How do I create a unified conditioning image?

[hwanhuh]

First of all, thank you for this great research.

I'm trying to reproduce your research, but I have one question, so I'm leaving an issue.
Using the blend_script file you provided, I got a series of m0r0~m5r5 rendering images for each view.
According to the paper, it seems like you 'stack' these to get a conditioning image, but how exactly do you get this?

Thank you.

[dunbar12138]

Hi, thanks for your interest!

We concatenate m0r1, m5r5, and m1r0 into a 3-channel image by treating each of them as grayscale to get the conditioning image.