Pop2 versus photogrammetry

With proper lighting and a good DSLR, it is possible to get very good (higher resolution than Pop2) scans.

I used a cheap LED light with polarizing film and a polarizing filter on the lens to remove glare / reflections. Look for YT “scanning the void” for details of the setup. He uses an expensive flash, but I’ve found it’s not required under controlled studio conditions. The background is velvet.

I used Meshroom to process the images.

Pop2

Photogrammetry

Photogrammetry do not give you the proper object size as it can’t calculate that from photos .
the price of a good DSLR is the same as better scan than POP2 .
So why ruining a good DSLR for doing photogrammetry if you can get a better scan .

Proper photogrammetry require a lot of work as I was doing it for long time and of course the results are great .

I can scan smaller objects with POP2 that looks better than your preview , so what is the point ?
you are not showing what POP2 can do, you showing what YOU can do with it , and it can be much better than the preview above .

2 Likes

I don’t want this to turn into a measuring contest (double entendre intended). I am simply pointing out that there are other technologies.

Objectively:

I didn’t say it was cheaper.

Nothing about the process will “ruin my DSLR”.

Photogrammetry can absolutely give accurate size. Lookup April tags.

Also, there is an absolute lower bound to the size and resolution of the pop2 scans. I will challenge you (to take my inferior skills out of the equation) to scan a coin like a US dime with the Pop2. I will do the same using photogrammetry.

1 Like

Advantages to Pop:

WAY faster

Easier to use in different lighting conditions

Easier to scan objects overall (unless they are very big or very small)

2 Likes

Well my DSLR have a rolling shutter that is limited to how many pictures you can take it with , and each time you take a picture the value decreases to a point that you need to replace it and is no cheap.
So if your DSLR do not use rolling shutter then you are good .

You know that POP2 is limited to 0.1mm in resolution so what challenge it is ?

Photogrammetry can do better scans but it take time and effort to do it right and there is a lot of factors and processing that is involved.
Why you think I buy POP2 for , to make my work easier … and it works , and if I run into limitations with POP2 I am using different sources .
Comparisons between two different 3D scanning technologies are pointless .

You know exact the limitation of POP2 where Photogrammetry have a large spectrum of usage so this is not fair.

For what I’m doing with trying to scan people… A proper photogrammetry rig would be the correct solution. But it’s ungodly expensive.

To do an instant scan would take dozens of high end cameras all shooting at the same time. There are rigs out there that can do it… but was going to take me 10s of thousands to set up.

I also looked into using less cameras and projectors… But couldn’t find a good open source software solution.

So the Revo point is going to be a sacrifice in the time it takes to scan. But it was by far the cheapest solution that seems that it could work.

2 Likes

@Elfmaze you have no idea how the raw data looks like before cleaning from the systems with multiple cameras shutting at once . Well maybe you do , but it looks nothing as clean as POP2 provide .
We used the raw data from the “famous” company but the consuming time editing process was a time killer . All the fine details they sell after are the result from processing not scanning , I mean using the fine skin textures to produce the finest details that the scanning did not provided or it would be a monster file hard to even process .
For a human modeling for any usage I just want accurate volume of human form , I cares less about the fine details as for that you will need to shave the complete model and get rid of the fine peach fuzz as well or spending hours removing it manually affecting the volume on way or another , no thanks .

what about you do photogrammetry of the bust ? and then make comparison ? that would be fair, or a head shot , just the geometry .

This is the statement I was responding to with the challenge. You have already acknowledged that you cannot meet this challenge and that you feel it is unfair. The details of the bust are not sufficient to demonstrate my point anyway.

I want to emphasize this was not an attack on you personally or meant to diminish the Pop 1 and 2 capabilities.

I don’t plan to continue this thread. My point was made and readers can take away what they will.

I only wanted to point out that there are other processes (even though the Pop is technically photogrammetry as far as the underlying principles) that one should consider rather than being stymied by limitations of the device at-hand.

1 Like

Well that would be all 3D scanners then , since they all are optical scanners .

Well there is not a scanner in the world that will scan everything from a cars to a coin , it is not how it works .

If you try to make full body photogrammetry with just 1 DSLR camera you gonna fail compared to POP2 , and that is what I tried to say

it would be nice to have one solution for everything but on some point free photogrammetry is not so free after all if you want a real professional results .

And no worries , we just had a conversation , nothing personal against anybody .
If we were on the same page all the time , it would be boring after all .

At the end we will choice what is best for us .

I used photogrammetry 22 years ago for the first time scanning clients products like shoes and gadgets for their online shops for 360 degree presentations . The textures were amazing and fully automatic process with a push of a button that cost $10K, today I can do exactly the same for free .

Yeah esp as I’m trying to make 3d printed wearables. scale is critical. My SLT was indeed a mess with my first dslr/ studio lighting/ turntable test… But I assumed its still user error and motion in the model as it still took 5 mins to get all the shots. Eric Pare has a fantastic 360 system in canada north of you that I’m wondering how the results would be.

1 Like

Hi!

Does someone think I can tweak the scan and studio softwares to get close results as the attached picture using Pop2?

For the moment, I’m really disappointed with POP2… I not even closer of this. My model’s aren’t clear as photogrammetry. I’m creating parts for online training. Photogrammetry really is time-consuming.

My goal with POP2 was saving time by scanning items.

thanks!

How large is this part and the other parts you want to scan? I’m guessing you need texture as well?

Hi,

part is 4,25in x 4,25in x 2,25in. larger parts are 24in x 18in x 10in , cylinder.

Texture is really important.

thanks!
frederik

I don’t see it being an issue. Cylinder can cause issues with tracking if it lacks features.

Can you share a screen shot of the scan or scans you have made?
Are you using a phone or PC to scan?
Are you using a turntable for the smaller parts?
What mode are you using?
Are the parts shiny?
Would you share a screen shot of your scanning screen?

Without seeing your depth and RGB cameras and settings it’s hard to make suggestions to help you.

By the way, welcome to the community Fred.

thanks for welcome!

Here are answers to the questions:

Can you share a screenshot of the scan or scans you have made? Sure, see attached picture. Is the best render I had over 25 trials scans.
Are you using a phone or PC to scan? PC, I’m working with MacBook pro.
Are you using a turntable for the smaller parts? I’ve tried turntable alone, turntable with marker, no turntable, and finally, turntable with markers on scanned parts.
What mode are you using? The better mode I’ve found is Feature.
Are the parts shiny? No, parts without any reflections. Only 6 very small areas with metallic components. I have no issue with the metallic area.
Would you share a screenshot of your scanning screen? Here is a scan result followed by a screenshot of scan screen.

Thanks for help!

1 Like

1 Like

Sorry it took so long to get back to you.

First thing I see is the scanner is parallel to the face. Move the scanner so it’s aiming down about 45 to 60 degrees. Use markers with marker mode. Need at least 5 makers in the scan area. Here is a link to help explain more.

I’ll write more as you watch the video. You’re welcome. I’m more than happy to help. I have only done a couple of scans with texture, just playing around. I do have 12 years of 3d scanning experience and I’ll do my best to halp you.

If you are not needing really accurate scans or every tiny detail I suggest using fast scan. It’s still accurate, but has the benefit of being easier to keep tracking and saving lots of time with processing the point could and mesh. Once you get a betterr understanding start using the high accuracy mode. In the Revo Scan App, select guide, then under Scanning Tip and Tricks select Learn More. Page 7 has an example of scanning black and white. It explains the process. I’ll be up for awhile.