A question, please, about how to best use Masquerade that will give Zephyr the greatest possibility of success.
I'm involved with an on-going project to scan and create 3d models of historical Viking swords. (Some of this was described in an earlier thread.
Our test runs with modern replica swords have gone well, and now we are attempting to create models of historical swords. We've run into a new problem.
When using Masquerade to create masks with only the red stroke and blue stroke tools, I find that the software gives us a lot of false negatives, especially at the edges of the sword, the area where we need the most detail in our model, and the area that it seems to be most problematic for Zephyr to create a good model. The tools create masks that often miss parts of the edge of the sword.
Here's a screen shot from Masquerade showing a portion of one of the images at the tip of the sword before masking.
Using the red and blue stroke tools gives me this mask, which misses some of the point
If I try to refine the mask using the stroke tools, often Masquerade will mask the edge, but also include some false positives, masking a part of the background as an area to process.
The only way to assure a really tight mask is to use the polygon tool and laboriously trace out the edge.
Two questions:
Is there a better way to use Masquerade that will make it more likely to get an accurate mask?
And if there is an error in creating a mask, which is more likely to create problems for Zephyr: a false positive (including part of the background) or a false negative (rejecting part of the object)?
Thank you.?
I'm involved with an on-going project to scan and create 3d models of historical Viking swords. (Some of this was described in an earlier thread.
Our test runs with modern replica swords have gone well, and now we are attempting to create models of historical swords. We've run into a new problem.
When using Masquerade to create masks with only the red stroke and blue stroke tools, I find that the software gives us a lot of false negatives, especially at the edges of the sword, the area where we need the most detail in our model, and the area that it seems to be most problematic for Zephyr to create a good model. The tools create masks that often miss parts of the edge of the sword.
Here's a screen shot from Masquerade showing a portion of one of the images at the tip of the sword before masking.
Using the red and blue stroke tools gives me this mask, which misses some of the point
If I try to refine the mask using the stroke tools, often Masquerade will mask the edge, but also include some false positives, masking a part of the background as an area to process.
The only way to assure a really tight mask is to use the polygon tool and laboriously trace out the edge.
Two questions:
Is there a better way to use Masquerade that will make it more likely to get an accurate mask?
And if there is an error in creating a mask, which is more likely to create problems for Zephyr: a false positive (including part of the background) or a false negative (rejecting part of the object)?
Thank you.?
Comment