Thank you Sassi, as always this is a fantastic way of getting a deeper understanding of how this works.
One thing I'd love to understand better is the starting point. As you said, you didn't have any info on the room size, but somehow you are still able to make an accurate representation, and without distortion!
For this image, I didn't have access to anything other than a website that has a "doll house" of a space created by Matterport, so no real info on anything.
But I did have the room size, which is how I made my cube. However my camera placement was clearly off, and the geometry I was adding was really distorted. How did you get such a clean camera position and no distortion?
In this case, I don't need quality so much, as the work I'm doing is more illustrative

Thank you again! I really appreciate looking at these files and seeing your method.

Admin: This thread was split, as a different theme was started. Ownership transferred to AlexC.
Part 1
https://cineversity.forums.maxon.net/topic/2023/modelling-with-camera-projection

Hello Sassi
Yes, that helps a lot. One thing I noticed is that if I start the camera calibration using Redshift, I get a really strange texturing effect, whereas if I am in standard renderer I can see the camera projection much more clearly.

Now I have a bit of a challenge. I had to get images not from a camera, but from a Matterport 3d scan. Meaning that the images don't really have any lens information, they are screenshots from a 3d camera...I still seem to be able to get decent results, at least for quick and dirty placement of objects, but the lens is so wide that any cube I put in the scene looks like a rectangle, even though the objects in the Matterport view seem "ok".
Do you have any recommendations on working with that type of image?

Here is a scene I am working on now
https://www.dropbox.com/scl/fo/umvcogd3048r13hfuk74h/APGoIeCJ-6HgDhwJhosScIg?rlkey=rvpwwv1qlbltfm0cb92a66wpz&dl=0

And one important comment...The images are also manipulated in Photoshop to remove the furniture, as I have to place LED screens and projection screens etc..in the space, so that is why in the first project file you saw that strange "dipping" in the floor, it comes from the Generative Fill tool in PSD. Here I included both the "raw" screenshot (in the tex folder) and the modified version, but I don't know if it makes any difference in this case.

Thank you!

Thanks Sassi
Out of curiosity, when you first opened the project I sent, what seemed to be wrong? Was the overall camera projection and position correct?

thanks!

Fantastic, thanks Sassi, I will try this approach with another shot!

Hello Sassi
Wow, this is perfect! I need to do this for some other photos and so I wanted to double check on your process
1/You worked in standard renderer, is this the better way to approach than working in RS with interactive render mode in the viewport?
2/I created the camera using the camera calibration tag. Is that a good approach?
3/The cube that was in my scene was at the right dimension for the room, based on my plans of the space, but. I can see that yours is very close.
4/I noticed your window polygons are all using the Geometry Axis of the room cube. Is that the way to place things accurately in space?
Thanks for any insight on the process! And thank you for the scene file!

Hello
I'm working on a fairly simple (?) concept, to take a picture of a room, projection in on some simple geometry using camera calibation, and then adding some surfaces that are just supposed to act as "mirrors".
What I'm trying to do here it to try to make all the windows into reflective surfaces.
However I'm not sure if something has changed since Redshift integration, but I'm finding it pretty hard to navigate and place surfaces. In the attached file, you can see I started with the left windows, but they are completely off, and I'm not sure how to find the right position for them in space.
What are some tricks for navigating this projected space?
https://www.dropbox.com/scl/fo/y2z7pdh9d433fskovfovb/ANuUKuBdXLRWJOI2b2DnKaU?rlkey=3tu2kwi8u5co3l8oj95wg3jlq&dl=0

Thanks!

Hey Sassi
This is great! I forgot that you can add fields in selection tags...It is a bit hard to make it precise though, I find it very "on/off". I will play around with it a bit more, but this is a great approach, thank you very much!

Hello Sassi, I've been playing around with that approach, and it does work. Not exactly the way you set it up, but see my file. I think I'm still curious to know if there is any object or field or vertex map I could use to have a more interactive (rather than numeric) approach to making these selections, as turning the Polygon indices on and off is a bit tedious, but this approach does work!
https://www.dropbox.com/scl/fi/s6w39calhdfb7e6dl08rr/WallNodeExtrude.c4d?rlkey=7lv3wolegr75ikb00ygqervtm&dl=0

Thank you!

Hello
I've been teaching myself scene nodes, and have been trying to do some simple stuff with parametric objects and selections.
One of the things that I find quite frustrating is that I haven't been able to figure out how to make selections based on a "region"
What I'm trying to do is
1/Select a few polygons (I can easily do this with the multiple selection tools by just entering their indices)
2/Apply a subdivision
3/Apply an extrusion to just that set of subdivided polygons

Between step 2 and 3 of course my selection no longer works as the indices have changed.

Is there a way to create something like a bounding box or a field that would "contain" any polygons within it, and make that a selection? That way I could subdivide and extrude in sequence without having to create a new selection?

Thank you!

Fantastic! Thank you Sassi! This is perfect for my project
All the best!

Hey Sassi, thank you! So simple and elegant!
It seems like the key is figuring out the IOR on the reflective material? Would it be possible to build something like this with 3 images rather than just 2?

Thanks again, really great setup

Hello
I don't have a project file for this as I have no idea how I would even start...I'm trying to replicate the effect of a Lenticular film

You see this in postcards or prints some times, when you look at it from one angle you have one image, but from another angle you have another image, with a "fading" transition between the two.

I'm assuming this would have to have to be a shader effect based on camera angle?

Thanks!

Hey Sassi, thank you! I missed this. Super clear explanation. I was thinking I could find the curve via the arc settings, but obviously it is not a perfect curve and your method makes perfect sense.

Thank you again!

Thank you Sassi, this is incredible. And extremely useful! May I ask how you managed to get the exact shape of the truss curves? This is what I was getting stuck on...did you use the blueprint as a background to extrapolate from?

Again, thank you so much

Hello

I'm trying to model this opera tent for a project, but I'm getting stuck because it feels like the dimensions are not equal across the curves.
I'm trying to use arcs to generate the two facing truss structure that are composing the main "shell" (I don't need to create the covering, just the structure).
Is there a way to extrapolate the actual curve from these measurements?

Here is the project I started
https://www.dropbox.com/scl/fi/wl78tb0g3szrab8h4q887/OperaTent.c4d?rlkey=ksa9tyl7llnll089bd6km8ln0&dl=0

And here is my document with measurements and drawings.
https://www.dropbox.com/scl/fi/prrueqrb3lo3u5ogps7h9/1_ZSB_Opera-Tent_18X21-Single_Grundma-e.pdf?rlkey=xv341fquqdmcsutqmjp22teqg&dl=0

Thanks for any advice!

Amazing! Thank you Sassi!!

Hey Sassi

Thank you, yes it is a sort of half/quarter square because the real world object has those proportions, so I am "cheating" squares by using rectangular boxes. But thank you for this, it is giving me a new idea on how to execute this in real life!
Is there a way that I can extract the number of cubes used to create the full outline of the dog? Maybe with Xpresso?

Thank you!

Hello
I'm working on a project where I'm using a cloner grid with a 3d model of an object that is at the real world size .
I'm using the Shader effector to only make the clones visible where the image I am feeding the shader effector is black.

What I'm trying to figure out is that I would like to calculate the smallest amount of 3d objects I need to feed the cloner to make the image visible at different sizes.
Right now my cloner grid is quite large, but in the real world it would need to be half the size.
What is the best way to play around with this setup to retain the "readability" of the image but pushing it to the limit of what is possible to read in the real world.

File is here : https://www.dropbox.com/scl/fo/dljhbz5i6yl8slvenvoif/AL5wRfmswnBQlNiVAN6vz8k?rlkey=j4zajo3stp80xy4j8z43cgyrq&dl=0

Thank you!