RANGE, POP 2, POP 3, MINI Showcase (by the texture guy)

That looks beautiful Richard ! Well done !

1 Like

Three peaches, scanned with POP 3. It was very hard to get them right. I had to jump from Revoscan to Cloud Compare to Zbrush and do a hundred tricks to get them. It would be much easier if the devs could add the features I sugested :pray:

heledahn-peach-1-pop3

heledahn-peach-2-pop3

heledahn-peach-3-pop3

After made into lowpoly and rendered:

7 Likes

Hi @RichardWren ,

Excellent post.

I have recorded your suggestions and will discuss with team. As for the “retain the texture mapping after aligning/merging clouds somehow”, we know that many users are requesting this feature, but the dev team needs some time to find a solution.

Thank you again! Please keep up your great work.

4 Likes

I’d like to take the chance to add these scans I did for the MIRACO contest here, so anyone who’s interested can see my work in this thread :hugs:

They were all taken with POP3:

Now, the following was taken with Range.

I sat the actor in a computer chair in front of the scanner, and I spined the chair from Left to Right to capture 180 degrees of each expression. I didn’t need the scans to be perfect, since I was looking only to capture the facial expression and the wrinkles and folds, so taking each of these faces took less than a minute. I didn’t have to do any cleanup afterwards, only cropping away the hair and the chair :star_struck:

Then, I took all these expressions and projected them in the mesh of the character creator I’m developing (with ZBrush), and added them as shape keys, to make the face animation with mocap much more believable :sparkles:

In the video below you can see a small test I took.

(The character in the video is still a bit wacky (in beta) so it’s not as detailed as the final result will be :kissing_smiling_eyes:)

That’s all for now! I’ve finally fixed the issues with my MINI, so when I have a moment I will take some more scans of jewels and other tiny things.

See you then! :wave:

5 Likes

Hi Richard , great job

Do you know that you can now create textures after merging your cloud points ?

You can also edit your point clouds and clean them in CC and still merge them in Revo Scan 5 and generate textures after meshing the merged point clouds ?

I am working on some. Videos about that subject …will let you know when done .

Keep up the good work

2 Likes

That is wonderful news! I was looking forward to something like that!
GREAT JOB for the Devs!! :sparkles: :star2: :clap:
I will definitely test that!!

1 Like

Hi guys! I’m testing the Revoscan update and so far everything is working fantastic!

One suggestion for the devs though: In the merge section, can we have an option to save the aligned positions without actually merging the point clouds?

I would like that very much, so we can export the point clouds in the aligned positions to CC and merge them there (because I really need the Cloud registration option!)

Thanks in advance! You guys are really doing a wonderful job!! :100:

2 Likes

This is the result of my test. A very glossy paprika, yet the result was pretty good, and it took very little time to scan. Merging and editing, however, took a bit longer. I took three scans from different angles and tried to merge them in Revoscan 5. The first two merged pretty well automatically, but the third one was adding too many artifacts.

What I did was export the merge between scan1 and 2, and import it into CC, then exported the scan 3 and in CC I aligned it and registered it with the merge of scan1&2. The result was almost perfect (because of the very accurate Mesh register option of CC)

I merged it making sure the original merge of scan1&2 didn’t change position at all. This is important, because then, I could go to the Revoscan5 folder (located in C:\Users.…\AppData\Roaming\RevoScan5\Projects) and replace it with the point cloud of scan 1. Then, when I meshed and textured the point cloud, the paprika was complete :sparkles:

I think POP3 is very good. So much better than POP2. But the textures are still too low quality for professional use :pensive: but that’s the camera’s fault, so… there’s not much we can do about it.

1 Like

Another test. The automatic merging works incredibly well! This was the result of merging 3 scans of this tougan, and only by passing the overlap detection several times, the meshed result was this!

I did the same as before, replaced the original fused file with the merged one and asked Revoscan5 to mesh and texture again. The texture is good only in the parts that correspond with the original point cloud, obviously.

I imagine that repeating the same thing for all the other views would provide the remaining textures, and then with one final alignment in CC I could export everything to ZBrush and project the missing parts. :melting_face:

It would be awesome if somehow the devs could parent the camera positions to the point cloud and align them as well so the textures can be 360 degrees :pray: :sparkling_heart:

EDIT: Okay, I just realiced that the merged point cloud has access to all the textures and it uses them to make a new texture! That is soooo awesome!!! You guys are amazing, thank you :sparkling_heart: :sparkles: :100:

4 Likes

New batch of scans, three pumpkins from last Halloween, and these literally took less than 30 minutes.

I’m everyday more impressed with the evolution of the Revoscan software. The Align is perfect. The meshing is flawless. The texturing, is still not perfect but it’s because the camera resolution of the POP3 is very low.

But the fact that it textures the merged scan parts is a dream come true and I will forever love the Devs for it :cupid:

I don’t even need Cloud Compare any more!

There is only one thing that I would like to suggest: The simplify option of the Point Cloud does weird patterns. Why? Couldn’t it be just an even reduction all over the cloud?

Other than that, :100:

3 Likes

Hi Richard

Please never use reduction of pointcloud on a single scan , it will downgrade the accuracy and details , maybe the geometry based option but not the other and only after merging objects not for single scans .

Reduce your meshes, that is better idea so your keep the volume intact .

There is no need to reduce your pointcloud other than uniform it after merging …

1 Like

Yeah that’s what I want it for. After merging three scans, the point cloud was very dense, and I wanted to reduce it. Simplify created bald sports in pentagonal shapes, almost like a Voronoi pattern… so I reduced it in CC and replaced the cloud with the one in the Revoscan folder. Otherwise the mesh might have had artifacts…

Is this what you see with simplify? I was wondering about that myself. Uniform and Geometric options both do it, but Geometric seems to confine it to the flat areas. But it’s not consistent over the flat areas either.

1 Like

Simplifying a point cloud is a very delicate process , normally you would not want to use it on a single scan , you should use it on a mesh for better results .

The Simplify function for point cloud was created to use it in general on merged scans . To actually modify it by uniform the points before meshing , the values should be used at the minimal settings other way the accuracy can be altered dramatically .

The Geometry mode works well on low fusing models , if the scan was fused at very high pitch point settings the Geometry mode will not perform well , especially on organic forms .
I think the attempt can and could be improved.

However Geometry mode would works better in Simplifying a mesh , not a pointcloud .

Yeah that’s exactly it. Uniform and Geometric make the same strange patters.

I only use simplify after I’ve merged two or three scans, to make the point cloud less clustered. This option doesn’t work well yet in Revoscan, so I always end up doing it in CC with the subsample option, where i can specify a percentage (something like 0.25 for 25%) and it removes points evenly along the whole surface.

1 Like

Today I scanned with POP3 a very complex shape (maitake mushroon). I didn’t work too hard on this one, just two passes: one for the top and another for the bottom.

The resulting point cloud was missing a lot of information obviously, but since I’m going to make a low-poly model I think it looks very credible regardless.

Here is a comparison with another maitake mushroom I scanned a year ago using POP2.

It’s easy to see how much the tracking, meshing, and texturing has improved from POP2 to POP3 :clap:

3 Likes

Great job Richard !

After merging try only lose points and overlapped points cleaning to keep the accuracy . And eventual CC as long it don’t affect pitch point settings .

The Merging results are again messed up , there should be not overlapped points after merging … will check out with the team .

1 Like

New batch of scans, now that I finally have more free time :wink:

Papaya POP3:

Five Eggplants POP3:

This time I didn’t use CC, and so there are some artifacts visible in the places where the two scans (top part and bottom part) merged. I passed the Isolation and the overlap several times, and then meshed them with the highest resolution.

The papaya is almost perfect. The eggplants have a bit of noise in the intersection lines. The textures are amazing, but it’s very clear the camera of POP3 has low resolution…

But the tracking was incredibly smooth, and even without spraying the veggies, the results were very fast and good!

1 Like

I would note that simplify in Revo Studio doesn’t have this artifact. Also, instead of the percentage slider it tells you the current point spacing, and you pick a target point spacing. And it works.

@Revopoint-Cassie @Revopoint-Jane for awareness.