Author Topic: Questions and Answers on new equipment support, camera bodies and correction mod  (Read 2524 times)

Sophie

  • Administrator
  • Sr. Member
  • *****
  • Posts: 396
Hello,

  In response to the many user requests and questions on new module availability, we have conducted for you an interview of the laboratory and development team managers in order to provide some answers:
 
Q. Why won't you publish earlier the list of camera bodies you are working on?
 
A. We try to communicate on the camera bodies and lenses we are studying through the news available on our web site.
 
 In spite of that, it remains difficult to communicate systematically and with certainty on the whole planning for certain reasons:
 
Very often we must wait for commercial availability before starting with a given camera body calibration. Even if we can usually work on a pre-release body version, we avoid doing so to prevent errors caused by last-minute modifications.
 
We do not know beforehand how complex and time consuming writing code for a given body will be. We must also consider logistical problems related with equipment purchase and leasing.
 
 
Q. What does the calibration phase consists of?
 
A. The Raw conversion in DxO OpticsPro isn't a simple file format conversion. It involves sensor measures on noise and colorimetry to achieve automatic image optimization.
 
 The body calibration phase consists in profiling the sensor, that is to say measuring the sensor in order to carry the best possible corrections. This means even if your lenses are not supported, you will benefit from DxO's know-how on digital sensor flaws characterization.
 
Camera body calibration - especially the study of sensor noise and its colorimetric response - is a meticulous task requiring dozens of days. It requires over a thousand individual shots just to characterize and calibrate a single camera body (RAW conversion, noise reduction, etc...).
 
Supporting a new Raw format also involves additional development time since most Raw formats are proprietary and undocumented, thus requiring « reverse engineering ».
 
 
Q. Why won\'t you publish earlier the list of modules you are working on?
 
A. We try to communicate on the camera bodies and lenses we are studying through the news available on our web site.
 
 In spite of that, it remains difficult to communicate systematically and with certainty on the whole planning for certain reasons:
 
 We must consider logistical problems related with equipment purchase and leasing.
 
 The frequent release of new camera bodies forces us to regularly reconsider the planning of our measure labs.
 
 The calibration of a given camera body / lens combination is highly time-consuming and lab time is a limited resource. Consequently, lab time allocation is submitted to constraints that make it difficult for us to communicate on a long-term planning.
 

  Q. Why aren\'t there more supported camera body / lens configurations?
 
A. First of all, each correction module corresponds to measures performed on a given camera body / lens combination. Even if work can be optimized when either the camera body or lens has already been analyzed, we still systematically perform measures on the specific equipment combination. Obviously, we cannot support every possible combination.
 
 Each correction module represents considerable work.
 
 Thousands of shots are required to characterize and calibrate a camera body / lens couple. Calibration techniques have been optimized, improving each year our capacity to deliver new modules.
 
 Why so many shots? It\'s essential in order to take into account all camera body and lens parameters that influence final image quality. Once the shots performed on various targets and live scenes, a behavioral and flaw model can be extracted for the body/lens couple. This flaw model is then associated to correction algorithms developed by DxO' researchers in order to build the correction module. Finally, a quality assurance phase verifies the module\'s proper operation under "real life" conditions. Only then can it be considered releasing the module to users.
 
 Experience shows us many differences exist between measures and the ideal model. For instance, Exif data is sometimes inaccurately set (focal length, focusing distance, etc...) and differently set from one body to the other,
 
 Finally, DxO Labs is currently looking into significantly increasing its DxO Modules production capacity in order to follow the DSLR and lenses market growth. We hope to be able to produce DxO Modules significantly faster and in larger quantities to address our current and future users\' expectations.
 
 
 Q. Why can\'t new camera bodies be integrated in the previous version?
 
A. First aspect to understand is unlike the optical corrections which are purely data driven, the corrections or processes related to camera body are built largely into the code itself. This is why new bodies are only supported through software releases, most often minor.
 
  Second, in DxO OpticsPro v5, the raw conversion and noise reduction algorithms were completely rewritten, involving recalibrating all previously released camera bodies. Specifically, the raw conversion is sandwiched between two separate noise reduction operations. This involves a different set of calibrations using raw images shot with different sets of requirements.
 
  In DxO OpticsPro v4 all noise reduction was performed post raw conversion, and calibrating a new camera for the older noise reduction algorithms involves 5 - 10 days of technician time (shooting, processing and quality assurance/validation). In order to support the 4 most recent bodies (Canon 40D, Canon 1DsMkIII, Nikon D300 and Nikon  D3) it would have added 20 - 40 days of lab and technician time to the work already necessary to support them in v5 (about 5 - 10 days per camera body).
 
  Supporting newer bodies involves a lot work. Bottom line is we have to balance the cost of retro-supporting new bodies for users who do not wish to upgrade to the most recent software with the benefits of supporting users who upgrade with more lens modules. It really is a trade off and not just a simple marketing strategy.
 
« Last Edit: August 28, 2015, 07:49:21 pm by pwirth »

 

photography
photography