Does Contextcapture supports 48bit colors?
Oto, you're not asking the question in a right way (dont mean to be a pain) but anyway I assume you have 3 channel image with 16bit (12-14bit) each. In general CC does support RAW formats for various cameras - Nikon, Canon, Sony and so on but then my question would be why do you want to use 16bit images in CC ?
I mean is there any difference in aerotriangulation or it is internally down converted to 8bit/channel?
I see where you're coming from. Well in general in the "big photogrammetry" (aerial) we're sometimes using only a single channel to run TPM (tie point matching) and usually thats the Green channel (never use the Blue one). Would it make any difference in the AT - in your case I dont think so cuz CC is a black box plus if you use 16bit images for the AT then later on your textures in the final mesh will also be 16bit and they will look "black" to you. I dont believe CC has an algorithm to scale 16 to 8 bits automatically. So my suggestion in your case would be to scale to 8 bit images and use those for AT and Reconstruction, of course while scaling be careful not to overexpose or underexpose the output cuz then the matching will give bad results (noise in the point cloud). I hope I managed to help :)
Cheers,
Ivan
Exactly. The question raised because found out that one of aerial sets we have is TIFF with 16bit channels. It worked but not sure if it adds any benefits.
Using 16bit images in post-processing in aerial photogrammetry is definitely beneficial but in CC I have big doubts. Plus think about this - the size of the input images(much bigger in MB) and the size of the final texture. If I were you I wouldnt use 16bit images in CC processing.
FWIW, I asked Casey Bentley, during a webinar, if there was any advantage to using RAW imagery. He said that imagery is automatically reduced to 8-bits per channel when being processed by ContextCapture.
Answer Verified By: Oto
Hey Ron, just out of curiosity did they say how exactly the down scaling from 16 to 8 bit works ? Im just wondering how would they do it... from 16bit with its whole palette 0 - 65536 to 8bit with its 0-256 ? In case you have "16bit" image which has its histogram cut to 8bit and just saved as 16bit again (which doesnt make any sense) then CC could do it but otherwise... automatic histogram stretching... I dont know about that... :) Anyway if someone knows better, feel free to correct me :)
Ivan,
I edited my previous post to clarify that it was 8-bits per channel. This is 24-bit color (224) with 16,777,216 potential colors.
RAW formats are typically 12 or 14-bits per channel with billions or trillions of potential colors respectively.
Regards,Ron
At least from Pix4D reply it seems there is some precalibrated range for each camera used for conversion. Maybe CC uses same approach.
https://support.pix4d.com/hc/en-us/community/posts/202525579-16-Bit-per-color-channel-processing-problem
In CC there is some 3rd party library used for handling photos I saw some name in older version from technical log files before they were encrypted and maybe it doesn't handle 16bits at least it had issue with grayscale JPG
Ron, everything you wrote is I would say more than clear to me but I guess you didnt understand my question.
There is zero point in using 16bit (per channel) images in CC, simply because it doesnt know how to properly down scale the histogram. Even if you want to then your results would be unusable. As I wrote in my previous post the best is to stretch the histogram with LUT or any PS/LR curve, save the output as 8bit (again per channel) and then use it in CC.
Simple quick and durty example with 4 CityMapper images 16bit per channel, 3 channels in total (NIR is removed) and here how the result looks like:
Yes, the texture is 8bit(24bit color) but you see only the blacks and the variations of blacks till 256 pixel values coming from the 16bit input image.
So to recap - no point in using 16bit (per channel) images in ContextCapture.