Debayering Inside OpenCine
Debayer Algorithms Comparison.
"My task was to develop, implement and accelerate debayering algorithms into OpenCine. The chosen demosaicers were Nearest Neighbor Interpolation, Bilinear Interpolation, Green Edge Directed Interpolation (GEDI) and SHOODAK2, and were successfully implemented and accelerated with OpenMP. Towards the end of the project, I tried to accelerate with OpenCL, albeit with no significant advantages. The SHOODAK2 algorithm was also marked for needing a revamp, as ideally it should be faster than GEDI. The documentation can be seen here and the progress can be seen here. I also published an article describing my experience as a GSoC student." - Cláudio Gomes.
Bi-directional Packet Protocols
Bi-directional Packet Protocols.
"My task for GSoC 2018 was to develop a bi-directional packet protocol that uses a pair of MachXO2s as routing fabrics, extends the limited IOs of the Zynq SoC (system on chip) and utilises the bandwidth and support from various bus protocols on the Lattice FPGAs (field programmable gate arrays) e.g. I2C, SPI, GPIO etc. SERDES transceivers with 10:1 SERDES ratio were developed and the need for ECC has yet to be evaluated. The documentation can be seen here and a blog post documenting my experiences can be seen here." - Mahesh Chandra Yayi.
FPGA based Realtime Image Focus Peaking
"My task for GSOC 18' was to design an HDL based IP (hardware accelerator) that detects and displays high contrast edges of a video frame with a user defined color while maintaining the resource constraints of the available FPGA resources and the realtime constraints of the video. Towards the end of the project, the IP was successfully synthesized and simulated, while its instantiation into the AXIOM Beta pipeline is still under progress. The system code can be found here and the presentation, documentation as well as a blog post on the project can be found here." - Rahul Vyas.
Live Histogram, Waveform, and Vectorscope
"My Project Live Histogram, Waveform, and Vectorscope is intended to deal with processing of .raw12 image format. The first step was to store data from raw file in desirable way into the memory and extract data stored into the memory to perform the required operations on it. The histogram helps to understand if image is bright or dark. The dark pixels are at bottom and bright ones at the top. Vectorscope tool is able to measures the logarithmic ratio of Red and G component on v / s ratio of Blue and Green. These are chrominance. Overall, these tools will help in better understanding of RAW12 files. Documentation with source code of tools can be found here." - Iti Shree.
Video Container Formats
Testing the linearisation transformation with dummy data.
"My task was comparing and finding the right video frame containerization system that would be suitable for the AXIOM Beta camera. For this, many different formats have been studied. This includes but is not limited to Magic Lantern MLV, Cinema DNG, Cineform RAW, ProRes RAW, ARRIRAW and REDCODE Raw. Finally, a working prototype (Proof of concept emulation) was built and shown to work and can be found here. The task also saw changes in the code of mlvfs to incorporate PLR information in the form of a linearisation table in DNG file format. Finally, all the information regarding the final submission can be found here." - Supragya Raj.
Want to participate?
Want to provide feedback or suggest improvements? Want to try your own approach and need help?
Get in touch via IRC or email.