Raw, pre-processed data Intel D415

[ASriv7]

Are the various settings like exposure, brightness, and gain applied at the hardware level or in the software after getting the image? If the images are postprocessed in the software, is there a way to get the raw data from the Intel D415 camera without any processing whatsoever?

[idata]

Hello Leponzo

Thank you for your contacting Intel® RealSense™ Technology support. For future references, your ticket number is 03395430. Please keep this number handy in case you are asked for it.There is a community tread that can be the answer to your question:https://communities.intel.com/message/524610# 524610 https://communities.intel.com/message/524610# 524610The RAW format is not exposed to end users. To view all supported stream formats for each sensor use the command:rs-enumerate-devices -m We have submitted this request for support to the engineering team but we cannot say if and when it will be implemented.I hope you find this information helpful.Best regards,Josh B.Intel Customer Support

[ASriv7]

Thank you for the reply.

Pages 33 and 35 of the D400 series datasheet have rows called Format: 10-bit RAW for the left and right imagers and the colour sensor, respectively. So, does this refer to how the files are stored internally (but not available to end users)?

[MartyG]

A piece of Intel's RealSense documentation defined RAW10 as "Four 10-bit luminance values encoded into a 5-byte macropixel". In an earlier post that asked about RAW10, Intel support member Jesus G said "The RAW format is not exposed to end users."

[zfari]

During my first tests with the D435 Camera I observed that the measured depth points always building smooth surfaces. There is no noise in the depth data if you look at a smal selection of points. But you can see that there are structures like little waves in the depth data. It looks like some preprocessing to denoise the data, but this effect looks like there is a structure in a flat surface. It would be helpfull to get the raw data with all the noise and then I can look what points I would use or not.I tried to use different presets, the high accuracy delivers the best results (smalest waves), but I did not find anything to get depth data like I get from Kinect or Primsense.I think that post processing the output is not the solution, I just like to get a less preprocessed set of depth data. The problem is getting worse with further distance (>1,5m).I would be happy to get some hints.

see: https://showbox.onl/ ShowBox https://mobdro.onl/ Mobdro https://vidmate.onl/ Vidmate

[MartyG]

Depth accuracy can depend on the resolution you are using. Lower resolution causes greater error as the distance increases. It is recommended that you use 848x480 as your minimum depth resolution for good results.

The 400 Series cameras apply a distortion model called Modified Brown-Conrady to the data.