danwnewlands

guilhermecasagrandi

danwnewlands
I'm setting up a scene with both arnold and karma shaders. We have proxy and render mesh assets and I'm using the purpose to specify which one to render and which one is viewed in the viewport.

It seems that Arnold ignores this completely and will render both proxy and render meshes no matter what's specified in the render settings.

Anyone else had this issue?

Cheers

I've already sent a bug request for that. Purpose seems to be bugged.

Would you say its a sidefx bug or something on Arnolds side?

danwnewlands
I'm setting up a scene with both arnold and karma shaders. We have proxy and render mesh assets and I'm using the purpose to specify which one to render and which one is viewed in the viewport.

It seems that Arnold ignores this completely and will render both proxy and render meshes no matter what's specified in the render settings.

Anyone else had this issue?

Cheers

ashutoshm532

guilhermecasagrandi

MoltenCrazy
Currently using a ROP Composite Output for my text overlays.

Is there a faster/better option? Surprised how slow the node is at completing tasks when compared to the OpenGL ROP that generated the base images. The text overlay seems to be taking about 5x longer than the OpenGL render. And it's memory hungry, too.

Is this the best route to go for a text overlay that echoes wedge values? I'd hate to resort to rendering geo for this, but working on a lone machine and waiting so long for a text overlay gives me time for crazy thoughts... :P

--
mC

`
The fastest approach is using imagemagick for that. The main problem is their caotic docs.
I had to dig a lot and as far as I remember, I've used the generic generator for that, because the image magick TOP node doesn't give you enough options for working with text.

Can u share the Hip file please. I'm having trouble using overlay text looking for alternative.
Thanks

MoltenCrazy
Currently using a ROP Composite Output for my text overlays.

Is there a faster/better option? Surprised how slow the node is at completing tasks when compared to the OpenGL ROP that generated the base images. The text overlay seems to be taking about 5x longer than the OpenGL render. And it's memory hungry, too.

Is this the best route to go for a text overlay that echoes wedge values? I'd hate to resort to rendering geo for this, but working on a lone machine and waiting so long for a text overlay gives me time for crazy thoughts... :P

--
mC

mestela
I used the labs 'filter by value', but then I still had to use two of them to get the functionality of a split, and even then its more difficult than it needs to be because the operation mode needs to 'contains' (it fails on 'equal to' because its a string and not a number), and its another gotcha of do you or don't you use an @, or backticks, or make sure to spell the attribute right because it doesn't list the attributes in a combo box etc... too many ways to shoot yourself in the foot.

Looks like this if you're curious. I love that labs is trying to solve this problem, but it shouldn't be up to labs. Tops attribute discovery/comparison/manipulation should be core to tops and idiot-proof (cos I'm a high functioning idiot):

Image Not Found

mestela
I have a file pattern top returning a bunch of files. I want to split all the files that are webp images to convert them to a more standard format.

I put down a split top, and try the following, with increasing levels of irritation each time it errors:

@extension=webp
@extension="webp"
`@extension="webp"`
`@extension`="webp"
@extension='webp'

etc.

Eventually I remember there's a labs 'filter by value top', but even that doesn't work unless you remember to set the operation to 'contains', and I realise this is all way harder than it needs to be.

Can the tops expression fields be made more forgiving with text? Especially considering that comes up so often in tops graphs?

Also any kind of help would be appreciated within the UI when constructing tops expressions, and be easy to find in the docs. I always have to search around for too long to get the info I need, in the end its easier and faster to get the Labs team to write a helper top, or to vetch here on the forum.

-matt

Tim Crowson
Nah, I'm looking for a non-interactive drawing of it (I don't want to modify it, I just want to see it). Plus the F key option requires a camera lop be selected.

aswaab

mtucker
The temp file is created, rendered, then cleaned up by the USD Render ROP. So you will only see it if you go look for it while the render is still happening.

I actually watched my temp directory during render to see if it was putting anything in there, and I did not see any files happening. Maybe it was all too quick, but I thought I could spy the process maybe.

jsmack

guilhermecasagrandi
Thanks Mark. I´ve been using Karma for almost a year and that´s one thing that makes our work a bit harder sometimes. It would be very useful if we could use this kind of feature in the image viewport in Solaris.

Have you gotten the image viewport to show anything at all? I feel like I keep wanting a view that will show the framebuffer at 100% size regardless of the viewport dimensions but there doesn't seem to be one.

mark
Yes, rendering to mplay, you should be able to click to focus. The one tile you clicked on will be rendered to final quality (or almost) and the surrounding tiles will also get some further refinement.

Integrating this into the viewport rendering is a little tricky (mostly due to UI interaction). But we are investigating other options for future releases.

Oh, if you are writing your own delegate and want to use the mouse click information, it's described in: https://www.sidefx.com/docs/hdk/_h_d_k__u_s_d_hydra.html#HDK_USDHydraCustomSettings [www.sidefx.com]

No_ha

guilhermecasagrandi
Can you post your problems? I´m using Karma for a while and It´s working (mostly) as aspected (from Mantra).

Here is a file I made for demonstration, it's not the actual project, NDAs and all.

Mantra:

Image Not Found

Karma:

Image Not Found

No_ha
Big thanks to you Brian! Love seeing your results and tests. I'm always trying to use Karma for work but end up using Mantra in the end because of small little things I can't get working in Karma. My recent issue was using pscale in a shader to randomize emission color. For some reason, this won't work in Karma but perfectly fine in Mantra.
But seeing your results gives me hope that Karma needs just a little more dev time and extensive documentation but then it will become a great Mantra replacement.

tamte

rafal
Karma is a Hydra delegate, so it can access only the data that Hydra provides to it. And I don't think Hydra provides access to the whole scene graph. So, Karma may not be able to access that data.

doesn't have to be the whole scene graph, but would there be a way to access at least what's actually passed? cameras, lights, geo that's passed through (this relates to rfe#100498)

I was so happy when Mantra was able to finally reference geo using op: syntax just to be taken away by Solaris, while the expectations were quite opposite (to be able to access much more than just geo)
going back to referencing files directly is another unnecessary layer of complexity, given that asset paths may need a resolver to be correcct so hacking that in the shader feels like a workaround

rafal
Some workaround ideas:
- use geometry primvars to access geometry in a shader at render-time
- write out the data disk as Houdini geometry file, which then you can use in the Import Point VOP

rafal
Karma is a Hydra delegate, so it can access only the data that Hydra provides to it. And I don't think Hydra provides access to the whole scene graph. So, Karma may not be able to access that data.