PDA

View Full Version : MentalRay for Maya production questions


royter
07-11-2011, 10:34 AM
whats the main difference between : maya motion vectors (via maya software) - MR motion vectors (via render passes) - MIP motion vectors (via production shaders)? Does any method support moving fluids or can blur reflections?

whats the correct workflow of rendering motion vectors in MR via render passes and using RSMV (real smart motion vectors) or other plugins for after effects? is there anything particular i should know?

when composting the frames of a scene that contains a fast moving object or camera, the final movie (not playblast) is always clunky and not fluid. The only solution i found is to use the "animation" QuickTime codec because all the others even the JPEGs and H264 gave clunky movies, but even the animation codec (witch gives big files) was not perfectly fluid and is not very practical. here u can see an example (http://vimeo.com/17995931) (car sliding in the desert ), any ideas?


how can i optain this type (http://features.cgsociety.org/story_custom.php?story_id=5655&page=2) of bokeh effect? tried it in MR4Maya but even if i boosted the bokeh effect ( map (http://3.bp.blogspot.com/_pAxkf2lRVRw/SJcwUNXlRXI/AAAAAAAAADs/LB0Soomm7Tw/s400/bokehMapAdditive.png)) it gave very grainy results (even with high samples) and its not near the result in max?

how can i obtain a lens flare and glow effect for my mental ray sun? the fx glow effect dont apply to the maya directional light and is only supported in the maya software?

how can i genrate a z depth pass that takes everything into account even fluids? what the best way in MR4Maya? and how can i add haze in post if i have a clean zdepth pass?

how can i generate a chromatic aberration effect in post without having to shift the RGB channels and have those empty lines at the frame boundary?

how do i composite my contribution maps? witch layering mode should i choose?

how can someone use an hdr in a scene where a car is moving from point a to point b? wouldnt the static hdr give weird results? whats the most practical way to stay convincing?

in the fast sss shader, why does the shader comes out black if diffuse color (not weight) is set to black, i only want to see the front and back sss?

InfernalDarkness
07-11-2011, 11:08 AM
You're asking a ton of questions with no real specific aim, but I understand why and where you're coming from... You're going to get a plethora of disjointed answers, but I hope we can help once again. I'll focus on what I know and work with, and leave the speculation on stuff I've only dabbled in to others with more experience.


how can i genrate a z depth pass that takes everything into account even fluids? what the best way in MR4Maya? and how can i add haze in post if i have a clean zdepth pass?

You cannot. Fluids too? Good luck even getting decent z-depths in regular stuff with complex shaders. The trick is to get good enough z-depths; don't be too much of a perfectionist, especially if you're doing animations. Break your scene down as simply as possible; get used to writing out separate z-depth-only renderable scenes from your main scenes. The Puppet_z shader is your friend.

As for adding haze in post? Simply select your z-depth alpha channel, make a new layer, and add a gradient. This is the easy part. Use a killer cloud or sky photo. The process is the same; use the z-depth to control both the cutout and opacity of your haze, then play with layers. Screen is generally helpful, but Soft Light also rocks sometimes. Overlay is a great starting point. Use your "Refine Edge" tool a lot to help with selection issues.

Fortunately, I haven't had to worry about the creative aspects of haze and fog since even getting a usable z-depth out of Maya would require a whole separate work-schedule, career path, and paycheck, so all my recent scenes are completely devoid of depth and focus, which makes life a lot easier. One of the beautiful things about mental ray - after awhile, you don't even care that it sucks anymore. You just make crap and call it done; best you can do, sometimes. If the clients don't care, it's wasted energy anyway.

As for personal artwork... I just cry myself to sleep when I get to the z-depth stage of a project, and in the morning, I usually feel a little better.


On HDRs and image-based environments: This is a workaround technique, and always will be. Without and parallax or spacial information, you pretty much just count on the fact that there's a finite amount of pixels in your final image anyway, so it doesn't matter that the reflections are fake. If they look good, that's all that counts. I used to be a scene-purist; modeling even crap that's off-camera to enhance realism. But at 300dpi, you can't tell the difference, so I just wasted a ton of time I should have been wasting trying to learn z-depth instead!

Bitter
07-11-2011, 05:32 PM
Z-depth shouldn't be hard to generate, even with a normalized (0-1) pass in Maya it's fine.

I hear a lot how it's difficult but I don't understand why. Can someone explain?

Is it about being unfiltered? It should be unfiltered. It should also be floating point or the lack of precision causes artifacts.

Fluids cannot be used in z-depth because the color generated is a result of traveling through the fluid and combining for a final result. An actual begin and end boundary of a fluid isn't very useful and makes a z-depth value a bit hard to decide on.

InfernalDarkness
07-11-2011, 06:12 PM
I'm totally with you on the fluids not being z-depth-able; that make sense to me. What I struggle with is complex shaders such as trees with cutout opacity in the leaves; thus far only the puppet_z shader works properly for this, and it's still a nightmare to set up in a complex scene unless I start the scene with the z-depth in mind from the beginning. This tends to get frustrating for me and bogs down "artistic flow" a lot. I'll admit I need more practice, but here are my results (http://forums.cgsociety.org/showthread.php?f=87&t=941774) testing various methods to achieve a usable, relatively painless z-depth in mental ray.

Not everyone seems to struggle with this one Bitter; most likely, I just suck at passes and need to take a course or tutorial on the topic!

Bitter
07-11-2011, 06:53 PM
Ah, ok. Cards.

There's a mip card opacity shader for using a card and z-depth.

kiryha
07-11-2011, 09:09 PM
If you will get answers on all you questions, Royterr, it would be one of the most exciting thread in rendering, like "forests in MR":)

You asked about motion vectors (post MB) and 3D rendered DOF, that is a bit confusing. Why you deside do "honest" DOF, and MB in post? Why not vice versa, or do both MB and DOF in post, or do both during render?

Since this effects take long time to render (yes, unified sampling in 3.9 is promising to be good solution, but i have no chanse to try it now) may be it would be better to look for ways how to do both of them in post? The bokeh effects wich you try to get is easy task for post.

Frankly, i never had problems with motion vectors themself(exept MV for fur), there are a lot of ways to create and use them, RSMB works fine with standard maya MV pass(2D motion vectors), but combine MB and DOF in post seems to be realy dificult task. I made a lot of research in this topic, i even try to render mentalray image in renderman as camera projection on scene geo with MB and DOF, but i never get predictable and clean result. From the other side, all artifacts that appears after doing blur and dof in post seems to be not to much noticeble in animation. Here is (http://vimeo.com/13125887) my favorite work with post process DOF and MB, there are a lot of artifacts, but it looks acceptable even in cinema.
If you have puppet mega_TK shaders, look into docs, there is grate example of motion vectors workflow, for me its works fine, but you have to deal with node based compositing application such as Nuke(if u wish to go deep to compositing- you have no other way- doing this in afterFX is pain in the ass).

Any way there is no one "production" method to do CG, a lot of desitions deppends of many factors and you have to build you own workflow, wich is suit your tasks and resourses. I had a lot of pain with MB and DOF cos our studio often render "final" shots cople times per day to satisfy all clients wishes, and it would be absolutley impossible to do this effects in 3D even if we had huge render farm.
So may be it would be more better to build good pipeline for particular task, but not to try create general workflow for all cases?

mercuito
07-11-2011, 09:13 PM
What I struggle with is complex shaders such as trees with cutout opacity in the leaves; thus far only the puppet_z shader works properly for this, and it's still a nightmare to set up in a complex scene

MentalCore handles this well. If you use cutout opacity on a shader your depth and all other passes will behave how they should, and even occlusion can obey the cutout shape too. Fluids on the other hand are a different story. Sorry not trying to hijack with a product placement, but these kinda reasons are exactly why I development it, I know your pain!

Bitter
07-11-2011, 09:14 PM
mental images forum (http://forum.mentalimages.com/showthread.php?8552-Mental-Ray-4-Maya-productions-questions&p=34854#post34854)

Already mentioned mental core. :-)

kiryha
07-11-2011, 09:38 PM
For now mental core -the most promissing solution for maya and MR. I dinn`t test it too deep, but only reading docs was enogh to understand this. You made awesome stuff! When it will be finaly released?

InfernalDarkness
07-11-2011, 09:39 PM
MentalCore handles this well. If you use cutout opacity on a shader your depth and all other passes will behave how they should, and even occlusion can obey the cutout shape too. Fluids on the other hand are a different story. Sorry not trying to hijack with a product placement, but these kinda reasons are exactly why I development it, I know your pain!

@Mercuito: I've been following your body of work from day 1 here, since you first posted the Core utilities! Alas, I'm stuck with Maya 2009 still, so it's not an option for me yet. I know you feel my pain, just making sure Royterr also knows that I feel his pain! We have both tried numerous methods for z-depth renders; it's not as though we just started testing such renders yesterday. I've been struggling with z-depth in mental ray since 2005, myself. This simple roadblock for me has basically ruined my artistic progress all by itself, as I'm unable to produce accurate atmospheres in post anymore. Not your fault Mercuito, just explaining that I mirror Roy's frustration with the technique.

@Bitter: I agree z-depth should be very simple and straightforward, as it is in every other application except Maya. It's a piece of cake on simple scenes, but once you start making huge landscapes with billions and krillions of objects and shaders, z-depth is no longer a "piece of cake" for me... I can very much understand Royterr's frustration and echo it, but I do appreciate your positive words on the topic too. Perhaps in 2012 it's more realistic or maybe it just works after 2009.

Bitter
07-11-2011, 10:06 PM
For z-depth on a cutout you should either use the cutout opacity or the card shader and plug your other shader into it. Then apply it all over the place.

z-depth is based on the intersection with geometry. This is how it's done for every renderer I know of. The cutout and card shaders tell the ray to ignore that part and keep going.

If your other shader is already built then adding this and just saving it somewhere will save you time later.

InfernalDarkness
07-11-2011, 10:44 PM
It's okay if you didn't read the other thread detailing many of the procedures we've tried. M0z and MasonDoran were very helpful.

The cutout and card shaders tell the ray to ignore that part and keep going.

Yes, they are supposed to do this. I am unable to get cutout opacity or the card shader to have any effect in my z-renderings, thus the shift to the puppet_z shader, which works well but requires a separate render entirely of course.

Anyway, no sense beating a dead horse. I'm not trying to hijack Royterr's thread, but we have both tried all of the techniques presented for z-depth. If he's still not getting proper results, that's the breaks, but I'm not either... So if you're frustrated with how obnoxiously complex such an obscenely simple render should be, Roy, I'm right there with you my friend. I can get a workable z-depth, but not on big projects, and not in this lifetime. (end crybaby session)

Bitter
07-11-2011, 11:47 PM
Ah, with raytracing it gets odd in the newer versions. I'll report that but has worked for me in the past. And still works with rasterization oddly enough.

Image is with the depth remapped buffer in Maya.

kiryha
07-13-2011, 04:10 PM
Continuing post process DOF topic:
Some guys create amazing camera. Using it, you are able to change focal point after shooting you pictures.
They call this technology Light field. See details at www.lytro.com

So, going back from real world to 3D, may be there is a way to implement same technology to render software to get at least cool DOF without any limitation and artifacts. Say, new vertion of Z buffer- light field pass.
Anybody have some knowledge on this theme?

royter
10-15-2011, 09:49 AM
thanks alot guys for your feedback.
it seems to me that there is only work arounds, and hybrid solutions for a zdepth pass in mr4maya, no unified syandard safe method like vray4maya.

InfernalDarkness
10-15-2011, 06:09 PM
@Royterr: I disagree entirely. Just use the depth pass, or the depth remapped if you prefer. It works fine, just took me a long time to figure it out!

royter
10-15-2011, 06:20 PM
does it take motion blur and chromatic aberration into account?

CGTalk Moderation
10-15-2011, 06:20 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.