Originally posted by slipchuck Impressive! but how would these be implimented? would you sell the results to pentax as a firmware update?
Possible. One approach could be to deliver this as a firmware upgrade for existing cameras. Of course I know nothing about the architecture of their firmware and how it could be done and at what cost: would that impact FPS rate in continuous shooting, interfere with some other processing, or would that require significant changes in the processing pipeline are the questions I don't know answers to, but it can be done that's for sure.
There is also a possibility to make a chip for noise reduction using the algorithm either as a completely custom design fabricated from scratch or implemented as FPGA (Field Programmable Gate Array) which can be done in a garage and is somewhat close to fabricating custom chip but it may require support of an external microprocessor or imaging engine (similar to FPU floating point units in old days, like Intel i387 FPU helping i386 CPU with floating-point math). The downside is: you need a new camera for this so it is not an option for K10D or K100D (645D???), but I believe firmware version would be sufficient especially if PRIME engine on K10D can be reprogrammed to do the job.
Still, since I do not work on this full time I need at least three more months to complete more tests, analyze results, design preliminary hardware and software architecture, write technical and perhaps research paper, and check how all that affects printing, additional post processing and similar end-user related things.
Again, work in progress somewhere mid life cycle.
-= IVAN =-