Let's see some Prman 11 renders!

[aubec]

I've implemented your "untested" script into a full shader and it works fantastically well. I guess some people can just write code and have it compile first time

Once again, thanks very much, I think I'll be using this a lot since my ambient occlusion render times have now come down from 2 hours a frame at 1k to 12 minutes a frame at 2K (including shadow map render times)!

Brilliant.

[firewalkwithme]

yo

any chance you'll be posting that shader of yours?

i'm getting mighty fed up with ray traced occlusion
and would like to see what that depth based sollution
has to offer

thanks

[aubec]

Yes, but I thought it might be a good idea to wait till I've finished the companion shaders to take advantage of all that lovely bent normal data...

I don't promise to finish it soon (I've got a feature to finish imminently) but when I get the time I'll put it all together in a Slim template. I'll probably post it on my site when it's done but don't hold your breath.

Andrew

[aubec]

Well, the tutorial's finally done. When the company you work for goes bust you suddenly find yourself with more time on your hands...

Ladies and gentlemen if you point your browsers at http://www.andrew-whitehurst.net/amb\_occlude.html then hopefully you'll find all you need to implement Ken McGaugh's depth-map based ambient occlusion. Sorry that the pictures are a bit crappy but they were the test frames I was working on when the axe fell at my company so I had take those and get out! Since I don't have RAT at home I have had to finish the tutorial "blind" as it were. That's my excuse for any bugs you find anyway

Hope you find it useful. The techniques are a bit complicated and the workflow is not recommended for a newbie but all those with some experience shouldn't find it too tricky. Good luck and thanks again to all the people who made it possible, Ken, Hayden Landis' SIGGRAPH paper etc.

Andrew

PS If anyone wants to hire a TD in London let me know Will render for food.

[cleverimage]

hi there, thank you very very much Andrew W for your tutorial, i found it very very interesting, thank you for taking the time to write it.

[flash]

Can someone HELP me on one thing about this tutorial i dont get?

I am having troubles to get that secondpass rendering to work, i am not sure what and where i need to fill like the tutorial says: "Finally you need to add an additional output variable to your PRMan render globals. Add "outOcc" as the type"
I tried to type "outOcc" to the render globals secondary page, on the "type" section.

If someone could help me and tell me exactly what sections i need to fill in render globals secondary page to get also that outOcc file out.

I can render bendnormals data fine, but as soon as i fill secondary pass sections i get error during rendering that has something to do with "unknown display mode or" or something like that...

Here is a picture if that helps me to get any answers.
Sorry i am very new with renderman.

[aubec]

Type = tiff or image format of choice
mode = outOcc

Sorry about that folks. Rereading what I wrote I can see how you could have misunderstood it. I'm making notes of bad descriptions like that and I'm going to update the tutorial when I can get access to PRMan again to render some better examples and screenshots.

Once apologies for not being clearer in my description.

Andrew

PS You also need to declare the new output variable in the rib file. Look up info on this in the MTOR online docs. They explain what to do. I had assumed people would know to do this. That's something else to add to version 2 of the tutorial! Oops again.

[oliverj]

I would like to know the rendering time of the ambient occlusion pass on the Athena showed in the tutorial.
Is it mentioned somewhere ?

[aubec]

For the athena pass at 1K the ambient occlusion renders took 10 mins a frame including the time to generate the shadow maps. The subsequent image renders utilising the ambient occlusion and bent normal data took less than a minute at 1 K with good quality settings for the pixel filtration.

Hope that helps.

Andrew

[subjectego]

I'm guessing that outOcc would be of type "varying color" (correct me if i'm wrong). In this case it would be pretty simple to, without haveing to generate rib prior to adding this variable to it just create a RibBox in slim, call it "declare" and fill it with this line:

Declare "outOcc" "varying color"

the name "declare" for the template is the built in way of letting RAT know to append the rib it generates and add this line in the place where other things are declared.

I would like to know, however, how did you set up your "environment map fill" in your Athena render? I have mine as [txmake /home/me/shader/hdri/stpeters_probe.hdr -envlatl ] Is "-envlatl" what you used?

Also for "environment space" how did you set it up? "_environment" or? I had slim default to "_environment" but that I'm getting the St.Peters light up my object kind of screwy What other options are there? world and _world_lefthanded don't work that good in my case either

Also, I've seen on your website you mentioned that you've created a toon shader for Prman. Any chance you could post that on your site or is that an in house thing of your studio so you can't post it? I wasn't too happy with the Deathfall tutorial since it requires you to use find edges algorithm in Photoshop or Shake and that didn't work that good for me even if I rendered my scenes 3 times the size for the "pre-findEdges" image, which kind of defeats the purpose of using this technique at all.

Thanx a lot!

[subjectego]

Also, Andrew, when using your "occlusion_constant.slo" as backplane (by naming the shader template "backplane" in slim and enabling backplane in mtor globals) I don't seem to get any alpha. Moreover I get the background to be white unless I set "greyLevel" to black, in which case I do get black background but no alpha still. How would you need to change "occlusion_constant.sl" in order to make it output alpha channel info at +1?

[aubec]

OK, one answer at a time

1) Yup, you're right it's a varying colour, so you've done the right thing with the declare RIB box.

2) I use _environment and a latlong environment map, so I'm a little puzzled why your images are coming out weird. Maybe try making the latlong map in a different way? Is the txmake process doing strange things to your image, like not stretching it correctly?

3) I'm afraid my toon shader is proprietry so I can't give you the .sl But, I can tell you it's based on faceforward tests and using the diffuse() value to drive a ramp colour. That should put you on the right course...

4) You don't need alpha channels for your ambient occlusion passes. I think I may have screwed up the shader I posted and put the stuff the wrong way round. The bent normals should have a black background and the ambient occlusion should have white background. Apologies for that. Should be really easy to amend that in the .sl though.

Hope that helps.

Andrew

[subjectego]

hey andrew, thanx. What I thaught was the problem with env maps is not in fact a problem just on one particular geom it looked screwy

One prman question though, say I wanted to reverse the _environment picture in y would there be a command or would I have to rotate it manually in an imaging package?

thanx

[aubec]

I usually reckon it's best to do tweaks like that in an imaging package just to minimise the amount of maths the shader has to do. It makes sense that the fewer tweaks the shader has to make to an image the faster it should render.

[firewalkwithme]

-Andrew

First up thank you VERY much for your nifty occ shader
and workflow, i have finally started to get some nice(efficient) occlusion renders.

One question though, that light dome you use for
gathering the dmap info, how do you generate it?
I'de like to do some comparison tests for qualityspeed vs. no. of lights used and i need to keep the sphere shape consistent throughout.
I suppose its a pretty simple script setup, but if you have something already i'de appreciate you throwing it at me.

Also , if you have any conclusionstips regarding the
workflow i'de love to hear 'em.

thanks again
g

[aubec]

Hi, sorry I've taken a while to get back to you on this but I'm finishing up work on a movie at the moment and it's all been a bit hectic. I'm afraid to say that the lightdome I used I made "by hand" with the calculations made with a calculator, a pencil and a piece of paper, though I think there is a light dome MEL script on Highend that might suit.

As to workflow tips... Well, I'm working on some revisions at the moment, so hopefully there'll be a version 2 tutorial and shaders soon. If anyone makes any exciting changes to the stuff I wrote, send 'em to me via my site and I'll try and incoroprate them. I know of at least 3 features which are currently using my shaders so maybe the big brains at those companies will have some bright ideas...

Anyway, I'm glad these little tools have sped up your workflow, and as I say Version 2 is in the pipeline. I'll post here when the new page is up and running. I'd be really interested to see any renders produced using these tools, maybe you could e-mail them to me? It might be nice to include a little gallery of work that other people have produced to show what it's capable of. If I get enough feedback (ie pictures) from you lovely people then I'll put a gallery page on the site.

Cheers

Andrew

PS there are a few inaccuracies and omissions in the current tutorial which I'll also iron out.

[firewalkwithme]

ok thanks

i personally am having a bit of trouble with flat surfaces
(some nasty artifacts that dont wanna go away)
but i'll just wait for your update.....

I'll be gald to send you some nice pics as soon as they
are worth showing......

g

[aubec]

By flat surfaces, I'm presuming you mean the floor? I have to say that one limitation of this technique is on big flat surfaces, in much the same way as environment maps don't cope well with flat surfaces. My workaround has been to render the floor occlusion layer as a seperate layer and blurring the shadow-maps for that layer render A LOT (like 25 or something). By doing this, you can get nice, soft, contact shadows which don't have that weird stripy atrifact in it. I'll document this better, in the much promised, but as yet unwritten, v.2 ambient occlusion tutorial. Hope that sorts that for you.

Andrew

[firewalkwithme]

-Andrew

Sounds good to me... i'll check it out.

thanks alot

g