16bit instead of 8bit source images- better mesh quality?

[mraw]

Hi guys, I couldn't find a clear answer to this question. We did some tests with Agisoft, feeding in over 300 images from a Canon and a Cell phone. We used jpgs on the Canon as well. If we would shoot RAW images and convert them to 16bit: Would this help the software to build a better mesh with more details? Thank you.

[DayGeckoArt]

I don't know the answer to this but I have learned to only use raw photos, because of the color correction and adjustment possible. I tried using JPEGs collected with my Parrot Anafi using Pix4D Capture and it turned out terrible because white balance and exposure were inconsistent, with no way to get all images looking the same.

With the raw files, I can use my Colorchecker Passport to get an accurate profile for the lighting conditions, and also apply the same white balance. With 2 different cameras I think it's even more important to capture raw so you can get colors to match as much as possible between cameras

[PROBERT1968]

Mine would not allow my aerial photos with 16 bits in Metashape. I had to convert or export to 8 bits in order to work with them in the software.  I don't know about the  colors. I asked Alexey about that and he said only 8 bits works...  BTW, they are in black and white phots .

You can try and add a few of them to the software and see what happens. 

[Bzuco]

RAW can be converted to tonemapped 8Bit JPEG. In tonemapping process you can adjust brightness in over/under exposed parts of photos and bring this way more details(e.g. better filtering noise) for Metashape.
You can check what bit depth is storing your cannon in RAW(12/14/16 bit).
It also depends on what are you shooting, if it is sunny or cloudy day, ...does your camera allow you to see histogram during shooting?

[mraw]

We're shooting on a Canon RP in a (small!) studio environment. I checked the spec and C3-Raw should deliver 14bit. I don't see the point of rendering tonemapped jpgs from that. The question remains: Rendering 16bit tiffs or exrs from that raw-footage. Does this help Metashape in aligning cameras, conserving more details in mesh?
Thank you, guys!

[Bzuco]

If you are in studio environment and also have good lighting, then I see the point of using RAW only for reason better eliminate CMOS sensor noise as for JPEG noise reduction is hard coded in camera firmware.
With RAW you have more options how to manually reduce noise....but the noise level is also questionable as your camera is Full Frame and can boost ISO as low as value 50.
Mesh aligning process only compares pixels from photos and with good studio lighting setup all color values can be easily represented in 8bit value range.
So I think in your case using 14 bit color information does not make camera alignment better and the mesh results will be the same.

[mraw]

The last test we did 1/2 second exposure at 400 ISO. We will have a look to bring down the ISO. Thank you so much.

[DayGeckoArt]

We're shooting on a Canon RP in a (small!) studio environment. I checked the spec and C3-Raw should deliver 14bit. I don't see the point of rendering tonemapped jpgs from that. The question remains: Rendering 16bit tiffs or exrs from that raw-footage. Does this help Metashape in aligning cameras, conserving more details in mesh?
Thank you, guys!

There would be a benefit for your textures because you can adjust color, exposure, bring up shadows, etc