I have a unique challenge that hopefully someone can help me with.

I am trying to batch render a bunch of objects of different sizes in solaris. I need each object to be in the exact center of the camera and an exact size within the raster of the render. For eg. a render of 2000 x 2000 and the object needs to be 1600 pixels on its largest axis (either width or height). I don't care if I have to scale and re-position the object or if I have to manipulate the camera to achieve the result.

Can anyone suggest a good procedural way to achieve this?

Much appreciated!

You could use the Match Size SOP to scale each object to the 0-1 range. Afterwards, manually adjust the camera to get the object's desired resolution. It might require some trial and error, but you should do it only once.


Or with an orthographic camera, set the orthowidth parameter for framing. Its value indicates the camera view size.
For instance, with an orthowidth of 2 and a camera resolution of 2000, you should adjust your object's size to 1.6 to achieve a width of 1600 pixels in the render.

Andr
You could use the Match Size SOP to scale each object to the 0-1 range. Afterwards, manually adjust the camera to get the object's desired resolution. It might require some trial and error, but you should do it only once.


Or with an orthographic camera, set the orthowidth parameter for framing. Its value indicates the camera view size.
For instance, with an orthowidth of 2 and a camera resolution of 2000, you should adjust your object's size to 1.6 to achieve a width of 1600 pixels in the render.

The match size doesn't quite give me what I need. It will normalize the scale of the objects within standard x,y,z axis, but it doesn't guarantee that it will be correct from the view of the camera. I need pixel accurate framing of different objects from the view of any given camera.

I'll play around with the orthowidth, thanks for the tip!

Maybe this can help you:
https://www.scantips.com/lights/subjectdistance.html [www.scantips.com]

You can try with my file.

I'll play around with the orthowidth, thanks for the tip!
indigo card [indigocard.ltd]
jjsploit.click

jiejia
You can try with my file.

Wow, this is cool! Thank you! This definitely works. I'd prefer not to have to run it through a solver, but it's quite a nice solution!

Some of the limitations here is that it's doing an offset directly in the z axis, so if the camera is not looking directly down z and is rotated differently the offset doesn't work properly. For now, I'm just applying the inverse rotation of my render camera to the geometry and then applying your offset to it, and then rendering from the not rotated camera. Jumping through a few hoops, but at least it solves my problem.

Thanks again!