Render Passes 2009

[itriix]

wahoo again. it was the matte mults on the switch matte that was causing the eating of the edges up. GOOD CALL DUDE! Thanks... Hmmm, that's very odd the .exr's won't open. I'll try playing around with the "object_id" myself. I'm currently rendering it out right now. It seems to be taking almost about the same time as "rendering" my beauty pass... might be because the pixel samples are at 12 x 12 like you're example Anyways, it had me thinking. Would it be possible to "shove" this into my actual "shader" and have it output like an "aov"? Or is that not possible because it HAS to be plugged into the "output" color/opacity/alpha slot?

Second, Did you take a look at the "shadow" pic that i posted? Doesn't that look a bit "strange"? "the shadow on the can" seems verrrrrrrrry odd. any thoughts about that one?

Third, when you were talking about the "illuminance loop" before, i'd like to attempt to get that set up in my shader so that i can break out the shadows... Mainly for learning purposes. If it's "overly" complicated, then I can research it on my own. But if you're up for a little bit more teaching, i'd love to learn if you got the time

Fourth, I'm curious about why the exr's aren't working out for you there. When you say that it doesn't open them. Does that mean you don't see anything when you file-in them? Or, you get some error when you try to load them in? One thing popped in my head, and (i'm sure you noticed this but thinking just in case)... The exr's were frame 50... Were you on frame 50 in you're timeline? (i haven't loaded single #'d frames before in shake, in quite a long time) so i'm not sure if that would cause the issue or not with it. Just has me a bit curious though as to why it won't load up there!

Thanks again for your time!

Cheers,

Jonathan

Edited November 11, 2009 by itriix

[sam.h]

Shake just seg faults when I try to load it

[itriix]

Shake just seg faults when I try to load it

LOL... That sounds sketchy!!!!!!!

Edited November 11, 2009 by itriix

[sam.h]

LOL... That sounds sketchy!!!!!!!

will reply tomorrow, for now, sleep!

[sam.h]

wahoo again. it was the matte mults on the switch matte that was causing the eating of the edges up. GOOD CALL DUDE! Thanks... Hmmm, that's very odd the .exr's won't open. I'll try playing around with the "object_id" myself. I'm currently rendering it out right now. It seems to be taking almost about the same time as "rendering" my beauty pass... might be because the pixel samples are at 12 x 12 like you're example Anyways, it had me thinking. Would it be possible to "shove" this into my actual "shader" and have it output like an "aov"? Or is that not possible because it HAS to be plugged into the "output" color/opacity/alpha slot?

Yup, to export to a secondary pass just use a Parameter node, set 'export' to 'always', turn on ivisible, name it what you like and add it to your mantra ROP.

Second, Did you take a look at the "shadow" pic that i posted? Doesn't that look a bit "strange"? "the shadow on the can" seems verrrrrrrrry odd. any thoughts about that one?

It does.... I *think* your shadow if going white (white being opaque) where there is no diffuse contribution ... ?

Third, when you were talking about the "illuminance loop" before, i'd like to attempt to get that set up in my shader so that i can break out the shadows... Mainly for learning purposes. If it's "overly" complicated, then I can research it on my own. But if you're up for a little bit more teaching, i'd love to learn if you got the time

I thought about this more and for it to be useful you would probably need per light exports, there are a few topics on the forum about it, personally I have never tried it ...

Fourth, I'm curious about why the exr's aren't working out for you there. When you say that it doesn't open them. Does that mean you don't see anything when you file-in them? Or, you get some error when you try to load them in? One thing popped in my head, and (i'm sure you noticed this but thinking just in case)... The exr's were frame 50... Were you on frame 50 in you're timeline? (i haven't loaded single #'d frames before in shake, in quite a long time) so i'm not sure if that would cause the issue or not with it. Just has me a bit curious though as to why it won't load up there!

Thanks again for your time!

Cheers,

Jonathan

Sorry about the late reply, I have been distracted by ... other VERY important things :

[itriix]

Sorry about the late reply, I have been distracted by ... other VERY important things :

no need to apologize! that looks well worth the time

as for you're comments:

Yup, to export to a secondary pass just use a Parameter node, set 'export' to 'always', turn on ivisible, name it what you like and add it to your mantra ROP.

yeah, i don't know why i didn't just try it. i'm familiar with exporting params but for some reason my brain wasn't working i guess. thought that it might not work! i'll add the "object_id" inside the shader then and export it all at once as an aov

It does.... I *think* your shadow if going white (white being opaque) where there is no diffuse contribution ... ?

Hmm, well I feel like the "backside/in the example pic/the left side" of the can, should be having shadows (so that would be WHITE) but instead it's pure black. And the shadow has a "stepped" look to it on the can. *makes me think something is wrong with the can. I looked at the normals and they are pointing outward *as they should be*...) not really sure what else would cause it to do that though.

I thought about this more and for it to be useful you would probably need per light exports, there are a few topics on the forum about it, personally I have never tried it ...

ah i see. well, i've been researching it today. i'm going to take a stab at (maybe a lil later though, for now, maybe i should just turn off shadows in one take, and then create another take, where i turn on shadows, and add the shadow matte material)... sound like a good idea?

i threw together a very cheesey basic background scene just so i could start playing with the shadow pass, it's in this topic: This isn't even using passes though, was just the beauty pass! LOL... since I was still thinking about how i wanted to handle the shadows.

There was a topic posted today about object_id. I referred him to you're post... but he was using like 50 pieces of geo

As for trying the object_id in comp... If you remember I was saying I was getting some "purple" overlapping areas in the object_id pass. Seems like it's okay in comp, as long as i use an OVER instead of just adding the results together. It made me curious though, what if I had like 5 objects? Would I just save them as different combinations of 1,0,0... 0,1,0... 0,0,1... 1,1,0... 1,0,1... kind of like this????

Anyways, in the meantime, i've learned some "strange things" about the delayed load shader. As I was pointing out the other day, i definitely thought something wasn't working as expected. I found out it seems like using a GEO ROP, doesn't fully save out the material "correctly"... I don't know i'm a bit confused on "what's happening". anyways. I tried the mantra archive, saved out the particle sim geo to disk and the material .ifd... I used the delayed load with it pointing to these two disk files, and the color was in fact correct. it worked correctly! BUT this lead me to a NEW problem. To get bulky sims, and easy, I would just create one popnet, use a WEDGE ROP to tweak the SEED param in the popnet, and then use the GEO ROP with the WEDGE variable. That way I could get multiple small versions of the same sim with a couple nodes. THE MANTRA ARCHIVE doesn't seem to be accepting the WEDGE. Therefore, I can only get ONE SIM saved out from it. I tried to create an OBJ level "seed" param that controlled the sop level seed, then point the WEDGE to the OBJ level seed, but the mantra archive STILL would not accept the wedge. So either i'm doing something wrong or it simply just can't do it like that. So that sucks. Lastly, ALSO, the mantra archive doesn't seem to be supporting the SWITCH SHOP (when i was telling you how i had the main shader, the object_id, and the shadow shader) all plugged into the switch shop, and i could switch between them with the RENDERPASS variable in the pre/post render scripts in mantra rops. Anyways, yeah, it seems as though the material .ifd that is created by the mantra archive is only "remembering" the current input...

Oh well, as for now, that leads me to thinking if i'm doing bulky sims that aren't being point rendered and require a material, then i'll just have to load them in from disk and render as normal without using the delayed load. I hope maybe i'm just doing something wrong, but i haven't heard/learned anything that states otherwise!

Cheers,

Jonathan

Edited November 13, 2009 by itriix

[sam.h]

Hmm, well I feel like the "backside/in the example pic/the left side" of the can, should be having shadows (so that would be WHITE) but instead it's pure black. And the shadow has a "stepped" look to it on the can. *makes me think something is wrong with the can. I looked at the normals and they are pointing outward *as they should be*...) not really sure what else would cause it to do that though.

Check out this example:

shadowMatte.hip

The shadow matte isn't apply shadows to the 'darkside' of the sphere in the middle because there is no diffuse contribution on that side to shadow.

ah i see. well, i've been researching it today. i'm going to take a stab at (maybe a lil later though, for now, maybe i should just turn off shadows in one take, and then create another take, where i turn on shadows, and add the shadow matte material)... sound like a good idea?

i threw together a very cheesey basic background scene just so i could start playing with the shadow pass, it's in this topic: This isn't even using passes though, was just the beauty pass! LOL... since I was still thinking about how i wanted to handle the shadows.

There was a topic posted today about object_id. I referred him to you're post... but he was using like 50 pieces of geo

As for trying the object_id in comp... If you remember I was saying I was getting some "purple" overlapping areas in the object_id pass. Seems like it's okay in comp, as long as i use an OVER instead of just adding the results together. It made me curious though, what if I had like 5 objects? Would I just save them as different combinations of 1,0,0... 0,1,0... 0,0,1... 1,1,0... 1,0,1... kind of like this????

I replied to that topic, the method you mentioned wouldn't work, you would get issues where the objects intersected / overlapped. Because ... say you had a 1,0,0 object, 0,1,0 object and a 1,1,0 object, where the first two overlap and antialias together, it would appear to be a 1,1,0 ... did that make sense?

Anyways, in the meantime, i've learned some "strange things" about the delayed load shader. As I was pointing out the other day, i definitely thought something wasn't working as expected. I found out it seems like using a GEO ROP, doesn't fully save out the material "correctly"... I don't know i'm a bit confused on "what's happening". anyways. I tried the mantra archive, saved out the particle sim geo to disk and the material .ifd... I used the delayed load with it pointing to these two disk files, and the color was in fact correct. it worked correctly! BUT this lead me to a NEW problem. To get bulky sims, and easy, I would just create one popnet, use a WEDGE ROP to tweak the SEED param in the popnet, and then use the GEO ROP with the WEDGE variable. That way I could get multiple small versions of the same sim with a couple nodes. THE MANTRA ARCHIVE doesn't seem to be accepting the WEDGE. Therefore, I can only get ONE SIM saved out from it. I tried to create an OBJ level "seed" param that controlled the sop level seed, then point the WEDGE to the OBJ level seed, but the mantra archive STILL would not accept the wedge. So either i'm doing something wrong or it simply just can't do it like that. So that sucks. Lastly, ALSO, the mantra archive doesn't seem to be supporting the SWITCH SHOP (when i was telling you how i had the main shader, the object_id, and the shadow shader) all plugged into the switch shop, and i could switch between them with the RENDERPASS variable in the pre/post render scripts in mantra rops. Anyways, yeah, it seems as though the material .ifd that is created by the mantra archive is only "remembering" the current input...

hmm, a simple thing like this:

delayedLoad.hip

is as far as I have gone with delayed load... :/

Oh well, as for now, that leads me to thinking if i'm doing bulky sims that aren't being point rendered and require a material, then i'll just have to load them in from disk and render as normal without using the delayed load. I hope maybe i'm just doing something wrong, but i haven't heard/learned anything that states otherwise!

Cheers,

Jonathan

Edited November 13, 2009 by sam.h

[itriix]

The shadow matte isn't apply shadows to the 'darkside' of the sphere in the middle because there is no diffuse contribution on that side to shadow.

i went back and looked at the "shadow pic" of my coke can and particles to compare and contrast it to the one you posted... and what you said about the back side not "being shadowed" because it doesn't have any diffuse contribution... BUT on my shadow pic, something still bugs me out a bit, and that's that the particles on the "back side" of the can are shadowed, but the "backside" of the can itself isn't? only thing i can think of is, is it because the "particles" are ON TOP of the can at that point, and are recieving diffuse lighting from the "top"? therefore they are receiving diffuse contribution, and therefore getting shadows. whereas, the back side of the can is "hidden" from all of that because of the particles, and therefore isn't getting any diffuse contribution, and therefore no shadow?

I replied to that topic, the method you mentioned wouldn't work, you would get issues where the objects intersected / overlapped. Because ... say you had a 1,0,0 object, 0,1,0 object and a 1,1,0 object, where the first two overlap and antialias together, it would appear to be a 1,1,0 ... did that make sense?

gotcha, that's kind of what i figured. so how would i break out something like 5 objects?

as for you're dl

is as far as I have gone with delayed load... :/

i took a look at you're hip file. yeah, in that example delayed load is fine because you're rendering as points, and not only that, you're just ROPing out one sim. It's quite common though to use a wedge and rop out multiple sims of the same sim but just change the seed, that way you can bulk it up. it's much more efficient than just doing one LARGE BEAST of a sim. however, seems that delayed load doesn't play nicely in that respect! Don't render your particles as points and have a material applied to you're particles and you'll end up with the "material" problem I mentioned, where the render looks different than not using the delayed load. "since i wanted to have "specular highlights" on my particles for the coke can, this required me to "not" render as points. (i had tried to do "texture baking" for the coke can, and was going to inherit the colors from that for the particle color (which wouldn't really be correct anyways after the particles actually started moving through the scene, because they would still have the color of the "baked texture", but it was a thought and something i wanted to try, because that way i could render as points and still get the "lighting" and "color" to look just like the coke can when they were emitted. However, this wasn't really acceptable, because i needed that "interactive" lighting as they traveled through the scene anyways. So I scratched that. However, it still seems to be a problem though, getting the texture baked out correctly (just for my own knowledge)... In fact, the "SHADOW" was doing something weird in the baked texture. It looked as though it was coming in on the opposite side of the can.

Oh well, i'll figure out a workaround i hope for now, i'll just stick to the oldschool method of just rendering out as normal with no delayed load in this type of instance

Edited November 13, 2009 by itriix