I am familiar with merging components and have been successful. Here is the issue: I scan many fossils and artifacts and need a top and bottom, basically two scans with the same background. I set weighted control points and align components only to true. But, the alignment phase keeps trying to align the background and not align based on the control points. so the top and bottom are merged incorrectly. This is an easy fix in a controlled studio or with masks, but out in the field, this is an issue. Other packages can weight the manually places control points to create a manifold model. I am sure RC can do this easily, I am just missing something. BTW, I will post a tutorial as soon as I know how to do this, this is a very common Archaeology and Paleontology need. I would like to ween the universities and GO’s I work for off the slower competing apps, but this is a must to be able to do quickly.
The last thread has been unanswered fro 9 months. https://support.capturingreality.com/hc/en-us/community/posts/115001943912-Merging-2-Scans-together-
I think it should be easy with a photoshop action. Select subject, Make mask, Save as Tiff. Done next pic. 200-300 pics done in 15mins.
As long as there is enough contrast between subject and back ground it will be good enough. Don’t worry about the mask being perfect. RC can handle it.
I’ll try to make a video
Steve,
Thanks I have that process all down. I was just hoping the non mask feature existed. Seems not to.
Hi Montrose,
you can set the weight of the control points, just click on them and then the settings pop up.
Unfortunately, it does not have as big an effect as it seems to have with other software.
The problem is very likely that there are more suitable features on the background than there are on your object.
There are 3 ways that com to mind how you can deal with this:
-
Get more of the object and less of the background in your shots. I’m not sure about your image sets, but this is still surprisinlgy often the case. The object should fill at least 80% of the frame. If that is the case, then you should not have too much difficulties since your objects seem to have a perfect texture. With this method, I bet that it is a matter of numbers - why not take only a few all-over shots with background and then take many more images wiht very little or no background at all? Also, try turning the object not just twice but 4 times, so that you get nice overlaps. Also, try to change the background a bit - e.g. don’t just turn it on the spot but put it somewhere else for each side. That worked perfectly for me the other day even with a trickier surface. If the background is totally different, then RC wil automatically align the object.
-
Alter the background so that there are fewer features. Even out in the field it should be possible to cover the ground with white or black (or grey) featureless cloth.
-
Use the max features per mpx setting. Calculate it so that the value is the overall number of features divided by the mpx of your camera. This will force RC to distribute the features more evenly and so hopefully not favor the background. Of course that only works if the object covers significantly more than 50% of the frame.
I can’t think of an in app way to do it. I think I have a masking solution that’s easy and automated. Will post results in about an hour. Even if it doesn’t help in this specific case I think this could help many others looking for something close to the same problem.
PS could I see those fossil pics. Looks really cool
Gotz, I think you will enjoy this. It should complement and improve what you have already said, and relax some of the requirements at the same time. I read somewhere about Photoshop having a new AI for automatic subject selection, and I can say it works rather well.
Well, better to just let you watch.
Hi Steven, I think that’s quite interesting but I had no idea what exactly you were doing… 
Do you have a summary?
Also, I did a quick test (10 minutes) to illustrate what I mean. This is also in reply to a different question here:
A piece of rind (how spectacular) in 3 sessions, each on a different background. 2 Sessions aligned with the first try, for the second I had to add 3 control points. I think the mistake here was that I moved to a different location where the light was too different. Anyway, the result speaks for itself, I think. In the top 3D window you can see the two different layers of newspaper. After the reconstruction, there were a few blobs here and there floating around, but none actually connected to the main object - easy to delete.
Thanks for the detail. I was working much too hard. Your method is much simpler and works much better.
Best Regards,
Edward
Gotz,
In Photoshop, there is a new feature under the quick select wand. At the top there is a “select subject” button. Photoshop will attempt to find the subject in the image and select it. It works well enough.
I recorded an action in Photoshop (a micro) of select subject and mask selection save as tiff and close.
Then I closed out that image and went to “file”, “automation”, “batch”. In the dialog that pops up I selected the action I had just recorded, the folder that the source images where located and picked the folder where to save the new tiffs. It is important that I picked on error save to log, else on any error it would stop the batch job.
I ran the batch and 168 images masked with very little effort on my part. XD
This all makes sense and works, unfortunately masking is the answer when we get photos we did not take in bad inconsistent lighting. I am buying some vantablack and black 2.0 for my turn table to see if I can speed up the process.
Montrose, great idea. I’m stealing it!