Spektrální syntéza s omezujícími podmínkami
| Thesis title in Czech: | Spektrální syntéza s omezujícími podmínkami |
|---|---|
| Thesis title in English: | Constrained Spectral Uplifting |
| Key words: | spectral rendering|spectral uplifting |
| English key words: | spectral rendering|spectral uplifting |
| Academic year of topic announcement: | 2019/2020 |
| Thesis type: | diploma thesis |
| Thesis language: | čeština |
| Department: | Department of Software and Computer Science Education (32-KSVI) |
| Supervisor: | prof. Dr. techn. Alexander Wilkie |
| Author: | hidden - assigned and confirmed by the Study Dept. |
| Date of registration: | 27.03.2020 |
| Date of assignment: | 27.03.2020 |
| Confirmed by Study dept. on: | 29.04.2020 |
| Date and time of defence: | 29.06.2021 09:00 |
| Date of electronic submission: | 20.05.2021 |
| Date of submission of printed version: | 21.05.2021 |
| Date of proceeded defence: | 29.06.2021 |
| Opponents: | Mgr. Tomáš Iser, Ph.D. |
| Guidelines |
| Spectral rendering is a technique in which light transport is simulated across a continuous range of wavelengths, as opposed to traditional (RGB) rendering, which uses tristimulus color values. Its goal is to realistically visualise natural phenomena which are impossible to simulate with a traditional renderer, such as fluorescence, interference, dispersion or metamerism.
Textures and materials of objects in a spectral renderer must be described as spectra values, usually for the visible range of 380nm - 780nm. However, many existing graphics assets are described by RGB values, which are not directly usable in spectral rendering. In addition to this, it is not conceivable that asset creation will ever move from colorspace workflows anyway: artists prefer to paint using RGB, and will not directly work with spectral measurements. Over the last few years, a process called spectral uplifting has been introduced for the purposes of RGB tristimulus to spectral domain conversion. Although reliable algorithms for spectral uplifting already exist, they do not provide the user with the ability to constrain the process so that it matches certain RGB values to measured spectra of real objects. The goal of this thesis is to extend an existing uplift technique to provide such capabilities. The process will then, upon for a given sRGB value, compute a spectrum with respect to the given constraints which ensure that all those RGB input values for which measured spectra exist yield exactly these in return - and that all other RGB input values return plausible synthetic data which smoothly interpolates the measured spectra. The resulting system provides the key benefit to users that textures for real-world assets can be painted in an RGB editor, while the spectral reflectance still corresponds to the measured behaviour for those colours that correspond to real measurements. |
| References |
| Wenzel Jakob and Johannes Hanika
A Low-Dimensional Function Space for Efficient Spectral Upsampling Computer Graphics Forum (Proceedings of Eurographics), volume 38, number 2, year 2019 |