Jump to content

Fast Gi Anyone?


Recommended Posts

  • Replies 137
  • Created
  • Last Reply

Top Posters In This Topic

Yep, it looks good! Also, thanks for posting the computation times. It's good to know how fast this baby goes. :)

Cheers!

steven

20157[/snapback]

I wonder if Sesi would concider introducing a method like this natively into the render pipeline? It seems like it's a real possiblity for making GI more useable, and I imagine it could be even more efficient if it were built in at a grass roots level. Does Mark Elendt hang out here, he'd be the man to ask I guess, wish I thought of it the other day, could have asked him. Mario have you been using this successfully in production?

Link to comment
Share on other sites

Good job MADjestic! :)

Mario have you been using this successfully in production?

Yes. I keep thinking of ways to improve it but haven't had a chance to implement them yet... production getting in the way <_< But we used my first crappy implementation on a job recently and it worked out quite well.

Maybe we can convince Mark E. to add a vertex shading context in SOPs, with vertex area and things like that as globals.... (optional camera space, derivs only for parametrics, mostly just vertices and lights)... hmmmm... useful? :D

Cheers!

Link to comment
Share on other sites

Maybe we can convince Mark E. to add a vertex shading context in SOPs, with vertex area and things like that as globals.... (optional camera space, derivs only for parametrics, mostly just vertices and lights)... hmmmm... useful?  :D

20161[/snapback]

Whoever sais that houdini is looking more and more like Maya - we'll convice him in the opposit, no matter what is imposed by that statement :ph34r:

Link to comment
Share on other sites

Maybe we can convince Mark E. to add a vertex shading context in SOPs, with vertex area and things like that as globals.... (optional camera space, derivs only for parametrics, mostly just vertices and lights)... hmmmm... useful?  :D

Cheers!

20161[/snapback]

Use the force! :ph34r:

Fast GI gotta be useful.

Link to comment
Share on other sites

EDIT: I've returned the number of leaves to 70 (as it was originally). So if it doesn't seem to work - download the files again.

Thanks, hoknamahn for the bug report =)

EDIT: I've also added a parameter to vary the number of leaves of a "kd-tree" and packe everything into a single archive.

Hope this works now.

Ok, folks. I think I am done for now. I haven't added the second pass so far - maybe later. Neither I compiled it for h8, nor tested it for Linux.

You are free to play with the code and if you get a chance to optimize it - you are welcome.

Read the coments of occlusion SOP in the scene to set up the parameters. You'll also have to add a rabbit.ply model from 3DScanner site - if you want to use the rabit model.

It may be messy a bit but it is not a ready-made tool anyway. Just if you are eager to be a masochist - go for it. Otherwise - wait untill it's completed.

Cheers. And thanks everybody, that was a great experience!

occlusion.rar

Link to comment
Share on other sites

haha.

hoknamahn compiled mine code for linux.

We are working very close and often share ideas - but in this case I simply worked under windows and coulnd't compile it for Linux. So I asked hoknamahn to compile it for linux. And we have different realisation of this algorithm, btw.

Hoknamahm has also told me that he couldn't make my code work under linux. I am not sure about that, though. So if it doesn't work - then it doesn't =) I'll try to investigate the case later, but I don't promise anything.

I think that hoknamahn may want to share his code later.

Link to comment
Share on other sites

You guys rule!

Do either of you want me to do an initial commit of your stuff to the odforce Sourceforge project?

20209[/snapback]

Sure, Jason. If you think that the code at this early stage deserves being published - please feel free to do it. This is nothing close to a finished product so far. On the other hand publishing the code can let people improve the code - so that could be beneficial.

And thanks for your comments. Hoknamahn did help me alot - so he certainly deserves his portion of fame. B)

Link to comment
Share on other sites

  • 1 year later...

Sorry for bringing this RIP topic back to life for a while, but I thought it would be nice to keep promises and finish the work.

Here's a slightly reworked and cleaned up (1.0beta) version (linux compile, tested for 8.1.704, 711 32bit houdini compiles) of the occlusion SOP. Don't get afraid at the number of params - most of them are disabled (will be reworked later - now you just don't need that buggy piece of code they are related to). Just use "occ_k", "dist_k", "etheta" and "rtheta" if you want - or simply leave the defaults and enjoy. Use it if you like, but don't blame me =)

(if anybody wants the sourcecode - it will be soon uploaded to my webpage, or contact me otherwise)

Occlusion.rar

Link to comment
Share on other sites

  • 1 year later...

Hey Mario,

I found my old code and now have a time to play with it again. :lol: I'm thinking about acceleration structures and the question is how to calculate the normal for a leaf of tree? Average value sounds reasonable only for the first look (as we do for area, position etc.) Imagine a leaf which contains a sphere. Average normal on the leaf will be equal to {0.0f, 0.0f, 0.0f} which is bad. So how to cheat this?

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...