noise in measured material

[andy.engelkemier]

We were getting some Really heavy noise in a metal material when doing a bunch of different knobs.
I noticed the limiting factor was the noisy knobs all had the material set to Measured and Chromium. Gloss was anywhere between .02 and .2 (I guess they were going for a pretty satin metal)
Changing the materials to color instead solved the noise.

Is that a bug? Seems like a bug to me.

[DMerz III]

Hmm that's an interesting find, do you have a screenshot to demonstrate, we use measured metals pretty often, wondering if I am missing something.

[andy.engelkemier]

I had some time to test. I tried to change as little as possible from my scene so the team could look at what's going on.
I've attached an image. It's a crop of the file, which is 7.5MB so I had to put that on oneDrive.
https://teknainc-my.sharepoint.com/:u:/g/personal/aengelkemier_teknateam_com/ES3ggV1s2ptClqwmOtuad54BqzOJWrJzWeUd9v0RgJBYcw?e=Vo6rqx

I just created a fake knob quickly in solidworks and put it roughly at the angle and position the others were. I left all the cameras alone and sitting there in the file.

The image, from left to right.
measured Chromium .01 gloss
measured Chromium .1 gloss
color HSV 0,0,75 .01 gloss
color HSV 0,0,75 .1 gloss

Notice there is significantly more noise for the measured matieral. It's obviously more noticeable with the more satin materials. But you can see it in the glossier ones as well.
Render settings are pretty standard. I was getting away with only a few samples for some of the renders, then noticed a couple were Terrible. We had a couple people going back and forth between setting up a model in the scene, and photoshopping graphics and other edits, so materials got swapped on some of the knobs. That's how I realized the issue.

So I'll let other people test it further than this. I'm not sure exactly why this is happening in this specific instance, but I'm going to go out on a limb and say it shouldn't matter which setting you use. The number of samples for the material shouldn't change between the two. Also, changing the render samples doesn't seem to have any effect on the noise. 2 samples and 16 samples looks essentially the same in this example.

I hope that helps.