A few questions on getting a good scan of small objects

[Samadan]

Hello all, I wonder if anyone can give some advice on helping me get the best possible results from a dwarf hamster skull only 28mm long.



I'm using a reverse lens focus stacked (Helicon) macro technique, so have no lens data to give Photoscan. I want to help it with alignment while taking as many shots from all angles in one go.

Photoscan seems to work best when it uses some of the ground to help with alignment. What would be a good pattern to create and print as a base plate? Something like this ? ..

http://conservancy.umn.edu/bitstream/handle/11299/172480/Photogrammetric_scale_noncoded_markers_plus.pdf?sequence=28&isAllowed=y

I'm thinking high-contrast irregular shapes would be best? Does colour matter?

Also, thinking about image overlap and sequence of images. Does the order of photos matter? Should you always overlap with the previous shot? What about moving in for closer views filling the frame? I can get really really close if I widen the lens. Would this confuse Photoscan as it won't know the focal length from exif data?

Is there a way of manually helping with initial alignment?

Many thanks,

Stu

[lvargo13]

Hi Stu,

I'm not sure about your first few questions, but for your last few- in my experience, the colour won't affect PhotoScan (I change the contrast on images of glaciers so that the ice is not to bright), it will just influence the colour of your final model.

Photo order doesn't matter either. And you need overlap within your set of images, below is a link to a thread about overlap, but I think it's recommended to have at least ~60%. 

http://www.agisoft.com/forum/index.php?topic=1732.0

Finally, you can help with manual alignment by adding markers- here's an explanation: http://downloads.agisoft.ru/pdf/PS_1.0.0_Tutorial%20(Intermediate%20level)%20-%20Marker%20Based%20Chunk%20Alignment.pdf

Hope that helps a bit