jsmack
is it precision, or ray bias that's the issue? The default ray bias is 1mm, which would miss the other side of thin refractive surfaces like that.

err wasnt aware of ray bias, that fixes it. thanks!

So i have had a few issues with my characters scale, however I believe they are scaled pretty well for houdini. I could and maybe should scale them up a bit, but they are human biped characters exactly the same size as tommy (test geo). I have this issue where the charactrs come from daz and they have a "eye moisture" thin layer above the eye, but they are in such a similar spot that I end up having issues where karma will only show one or the other. I also had artifacts in a previous project where I was using redshift, where I would just nudge them both a bit using a peak sop, but it seems harder to get it to work with karma.

I guess my question is, should I scale up everything ? Is this the best fix? Even though it seems like things are the right scale in sops? Or should i scale my whole scene up just in karma? or am i being silly and should I not have an "eye moisture" layer on top of the eyeball and instead just use a coat or something over the eye in its own shader? or is there some other kind of fix I could try?

brians
I've fixed this now, and will push a fix out in the next day or so.
I'll let you know the release number once that is done

woah! sweet! Thank you Brian!

jsmack
lol dead pixels, are they NaNs?

they could be, im really not sure what the best way is to figure that out.

if i dolly in, it looks like this. if i rotate the camera slightly it goes away completely.

I've noticed that in some of my renders ive been getting occasionally dead pixels, almost like reverse fireflies with xpu (and it seems to only be xpu). a friend mentioned they were having this issue recently and now i'm seeing it as well. Seems like a bug!

im also trying to use the hair procedural, and seem to be feeding it everything it wants, but it is not showing anything in render!

my understanding is that if you say things are in the same trace set together they get path traced as one object or something? anyway, the only time im really familiar with using it is with SSS objects that intersect. you want to make them part of the same trace set if you dont want to get an ugly dark area where the two objects intersect. i was told at some point that this isnt yet supported in karma xpu, but i have no idea! was also just an example.


thanks this list is handy brian!

i think what you want to use is this? https://www.sidefx.com/docs/houdini/nodes/lop/backgroundplate.html [www.sidefx.com]

i have a lot of questions generally about what is supported by karma CPU, vs xpu, and theres an old chart of mantra vs karma vs xpu from the houdini 19 release i think, which is way outdated. it would be nice if this got updated, and / or if there was a good place to see all the differences clearly?

for example, im curious to know if trace sets are supported yet in karma CPU or xpu, i have no idea where I could check that though. i suppose i could use the changelog but that definitely isnt quite as clear.

the viewport here is GL, but this project is in RS. it's happening regardless. I can't easily share unfortunately, the project is kind of a mess, id be happy to show over zoom or give remote access to one of my machines. in the meantime the solution for me sems to be like just re-exporting everything in 19.5 as i need to. very confused here. i could provide a usd but it would be missing a bunch of texture files.

im noticing that if i go into the original asset file for the enviornment, and go to the original stuff that the first usd file was exported from, everything is there correctly and textures show up.

any .usd files that were exported with RELATIVE PATHS enabled from Houdini 19.0 do not show textures in 19.5.

HOWEVER, if i export them from 19.5 instead with relative paths OR without relative paths, it reads fine in 19.5.

So i guess the question is, is there some way to just correct these without having to dive into all my old files and update everything re-exported from 19.5?

so currently im having an issue rendering USD files with TOPs in 19.0, random frames are giving me exit codes and i can't diagnose why they aren't rendering.

i ended up trying to render the same usd files in 19.5 and it's nearly flawless. so i'm hoping to switch over to that.

however there's one major issue i'm having which is that a bunch of textures are disappearing when i open the usd files in 19.5.

so here's an example of one of my usd files in 19.0



you can see it says "@../ASSETS/Home Master 2022.fbm/FG Eight Figurine_EightFigurine_albedo.jpg@" and it shows up correctly in the viewport.

here's 19.5:



you can see it says that and then it makes a little -> arrow that points to the full absolute path (which i checked and is 100% correct).

so i have no idea what is going wrong here or what i'm missing! any help would be greatly appreciated.

wow! thank you! great to know.

jsmack

ChristopherRutledge
meant to attach this image but here's a hip file simplified, you can clearly see the faceting artifacts on the characters hands. if you play with the displacement/normals/SSS radius youc can see it more or less clearly all over the model.

Looks like subdivision wasn't enabled.

ah, that would explain it, where am i missing the subdivision setting? ive been looking for it in karma. i thought that it was "dicing" and it worked differently from subdiv in other renderers like redshift or arnold?

meant to attach this image but here's a hip file simplified, you can clearly see the faceting artifacts on the characters hands. if you play with the displacement/normals/SSS radius youc can see it more or less clearly all over the model.




https://drive.google.com/file/d/1o0XC73JQehmChjwEQCaLY-fJXaCcp_5F/view?usp=sharing [drive.google.com]

I'm noticing that I get some weird faceting when i use the SSS + detailed displacement, or normal maps sometimes.

as an example here's voronoi noise used to drive displacement. I see this faceting with either normals or displacement.

no idea how to get rid of this with karma, it seems to be an issue with both cpu and xpu. have tried all sorts of combinations of things. am i missing something or does this need a fix? its more clearly visible when the SSS scale is lower.

im confused, i dont think i understand what the houdini specific env var / packages have to do with the windows system level env variables? theres nothing in my windows system enviornment variables under path that seem to pertain to houdini in any way that I can tell, i didnt think these two things were related...

if using packages, how would that work?

yep, that did it. latest redshift is breaking karma somehow...

im also having this issue, im wondering if updating to the latest build of redshift caused it? trying without the redshift package to see if it makes a difference.