search

anderslanglands / alShaders2

BSD 3-Clause "New" or "Revised" License
75 stars 38 forks source link

Issues using cryptomatte in htoa #14

Closed javiertoledovis closed 6 years ago

javiertoledovis commented 6 years ago

Hi there.

We are testing cryptomatte in arnold for houdini (htoa 2.1.4, arnold 5.0.1.4, houdini 16.0.736). Everything works fine except for a few things.

1. Material names in nuke: When we read cryptomattes in nuke we can select objects by name, such as /obj/sphere1, but with shaders it's not working. Instead of /shop/blueMetal this appears: <-2.99242719543e+16>

2. User defined cryptomattes: We use user data to identify different groups of polygons or points (i tried both polygon and point data) inside of an object. When I define User cryptomatte 0 with an AOV name and the user data name the aov renders black. I think i followed the git wiki but I cant make it work.

3. Not clean crypto_asset_offset: When i use an integer offset in an object to select a subset of polygons it works but the colors are not clean, when I select an object parts of the others are selected as well.

Thank you for your time!

Cheers

jonahfriedman commented 6 years ago

Hi @javiertoledovis, thanks for reporting.

Could you double check which version of AlShaders you're using? Your first issue in particular looks like something we already fixed. If you are on latest (2.0.0b2), reproductions for all of these would helpful (ideally in the form of ass files).

  1. (Material names) This sounds like a problem that was fixed.
  2. (Custom Cryptomatte) Is your user data of type string? That's the only supported type for custom cryptomattes at the moment.
  3. (Non-clean offsets) I'm not sure what could cause this- could you send an ass file that renders this way and an EXR?
javiertoledovis commented 6 years ago

Hi @jonahfriedman

Thanks for answering. I noticed that i was using the b1 version, I just updated to b2.

1 (Material names)- Solved! 2 (Custom cryptomatte)- I just tried setting the user data in polygon level instead of point level and it's working. It didn´t work in b1 but it's working in b2. Great! 3 (Non-clean offsets)- Same solution as previous point.

Everything is working as expected. Thanks so much!

jonahfriedman commented 6 years ago

Ah good, thanks for testing and confirming.

I still don't know what issue 3 was, but I'm glad it's (seemingly) fixed. If you run into it again please don't hesitate to open another issue!