3DF Zephy

[KenmoJr]

I downloaded the trial of 3DF Zephry shortly early in the New Year when the Holiday Season price was being promoted.

Of the software I've tried - Meshroom and RealityCapture, Zephyr has the best and easiest to use interface. However I feel the trial period of 14 days is way too short. Photogrammetry is way too complicated to get a comfortable feel in. Most apps offer a 30 day trial.

There should have been a post when the Holiday Price would be expiring. During the "Holiday Season" is too broad and open to interpretation.

My intent for photogrammetry is to scan my collection of over 100 model and diecast 1/24 & 1/18 scale cars into 3D objects to be used in my Daz Studio and e-onsoftware's Vue Creator 3D renderings.

I also use Moi3D, 3DCoat 2021 & Silo 2.5. Currently trying to learn Blender 3D.

However I understand shiny metal surfaces can be difficult to capture in photogrammetry. I am also an avid photographer and have 7 DSLRS - Nikons D3300, D5200, two D7000s and D7100 and two Olympus E500 & E620. I mostly do automotive and landscape photography. Through experimenting I found one of my Nikons with a 50 mm lens and tripod give me the best results. I am photographing in a room with controlled lighting and a backdrop.

Despite the recommendation to use powder on metallic objects, I will not pursuit it. I keep my models in individual display cases to keep dust off of them. I've been building scale model cars and collecting diecast models since my youth and I am now close to 70 years old. Too many times in the past I've destroyed a model while dusting, by knocking off a small mirror, antenna, headlight etc.

IMHO to recommend dust or powder on a collectible is crazy.

Is there anyone else on this forum who can recommend how to address the shiny surface of collectible model/diecast cars. In the summer I may try it on my own real vintage car. But again coating it with powder is not an option and I will not take kindly to any who recommends doing so and will consider it an attempt to troll.

Thanks kindly

[KenmoJr]

So, can I assume no one is using 3DF Zephyr on metal diecast model cars?

[cam3d]

KenmoJr - Photogrammetry works by matching features so surfaces which are metallic, transparent or homogenous are not a good fit for the technology.

As you have mentioned, dusting comes with inherent risks of damage. Personally I use AESUB which sublimates, gives great results, but still runs the risk of damage on some surfaces (typically from the propellant, either butane or propane) if I remember correctly.

The next best option would be cross polarization - By polarizing the light source (strobe) and using a polarizing filter on the lens you can remove almost all reflections - This can be handy for stabilizing the photographic input into Zephyr by way of removing the reflections which move across the surface of the subject.

Other than using a polarizing filter, you need to take many more photographs to 'fill in the blanks' but this is non-ideal and the returns, as always, are diminishing.

[Dieter_Weber]

On this thread https://www.3dflow.net/forums/forum/...f-shiny-object the silhouette option was mentioned for shiny objects, but it is not in the cheaper versions of Zephyr.

I cannot recommend AESUB Green for your project, because in my case only half of it sublimated after 1-2 weeks (cold weather). Maybe I did use too much, and I used a different spray gun than AESUB sells, so I do not know if that might create the problem, but I do not think so. AESUB stayed on the surface like dust, so I had to use water and a brush to remove it.


Concerning Zephyr Free, I think you can use it forever with 50 pictures. Only Lite (500 pictures) is confined to 14 days.



When I saw your other post with the car, I thought about showing you my Zephyr reconstruction from yesterday (almost 200 pictures), which I rendered in blender as well. It looks as perfect in blender as it does in Zephyr. The problem is not Zephyr but the glossiness of your model:

[KenmoJr]

Thanks for all the replies. I truly appreciate them.

However I find the replies that photogrammetry is not for shinny surfaces very disturbing and should only be used for organic shapes or objects of a matte color. I'm sure there are many in the automotive industry who would love to use this technology and excluding them because your technology is not compatible is inexcusable. I see vehicles, spaceships, boats, bottles, metal cases 3d models, etc used in the game industry & film industry. Why not adapt your technology to those people.

It would be akin to making personal computers only for the word processing crowd.

By the way, I find 3DF Zephyr a very user friendly piece of software, especially compared to competing products.

I'm near 70 years old and recently retired from the computer industry as a server admin (Novel Netware, Citrix Winframe, SUSE & Windows) after 35 years. Photosgraphy, 3D modeling/rendering and vintage cars are my hobbies.

[KenmoJr]

On this thread https://www.3dflow.net/forums/forum/...f-shiny-object the silhouette option was mentioned for shiny objects, but it is not in the cheaper versions of Zephyr.

I cannot recommend AESUB Green for your project, because in my case only half of it sublimated after 1-2 weeks (cold weather). Maybe I did use too much, and I used a different spray gun than AESUB sells, so I do not know if that might create the problem, but I do not think so. AESUB stayed on the surface like dust, so I had to use water and a brush to remove it.


Concerning Zephyr Free, I think you can use it forever with 50 pictures. Only Lite (500 pictures) is confined to 14 days.



When I saw your other post with the car, I thought about showing you my Zephyr reconstruction from yesterday (almost 200 pictures), which I rendered in blender as well. It looks as perfect in blender as it does in Zephyr. The problem is not Zephyr but the glossiness of your model:

I disagree. Cars are meant to be shiny. So the problem is not with the car. It is with the inability of photogrammetry to deal with their glossiness. This issue must be addressed as the technology evolves.

Few people who own classic cars would take kindly to anyone dumping their car with dust or powder at a car show so they can take a photo of it.

[KenmoJr]

KenmoJr - Photogrammetry works by matching features so surfaces which are metallic, transparent or homogenous are not a good fit for the technology.

As you have mentioned, dusting comes with inherent risks of damage. Personally I use AESUB which sublimates, gives great results, but still runs the risk of damage on some surfaces (typically from the propellant, either butane or propane) if I remember correctly.

The next best option would be cross polarization - By polarizing the light source (strobe) and using a polarizing filter on the lens you can remove almost all reflections - This can be handy for stabilizing the photographic input into Zephyr by way of removing the reflections which move across the surface of the subject.

Other than using a polarizing filter, you need to take many more photographs to 'fill in the blanks' but this is non-ideal and the returns, as always, are diminishing.

By cross polarization do you mean using a color polarizing lens? If you do, the photos I took in this thread https://www.3dflow.net/forums/forum/...n-export/page2 were indeed with a CPL attached to the front of my lens.

[KenmoJr]

By the way I did visit your Facebook group last night and saw the Post Holiday Sale of your software would expire after January 8th.

Cheers and many thanks...

[cam3d]

KenmoJr

Photogrammetry can be broken down into a several processes, with SFM and MVS being the most well known and established. We have a range of reference publications which you can read over if you like: https://www.3dflow.net/technology/ - Photogrammetry requires reference data for reconstruction. This reference data takes the form of features which are matched image to image.

Cross polarization requires that the light source is filtered, and that the CPL and filtered light cancel each other out, removing reflections much more effectively than just using a CPL.

[KenmoJr]

KenmoJr

Photogrammetry can be broken down into a several processes, with SFM and MVS being the most well known and established. We have a range of reference publications which you can read over if you like: https://www.3dflow.net/technology/ - Photogrammetry requires reference data for reconstruction. This reference data takes the form of features which are matched image to image.

Cross polarization requires that the light source is filtered, and that the CPL and filtered light cancel each other out, removing reflections much more effectively than just using a CPL.

Thanks kindly for clearing up what cross polarization is.

And thanks for the link. I would gladly spurge for Zephyr Lite in a heartbeat if it did not have the issue with shiny metal surfaces. I am also aware that the competition Reality Capture, Metashape, Photomodeler, Meshroom, Autodesk, etc all have this issue.

My sole purpose in using photogrammetry would be 90% for scanning in metal scale model diecast cars, model cars and real life cars. I am a car nut. And own a vintage car as well. I did try the demo of Zephyr light on a couple of miniature figures ie: Spiderman, a sheep, a owl and a knight and they turned out quite well. They were all plastic except the owl which was porcilain.

I even retopod the sheep via auto retopo in 3DCoat but did not bake any normal or texture maps to it.

Just extremely disappointed and frustrated by the fact the technology does not accommodate shiny hard surface objects
Appreciate the replies very much.

[KenmoJr]

I did post some of my observations and concerns about photogrammetry and Zephyr on the DAZ 3D forums.

Perhaps a member here may also be a poster on DAZ3D, so I will include a link to the DAZ3D thread here.

https://www.daz3d.com/forums/discussion/comment/7009316/#Comment_7009316

Cheers & many thanks...

[cam3d]

KenmoJr Thanks for the feedback - There are other technologies which are being developed which have a different approach (such as Neural Radiance Fields) but are yet be available commercially. It might be quite a long time before they are public facing, and still have quite a lot of issues to overcome. https://www.matthewtancik.com/nerf