Workflow Questions - New User

[wj247]

Hey all,

New to this program and have a few questions so I can figure out my workflow. I've previously used another program (photoscan) and had a pretty good workflow set up, wondering how much of it I can cross over here.

I primarily do, small object capture with a turntable and have been able to recreate through capturing both sides of the object by flipping it over. some questions I have about workflow:

1. Masking by Model: Can I create a basic 3D reconstruction (fastest settings, quick process) and export masks from a model. here's what i have done previously

• fastest reconstruction method for basic shape -> quick clean up -> quick 3D generation for shape
• create and export mask from model
• import masks from model for higher and more details reconstruction (i do the same for both sides of object).

2. Orientation/projection of camera: Because of my capture method, I usually have two sets of images (object one side up, and flipped over) of the object and gives me enough coverage.

• I import both image sets with created masks as above.
• I rotate the bottom image set (object upside down).
• create reconstruction (sparse cloud) and then I should get a whole connected object. cameras are calibrated and projections are correct.

This has worked for me for most of my object scans, and I was wondering if there are similar methods I can implement to improve my workflow as I continue to learn this program.

Thanks

[cam3d]

Hi wj247 - Welcome!

The rules of photogrammetry apply across the board, regardless of the software used to process the data set.

Masking by model is not currently an option within Zephyr. We have a bespoke masking program called 3DF Masquerade which can quickly mask by color or brushstroke. As you are primarily scanning with a turntable there are some image acquisition techniques that you can apply to your workflow which will exponentially speed up the masking process.

We have a tutorial dedicated to advanced turntable masking which can be found here: https://www.3dflow.net/technology/do...for-turntable/

Regarding the orientation of camera positions - This depends wholely on the subject itself and the complexity of its surface. Personally I shoot up to ten full rotations of the subject at between 7.5-15degree increments. Until recently I would move the camera up and down to get these angles, and for many subjects which can't be touched (because of subject shift, with for example a vase of flowers) this is still a requirement. However, banging together a little wooden ramp (and covering it with black material to ensure accurate masking by color) I can leave my camera in place and simply prop the subject on an angle, shoot, rotate, shoot etc...

If you have any more questions, feel free to ask away! More than happy to help if you get stuck at any point in the process.

[sam3d]

Cam it's cool to get insights like the above.

Maybe the team could pull together a bunch of updated blogs/tutorials showing a some real life capture scenarios? The existing ones are good but some specific examples would be great.

Of course it's subjective but areas could include:
Studio/in the field scene setup i.e. blackout materials, turntable setup/environment etc
Lighting, polarisation (cross) for different subjects
Zephyr settings
Mesh processing workflow i.e. smoothing, photoconsistency, decimation - what order to apply & when?

It wouldn't need to be formal - perhaps some interactive projects with shared datasets to follow?

[cam3d]

Funny you should request that. We are now working on some fresh tutorials for the next release