Rendering lightfields from 3DS Max

[Aaron Munro]

Hi again,

Thanks for your previous response. After having some success with the latest software. I also attempted to see how URP would go. While it may have been functional, it seemed as if each camera projected an emission onto the focal plane causing the entire projection to go crazy overexposed. You might know why and if that can be resolved. I couldn't tell 100% if stereo parallax was working or if it was still mono and shifting around gave the impression of it being stereo. Could only see glimpses of the projection in certain angles.

Anyway, I did have another question I wanted to find information about, as the subject suggests. If I were to export a light field array from 3DS Max to use in Colibri VR. What conventions do I need to use for example in file naming, image size, lens, information, camera position, scene origin, etc... 

I would like to test a number of different scenes and perspective exports.

Also, how well does the LF data compress down, can the data be streamed remotely, only access the data that's required. Do you have and interesting plans on your roadmap, if you have one?

Thanks again

Kind  regards

Aaron Munro

[Greg de Dinechin]

Hi Aaron,

Thanks again for your message!

Not really sure what goes on when using URP unfortunately, I would have to spend some time digging around to understand the differences, as I have not yet used it in any of my projects.

Concerning your second question, the best would probably be to convert the scene from 3DSMax into Unity (I do not have experience with 3DSMax, but this seems possible as there is a corresponding tutorial on the Unity website), then use COLIBRI VR's acquisition component to obtain your set of images with the right file structure.

If this is not possible, an alternative is to try to setup the file structure yourself. To do this, you need to emulate the file structure described here in the documentation, for which I'll provide some details here :

• The images should be in an images folder, and named 0000.png to <image count>.png. Image resolution should be lower or equal to 8192x8192.
• The camera models should be described in the sparse/cameras.txt file. Lens information goes here. More information is given here on the COLMAP documentation website, as we use their format for this file.
• The camera positions should be described in the sparse/images.txt file. Camera position relative to a scene origin goes here. Similarly, the same page of the COLMAP documentation website provides more information on how to structure this file.
  

Also feel free to use the acquisition component on a sample scene to get a better understanding of the file structure and how the information is formatted.

As for your question on data compression, absolutely, there is still much to be done! For now, I haven't focused on compressing the image data unfortunately, so COLIBRI VR is far from optimal from this standpoint, and the subject should definitely be on the roadmap. However, I unfortunately cannot find time to work on improving COLIBRI VR myself at the moment, so the project is very much on standby. Hopefully sometime in the near future!

Thanks again, and very sorry for answering so late: I haven't been able to answer questions on the forum in the past year, so I have to catch up.

Kind regards,

Greg