the viewpoint to a machine learning system previously trained to predict a factor that, when combined with a distribution function that characterises an interaction of light with the predetermined surface, generates a pixel value corresponding to the first element of the virtual scene as illuminated at the position; and

when the evaluation processor evaluates that the graphic processing unit should render the pixel using the machine learning system,

the graphics processing unit is configured to combine the output of the machine learning system with a distribution function that characterises an interaction of light with the predetermined surface to generate a pixel value corresponding to the first element of the virtual scene as illuminated at the position; and to

incorporate the pixel value into the rendered image for display.