Here is the result out of realitycapture with your photos:
A few observations with your photos. It looks like you moved the block around to get different sides. Generally you want to make sure nothing anywhere in your photos as been disturbed. When you move objects around you get conflicting alignment information between the background and the scan subject. Also when the block moves the lighting on the surfaces changes which results in mismatched pixel colors and difficulty making an alignment.
The main things to improve for better results would be a better camera system. The images are a little soft and the sensor noise is approaching the same level of contrast as the details in some areas- that makes it hard to find matching points.
Lonnie-S wrote:
asmasm wrote:
I did some testing for how accurate I can get the dimensions. Neither my method for measuring off of the 3d model or my cheap HF calipers are very precise. The scale factor looks to be somewhere in the range of 0.1% to 0.06%. I'm getting a block that threw a rod from a friend. I'll paint that one and upload the result.
. . . images removed . . .
Actually, I'd consider those dimensions to be very good. I'd be very interested in knowing more about the nuts and bolts of your photography process. I have a fiberglass Haynes Roadster nose from Jack at Kinetic. I'll be using it on my build initially. In the long term, I want to replace it with a custom nose build by me from aluminum or steel using metal shaping techniques. I'd love to have a scan of the fiberglass Haynes to use as a basis for that future custom shape done in MOI, Rhino or Fusion 360.
Cheers,
The camera is a nikon d800E with a sigma 50mm 1.4 being shot a f/11 on a tripod with a cable release. Scanning a fiberglass nose would be pretty straightforward but you would need to treat the surface so it wasn't glossy and had some details for the scan to pick up.
As far as shooting goes you want to take a lot of images in circles starting with the whole object in frame, then moving in closer.