Hi everyone,
I'm finalizing my SOP for the EMJM in Imaging (Immersive Track). My background is in software architecture (WebGPU/3D), and I'm trying to frame a transition to academic research.
I've taken some strategic risks in this draft (~2000 chars) and would appreciate a "stress test" on these specific points:
The Grades Risk: I explicitly mention my lower math grades but frame it as a past "academic imbalance" I've corrected. Does this show maturity or just flag a weakness?
The "Startup" Tone: I mention co-founding a lighting engine startup. Does this establish competence, or make me sound too distracted for full-time research?
Technical Accuracy: Does my description of "non-Lambertian surface failures" sound technically precise to a signal processing expert?
Grandiosity: Is opening with the "Plenoptic function" and "Ultimate Display" too ambitious/naive for a Master's applicant?
Any feedback, especially from current students or alumni, would be huge. Thanks!
Current imaging technologies are merely lossy samplings of the Plenoptic function. My long-term curiosity lies in the field's pursuit of reconstructing the full light field moving beyond 2D arrays toward the theoretical "Ultimate Display." However, my attempts to build such systems have hit a wall: my work relies too heavily on "black-box" approximations. I am applying to the Immersive Imaging track to replace my empirical heuristics with rigorous physics, transitioning from a software architect to a disciplined researcher.
After graduating in June 2025, I co-founded Orlume, a WebGPU lighting engine we just launched globally. I engineered a pipeline integrating Depth Anything V2 and a PBR system to convert monocular images into interactive 3D scenes. While I successfully implemented dithered ray-marching, the system failed when handling non-Lambertian surfaces; my screen-space approximations collapsed, creating artifacts that broke visual immersion.
This technical ceiling reflects an earlier academic imbalance. During my undergraduate studies, I prioritized implementation over theory, resulting in lower grades in pure mathematics despite high performance in applied modules. I now realize that engineering without derivation is a dead end. I have since revisited Linear Algebra and Probability through supplementary coursework not as a substitute for academic rigor, but to ensure I am strictly prepared for the mathematical demands of this program.
I now seek structured discipline to master inverse rendering, multi-view geometry, and computational optics. I need to derive solutions from first principles to solve challenges like the Vergence-Accommodation Conflict mathematically. I am specifically drawn to Prof. Atanas Gotchev's work on light field displays and Prof. Mårten Sjöström's end-to-end 3D pipelines. Furthermore, my work on a low-latency (<5ms) vision-driven Virtual Piano aligns with Dr. Emin Zerman's focus on HCI and Quality of Experience.
The EMJM in Imaging uniquely combines the signal processing rigor of Tampere with the applied systems perspective of Mid Sweden University. I aim to contribute to the Plenoptima ecosystem, ensuring my transition to research is built on mathematical truth rather than just engineering intuition.
I'm finalizing my SOP for the EMJM in Imaging (Immersive Track). My background is in software architecture (WebGPU/3D), and I'm trying to frame a transition to academic research.
I've taken some strategic risks in this draft (~2000 chars) and would appreciate a "stress test" on these specific points:
The Grades Risk: I explicitly mention my lower math grades but frame it as a past "academic imbalance" I've corrected. Does this show maturity or just flag a weakness?
The "Startup" Tone: I mention co-founding a lighting engine startup. Does this establish competence, or make me sound too distracted for full-time research?
Technical Accuracy: Does my description of "non-Lambertian surface failures" sound technically precise to a signal processing expert?
Grandiosity: Is opening with the "Plenoptic function" and "Ultimate Display" too ambitious/naive for a Master's applicant?
Any feedback, especially from current students or alumni, would be huge. Thanks!
Current imaging technologies are merely lossy samplings of the Plenoptic function. My long-term curiosity lies in the field's pursuit of reconstructing the full light field moving beyond 2D arrays toward the theoretical "Ultimate Display." However, my attempts to build such systems have hit a wall: my work relies too heavily on "black-box" approximations. I am applying to the Immersive Imaging track to replace my empirical heuristics with rigorous physics, transitioning from a software architect to a disciplined researcher.
After graduating in June 2025, I co-founded Orlume, a WebGPU lighting engine we just launched globally. I engineered a pipeline integrating Depth Anything V2 and a PBR system to convert monocular images into interactive 3D scenes. While I successfully implemented dithered ray-marching, the system failed when handling non-Lambertian surfaces; my screen-space approximations collapsed, creating artifacts that broke visual immersion.
This technical ceiling reflects an earlier academic imbalance. During my undergraduate studies, I prioritized implementation over theory, resulting in lower grades in pure mathematics despite high performance in applied modules. I now realize that engineering without derivation is a dead end. I have since revisited Linear Algebra and Probability through supplementary coursework not as a substitute for academic rigor, but to ensure I am strictly prepared for the mathematical demands of this program.
I now seek structured discipline to master inverse rendering, multi-view geometry, and computational optics. I need to derive solutions from first principles to solve challenges like the Vergence-Accommodation Conflict mathematically. I am specifically drawn to Prof. Atanas Gotchev's work on light field displays and Prof. Mårten Sjöström's end-to-end 3D pipelines. Furthermore, my work on a low-latency (<5ms) vision-driven Virtual Piano aligns with Dr. Emin Zerman's focus on HCI and Quality of Experience.
The EMJM in Imaging uniquely combines the signal processing rigor of Tampere with the applied systems perspective of Mid Sweden University. I aim to contribute to the Plenoptima ecosystem, ensuring my transition to research is built on mathematical truth rather than just engineering intuition.
