[Simone]

Hello everybody,
I'm a beginner in Torque3D but a competent programmer. I was able to run a basic project on macOS.

I want to couple a Torque3D world with an external AI (python) and I need to feed multiple camera views to my neural net.
What I need is to capture a frame buffer/ image buffer, whatever, of one or more Torque3D cameras.

I looked and searched for some API, but I was not able to find anything useful.
Anybody could help? Some link or information?

Thanks in advance
Best regards
Simone

[Azaezel]

https://github.com/GarageGames/Torque3D ... screenshot perhaps?

[Simone]

Dear Azaezel,
thanks for the link

The screenshot functionality requires to save to a file; this would not do because we are sampling multiple times per second and using the filesystem would be a huge bottleneck.

On the other side, If I understood correctly, this screenshot is taken on the current 'viewport' (I don't know if it is correct terminology); we require to take multiple camera views at the same time.

Does anybody know if it is possible to capture multiple cameras view in a fast way (not passing through a file)?

https://github.com/GarageGames/Torque3D ... screenshot perhaps?

[irei1as]

I've done some of that in viewtopic.php?f=10&t=866&sid=26ef5ec3b9 ... 05f9b1d9e7 but it's just basics and a bit outdated.

Feel free to check it and fix it.
The part about using it as bitmap comes from named textures and dump function, not sure.

On a side note, maybe in 4.1 we'll have camera-based texture render targets.

[Simone]

Thank you irei1as.
This link is more akin to what I need.

Best regards
Simone

I've done some of that in viewtopic.php?f=10&t=866&sid=26ef5ec3b9 ... 05f9b1d9e7 but it's just basics and a bit outdated.

Feel free to check it and fix it.
The part about using it as bitmap comes from named textures and dump function, not sure.

On a side note, maybe in 4.1 we'll have camera-based texture render targets.