Hasta La Motion!

Hi, i’m wondering if anyone can explain to me how the depth channel is calculating from 3d scene in nuke. It’s kind a weird, i did some test using a 3d card. I found that the relationship between the distance (from camera to object/card) and the output depth value is not linear. Here are the data of distance – depth pair (first column is distance, second is depth alue) :

0.1 – 10
0.5 – 2
1 – 1
15 – 0.06
30 – 0.03
60 – 0.016

the conclusion , within the distance range of 0.1 – 1, the relationship is not linear, kind like an ‘exponential’ ? but it’s start to linear start from distance of 1. So i’m wondering how they’ve ben calculated. Don’t you think it should be linear all the way from 0 – unlimited. also can someone explain why the closest depth value is 10 ?

As comparison i checked the way Houdini output depth z channel, which is exactly linear to the distance itself (not only linear, but the distance and depth even share the same value). So let say if i want to work with image rendered out from Houdini, i can zmerge with FG which is i created using nuke 3d, by placing obj in nuke 3dspace at exact 3d distance (this is why i need to have linear depth from nuke, so the two can match in 3d space) . I just want to try this way , which is ussually done by grading the depth channel manually.

thanx