Mental Ray: worth learning?

[Kerro Perro]

Hey if it works for you that's fine! I just personally like to use dramatic lighting with spots and such and that has always flickered in fg so far...

PS i'm crashing out - spent all day building an ikea bed so i might continue this conversation tomorrow ;)

[Nizar]

Thanks for so many info Rez007 I will try tomorrow and post here some results. I have an entire month for learn mental ray.

About Gustavo tutorial, I have the first one (demystify Render Tree). I think is a good tutorial and I like how he explain his workflow.

Falam@ modo render can be mastered in about a month (or lesser). Modo render output is very good IMHO. The book (a real bible) you mentioned is out of print, but is also outdated (correct me, but don't talk about unified sampling and architectural material)

[Rez007]

Hey if it works for you that's fine! I just personally like to use dramatic lighting with spots and such and that has always flickered in fg so far...

PS i'm crashing out - spent all day building an ikea bed so i might continue this conversation tomorrow ;)

Thats cool, yeah my setup was what I had to do for a client and the method seemed to work for me. I guess it might depend on the overall scene and its final intentions. I think brute force mode would be the all-around best solution with Unified, but that would take some time, and I am not to positive on that one...

[remydrh]

Typical rules of usage for Unified Sampling are:

Materials:
AO (in a material such as mia): 4-6 samples
Reflection: 1 (for insteresting/textured surfaces) for clean/perfect surfaces you may need 6-16, especially with high contrast environments
Refraction: 1 (however, refraction gloss is not as easy a 'diffuse' effect as reflection over long distances. For frosted treatments I suggest some translucency)


Area lights:
High samples: 4-8 (fewer rays are better with DOF and Motion Blur) This applies to user_ibl as well unless the HDR used is very complicated

Depth Of Field lens shader: 1 sample

The more DOF or Motion blur, the lower your local samples should be.

For product shots, environments, exteriors, and set recreation with HDRs, I would suggest native (builtin) Ibl, these are strings options. Or second choice, user_ibl. Secondary GI can be Final gather (FG) but greatly reduce the needed accuracy as the IBL is handling most of the light in these cases. (Each situation is different so play around some.) And use the fgshooter mip shader if possible to reduce FG variance in animations. It makes the solution more automatic to render with less tweaking while keeping the speed improvements of an interpolated technique.

If you render a sphere on a plane with Unified Sampling you will find it's slower than regular AA. Trivial scenes do not gain improvement from Unified except possible noise reduction. Try more complex scenes you might have ignored doing before.

Nice (non-blog) example: http://forums.cgchannel.com/showthread.php?t=15050
Newcomer (<20 posts) alert: please use the URL tags - HB

Above: 30 minutes at 720 HD with mental ray and Unified Sampling

[Soth3dlife]

In my opinion you should not really worry about that. Mental ray with unified sampling is so easy to set up, actually I find it quite easier than Vray ( perharps not as effective sometimes ). From my point of view I just care about:

1-. Unified Sampling

2-. Mia_material_advanced

3-. Spot, direction and portal lights

4-. Most of the time I use FG so depending on the project I use HDRI.

Guess what, I do the same in Vray. What I find easier in Vray? that all the passes and everything within Vray work perfectly together and there are presets for everything. For example, yo can just set the passes with Vray much easier than with MR. That's my point of view. Other advantage of Vray, by default is faster and with the proxies you con optimize renders quite a bit.

I'm just saying that you are most likely do the same in both of them...

[Maximus]

I have no idea why someone would bother learning Mental Ray nowadays. It has everything (no it really doesnt) but it has it old, not updated, poorly documented, not implemented, with a 10 year old workflow that really goes against softimage workflow itself.

Some aspects.

It seems like Unified sampling is a life saver, is it? No it isnt, it might be nice but mental ray has really other problems rather than this. In many scenes Unified can be very very slow compared to old method and here you have people say "yes depend on scene, depends if you know how to use it, depends how you set it up, depends on your materials, depends.. depends...." with Mental ray it always depends on something, and the bottom line is: Mental Ray is an insanely over complicated technical render engine that didnt get optimized in ages, Developers left it in pieces and gave the users the burden to figure out how to use it.

Bottom line is the question everyone does: "so when the hell to use unified and when not?" Answer: Figure it out by yourself wasting hours of render tests and previews. Welcome to mental ray. Everything in MR works like this.

Lets take some examples, it lacks of a modern GI (how many years is it now? And still nothing has be done, this say enough about the future development and the amazing trust you put on their developers. Really feel to invest time in something that doesnt get updated? check out other engines updates and timeframe for those). Even Cycle has a GI.

OH yes i saw some friend of friend of another friend of someone who did an amazing animation flicker free. Yes you can do flicker free animation with Mental Ray but for some reason always on other people scenes, not in yours. The amount of tricks/optimization/time wasted to make it work is simply ridicolous.

The whole physical light system is complete garbage. Portal lights/environment sky/ photographic exposure/cm factor this crap is complete mess. You cannot export portal lights/sky/environment and portals with correct lighting because portal lights require photographic if not they blown everything out, so you have to think about this in compositing (another damn step to fix a complete mess that was poorly designed and more poorly implemented) The sky doesnt get proper gamma so you need to replace it, its retarded, a nightmare and we are in damn 2012 not in 1990 anymore.

Subsurface scattering, the quality is no doubt amazing but lets see.
How many F***in nodes we have? skin1 skin 2 fast, miss_physical_ miss fast miss sss sss miss missy missisippi. Wtf? Clean this crap already? Make a damn single SSS updated and modern, speed efficient, GI efficient and get done with it. Try to use GI with SSS and see the render times, open the script editor and activate the log info, see what the hell is going on under the hood. It does a freakton of passes taking insane amount of time to render a damn sphere on a plane. Lightmap? come on the whole system was fine in 1990. Not anymore.

SSS was designed to work without GI, because in the old days people were used to light stuff without GI. Now its the complete opposite, thats why Arnold is kicking everyone asses.
Thats being modern. yeah Mental Ray can do it!!! at what cost?

Lights: oh my.... you have to connect at least 3 nodes to have the same light a standard render engine has. Physically correct (always laugh at this when speakin about mental ray cause it has nothing designed to be physical). Lets see, you have to connect: blackbody (because you have to put color temperature for your amazing physical light) > physical light > light node.
3ppg while you should just have one, with all attributes, kelvin presets, area lights etc.
I have an amazin character i created lets light it with physical light and blackbody. Get light > Spotlight > render tree > connect nodes. Oh Wait! Physical light should used with point light, not spot light. Too bad no-damn-one tell you this untill you end up in forums.

This is a glimpse of mental ray world, and i spare you about that crap GI system not even worth to be called Global Illumination system, actually i will tell you because i know lots of people are laughin hard at this, so i'll give you more:

IBL > Image Based Lighting > oh shi! You would think hey this gonna be amazing, will work like Vray domelight!!! NO ITS NOT.
They forgot to add bounces, so both IBL/usreIBL are without GI, meaning have fun to light a scene that requires bounces, meaning its just another fake.
See? They keep adding bandage over and over. UserIBL and rectIBL was added recently, lets see the amazing workflow in 2012.

You have to create a spotlight > Connect the useribl node to this spotlight > open the spotlight PPG and set the shape to "user" > at this point you are like "oh my! that wasnt hard, so fast!!!" NO you are not done yet. You need to open your render tree environment node and connect here another user ibl because otherwise the reflections wont work.
Seriously? is this a joke? I can assure you that i'm just scratching the surface of the mess this engine is.

I can go on for hours, but the bottom line is, you will find people who worship mental ray, i spent 6 years into this engine and i wish i did not. I'm using it because its inside Softimage and for what i do its crap but since i threw at it so many time i no longer have the patience to switch to another render engine that would just be better and faster for my works.
This is when there wasnt so many choices out there, you can see by yourself the current status of this engine, considering AD implementation will always be ready to fuckup everything they could.

AD + Mental Images = awesomeness.
Feel free to enjoy MR and trust people comments about its reliability, speed, etc, i cant actually wait untill you dig into it and you will realize how a joke this engine has become.
It has litterally NOTHING that another engine can do faster and better, it is slowly developed, poorly updated and implemented, poorly documented, there are few tutorials.
so again please someone enlight me on a single reason why you would want to use MR over any other render engine, what feature it has thats better or other dont?

I'm gonna use this quote from Remydrh to enlight some aspects
"For product shots, environments, exteriors, and set recreation with HDRs, I would suggest native (builtin) Ibl, these are strings options. Or second choice, user_ibl. Secondary GI can be Final gather (FG) but greatly reduce the needed accuracy as the IBL is handling most of the light in these cases. (Each situation is different so play around some.) And use the fgshooter mip shader if possible to reduce FG variance in animations. It makes the solution more automatic to render with less tweaking while keeping the speed improvements of an interpolated technique."

i've highlighted some stuff:
- native builtin IBL : we dont have this in Softimage, it works trough ctrl_ghost but it doesnt update correctly so you just cant use it, beside this as remydrh says it requires bounces so its not a global illumination solution, you still need Final gather (this applies to userIBL too as i described above) which brings you back to the splotches and flickering problem.
- play around some : as i said this is what you will end up doing most of the times with mental ray, tweak tweak tweak untill you get frustrated as hell, so yeah play around some
- fgshooter mip shader : i strongly encourage you to open softimage, hit F1 start reading FGShooter usage and use it. Call me back when you managed to do it Good luck.

I dont care if its AD fault or Mental images fault to implement/bugfix/update/develop stuff, the bottom line is the end user will always get fucked. So dont start with the "its AD implementation that fails" WE, users DONT care who fault is. we want working stuff.

Regarding that scene from Alex Alvarez. This is the usual misleading information I see every damn time MR comes out.
Yes Amazing scene, now lets analyze it a bit.
The guy says it took 30 minutes (but he forgot to say what CPU he had apparently). Beside that most people come to forum "hey! Look i did this render in 15 minutes at 1980x1080 with mr!! blazing fast!!" lets see, how much time did you spend to setup the scene? ah! This is the point.

if you spend 8 hours to setup a scene because your damn engine workflow is crap old and full of optimization needs/problems/ and then you spend 15 minutes in the actual render the total amount of time you threw at it is 8 hours and 15 minutes. Not 15 minutes rendering.
The workflow of a render engine is what makes you productive, not the actual render time. I wonder how many preview render that Alvarez did to achieve what he wanted? if i am not wrong that work took more than 2 months, without counting the amount of tries with instances/lighting.

nothing is as it seems
Just to explain this further, in maxwell for example your scene setup lighting and shading takes less than 30 minutes for example then you have a lot of time to render. But you dont bang your head on a wall tryin to figure out what the heck is going wrong with your render, throwing milion of previews to adjust displacement parameter that instead of being 0.001 needs 0.00008, see what i mean?

I've seen stunning works from any render engine, its not the renderer that makes a beauty work, but the workflow and the modern tools sure helps a lot, how much would help to have an interactive preview when you lighting a scene? Dont tell me its useless. Oh yes we have the progressive rendering!!! LoL. No we dont.
Mental Ray has nothing modern, its stagnant, old, still people can use it and produce nice images BUT it doesnt mean is actually worth to use considering how modern other render engines are too. Again after you see the actual situation (go check out mental images forum, sounds like a desert with always the same 4 people speaking) give me a single reason why you would want to use MR over anything else, i'm really willing to know this. Find me something MR has that vray/arnold/modo doesnt have, workflow included.

I can only be sorry for you if you decide to start spending time with it Good luck and i really mean it.

/end of rant (i know i sounded harsh but i'm pretty sure you will agree with me with time, if you decide to waste it on Mr).
oh btw sorry if i offended someone (which i doubt).

Enjoy

[remydrh]

In many scenes Unified can be very very slow compared to old method and here you have people say "yes depend on scene, depends if you know how to use it, depends how you set it up, depends on your materials, depends.. depends...."

This is true of all renderers. Performance is based on the architecture and choices made by the user. Please see the Vray docs where they describe the Adaptive Subdivision sampler occasionally performing better than DMC:

(Vray Docs) On average it takes fewer samples (and thus less time) to achieve the same image quality as the other image samplers. However, with detailed textures and/or blurry effects, it can be slower and produce worse results than the other two methods.

So I have said nothing that isn't true regardless of implementation. Hence my comments about "sphere on a plane" indeed being slower with Unified. This matches exerience and documentation in other renderers. Unified and DMC are both designed for complex scenes where control is best left to the core.

Yes you can do flicker free animation with Mental Ray but for some reason always on other people scenes, not in yours. The amount of tricks/optimization/time wasted to make it work is simply ridicolous.

Please visit www.themill.com for quite a few examples of flicker free animation that is consistent. The fact you have never been able to achieve it yourself may be a training or time limitation of some sort. But I would hardly say your experience is universally applicable and is as demonstrated above, a verifiably false assumption.

You cannot export portal lights/sky/environment and portals with correct lighting because portal lights require photographic if not they blown everything out, so you have to think about this in compositing (another damn step to fix a complete mess that was poorly designed and more poorly implemented) The sky doesnt get proper gamma so you need to replace it, its retarded, a nightmare and we are in damn 2012 not in 1990 anymore.

My solution to this would be that you should indeed think about two things: 1. linear workflow and color selection choices in the software that avoid tonemapping in 3D. 3D packages are not especially designed to handle this well since they invariably bake-in the solution. Hence the reason they are 3D packages. 2. You should rely more often on Nuke, Photoshop, etc for this. As shown by your Vray passes argument, you do rely on passes at times. To make it even more simple, try using correct additive passes only: diffuse, indirect (and indirect with ao), scattering, emission, reflection, and refraction. In post these are simply plus or add operations. This shrinks your options down to 6 passes. At Sony, the passes output for many shots of Hotel Transylvania were restricted to the passes I just mentioned.

SSS was designed to work without GI, because in the old days people were used to light stuff without GI. Now its the complete opposite, thats why Arnold is kicking everyone asses.
Thats being modern. yeah Mental Ray can do it!!! at what cost?

Very quickly typically. And the control set is much easier in SSS2. I'm not sure where the comment about SSS not designed to work with GI is from. It's consistently used with indirect illumination and calls irradiance in the shader.

Lightmap? come on the whole system was fine in 1990. Not anymore.

Vray also uses a cached mechanism for SSS. In Vray it is referred to as the "pre-pass rate" in the shader and is also resolution dependent. In mental ray 3.11 the lightmap has be removed and is now rendered on the fly without any caching or point clouds.

always laugh at this when speakin about mental ray cause it has nothing designed to be physical

The physical light, or any light with quadratic decay can be considered "physical". The mia_material, final gathering, irradiance particles, builtin BSDF, and user_ibl are all modern shaders with physical properties by default. You should avoid legacy shaders like phong, blinn, etc. Usually these are kept exposed by Autodesk for fear that legacy customers will have a stroke if hidden or removed.

Oh Wait! Physical light should used with point light, not spot light. Too bad no-damn-one tell you this untill you end up in forums.

From a mental ray perspective, the physical light shader can be used with any non-ibl light type. This is perhaps a documentation problem with workflow in XSI. Sadly Autodesk has not put a lot of effort into documenting integrated products. Instead they often opt to include the integration manual which is not the same thing. It's up to the OEM to do this because the mental ray user manual is different in XSI than 3ds Max for example. It's also possible the information on forums is incorrect. I often see the "solution" to a problem is to simply crank everything up. . .

They forgot to add bounces, so both IBL/usreIBL are without GI, meaning have fun to light a scene that requires bounces, meaning its just another fake.

I feel this is a misunderstanding of both Vray and mental ray. Neither dome lights generate indirect illumination (or "GI") they are simply smart area lights, or direct lighting only. To generate additional bounces you would still need an Irradiance Map or Final Gathering. Vray documentation shows it allows you to store the light in the irradiance map, but you still have to have it "on". In many cases for exterior or set recreation you may not need to add GI for a satisfactory image because the dome effect looks pretty good alone. This is obvious to see from the days before useful GI when lights were placed in a dome pattern to "fake" GI. Both Vray and mental ray (and even Arnold without a diffuse bounce or "GI") operate this way.

It has litterally NOTHING that another engine can do faster and better, it is slowly developed, poorly updated and implemented, poorly documented, there are few tutorials.

I completely disagree with a few of those points as I talk to many of the developers directly. You do understand quite a few technologies, including rQMC (used in DMC and Arnold at Sony) are in fact mental images patents. However, it's true the documentation and tutorials are very much lacking. Autodesk has not provided enough information on these features. This is one of the reasons the Elementalray blog exists, to fill that gap. I also find (sadly) that many schools and universities are teaching old methods and technology. We find interns that have freshly graduated are using very poor practices they learned in school. (Put down the Blinn and walk away slowly. . )

Final gather (this applies to userIBL too as i described above) which brings you back to the splotches and flickering problem.

I'm curious if you've tried this combination in earnest or are simply rejecting it by fiat. I suspect your frustration is fueled in part by the poor support XSI seems to have received lately. And if you have in fact found a better alternative then I wonder why you are in this thread. Or for that matter, why anyone devotes the time to troll on forums. As for flickering, any and all interpolated techniques will flicker without user interaction. This is a mathematical nature of interpolation. The alternative is brute force. Which is certainly easier (as evidenced by Arnold) but much more expensive in rendertime (as evidenced by Arnold again).

play around some : as i said this is what you will end up doing most of the times with mental ray, tweak tweak tweak untill you get frustrated as hell, so yeah play around some

The idea that a single button click solution exists is what permeates a lot of frustrated artists. Unless you use something like iray, there's a good chance you're having to tweak and push and pull. Having used mental ray, vray, and the current project on Arnold, I can say they all have this "problem" of requiring you actually, dare I say it, work some to get the best possible result. For mental ray the challenge can be navigating the integration. Vray it can be eliminating noise. And Arnold the challenge can be slow renders or unsupported features.

I think learning mental ray can be valuable for a few different reasons:

The flexibility many seem to loathe actually prepares you to use multiple techniques from rasterization to raytracing used by other renderers. Just as user_ibl and vray dome lights, etc are very similar, you can learn the processes without spending the money on another product (if people do indeed buy software nowadays. . .) This means you can use Vray, Renderman, or even Arnold type rendering without spending a dime because the concepts overlap.

Just to explain this further, in maxwell for example your scene setup lighting and shading takes less than 30 minutes for example then you have a lot of time to render. But you dont bang your head on a wall tryin to figure out what the heck is going wrong with your render, throwing milion of previews to adjust displacement parameter that instead of being 0.001 needs 0.00008, see what i mean?

And this render from Maxwell is perfect every time the first time? You have no client that asks for changes? Last minute changes? Expensive changes? How do you handle jobs where render time is at a premium where you may share a render farm or version 28 needs to be sent to compositing in 10 minutes? Are you using this for animation? In some cases where it's a single freelancer or usually an arch viz studio, these are not considerations and you may indeed find other software better suited. But I simply cannot hire someone and keep them when every frame they send takes greater than an hour a frame on the farm. Ideally we require 25 minutes or less a frame.

oh btw sorry if i offended someone (which i doubt).

No offense taken, but I do hope after typing all of that you feel better or maybe had a Xanax handy. I really don't think software is worth having a stroke over.

Moderator edit (and this is not a warning or anything, just a (sort of) friendly observation):
I understand this to be a heated debate, the "Xanax" remark was, however totally uncalled for IMHO!!! - HB

[Soth3dlife]

ok, this is becoming a little fanatic.

This is the thing. Mental ray has a lot of old legacy stuff and the Nvidia people take long time to upgrade stuff. But it works, it's efficient regardless of what some people say. I personally I starting to use Vray, but just because I work a lot on compositing and with Maya where Mental ray sucks so bad and the passes are even worse that Mental Ray's integration.

On the other hand what Maximus says it's true, until 2013 the IBL system was crappy in Softimage but now it's fixed.

oh, by the way:

I have no idea why someone would bother learning Mental Ray nowadays. It has everything (no it really doesnt) but it has it old, not updated, poorly documented, not implemented, with a 10 year old workflow that really goes against softimage workflow itself.

There are several reason why some people, like me, still use Mental Ray.

[Nizar]

OH yes i saw some friend of friend of another friend of someone who did an amazing animation flicker free. Yes you can do flicker free animation with Mental Ray but for some reason always on other people scenes, not in yours.

Thanks very much for your long post maximus.
Your sentiment is the some I have about MR. I tried it sometime ago (2010 release) when I decided to learn softimage, supported by i3d mental ray volume 1, I found it really complex (don't want said a mess, I have not the experience for said so).

As said by friend of mine "softiamge workflow is a dream, but mental ray is a nightmare"

I can only justify it with the legacy aspect mental ray has. I thought the direction taken sometime ago was wrong (there were no terms of comparison) and now mental image must respect the legacy technology and all the artist used to this workflow.

For some strange reason, I start using and studied modo rendering for my works (I say "strange" because I don't like this software), but I can ensure a month is enough for learn and mastering it. I did some simple animation (turn around and some walk cycle), no flickering at all and setting is insanely fast if compared to mental ray (or what I actually read and know about mental ray), needs only to set correctly antialiasing sampling and other two parameters (and is very simple because modo give you prefab setting with clear indications). Setting correctly a scene in modo is very simple, and believe me or not shader tree (the ultra criticized layer system inside modo) is very fast to setting. So, for me, flickering is a totally stranger.

But you don't bang your head on a wall tryin to figure out what the heck is going wrong with your render, throwing milion of previews to adjust displacement parameter that instead of being 0.001 needs 0.00008, see what i mean?

at 0.0001 render time become insane...

I agree with you, so far, my feeling with mental ray is about the some. Every time I return back and want try it but I feel frstrated by the many (too many) parameters.
The reason I want learn (a day) mental ray is I have purchased it, and I like work inside softimage.

About Alex Alvarex scene, he said his machine has 12 gb in total so I guess is not a so great machine. Also, he wrote the plant collection took about 2 months and execution time 1.5 days.

[Hirazi Blue]

ok, this is becoming a little fanatic.

As a Moderator I fully agree. I understand Mental Ray is one of those topics (like "is the earth flat?") that evokes this in people, but please let's all keep it cool...

There are several reason why some people, like me, still use Mental Ray.

QFA

;)

[Kerro Perro]

I actually also had a long winded post about the things i love about mental ray and the things i hate, but you know what? Who cares?

The only thing i want to do is answer Nizar:

- i you can light and render your scenes with full gi in Modo or Cycles and it works for you: great! I would say keep doing that and use mental ray for quick tasks like rendering masks or albedo or AO - anyhing but gi or fg! If i understand your needs correctly this should be fine, use what works best, easiest and fast for you and concentrate on what you create, not how. I wasted a lot of time doing that and regret it.

- if in the future you can use Arnold go for it, it rocks!

Message ends.

[Hirazi Blue]

I actually also had a long winded post about the things i love about mental ray and the things i hate

I hope my previous moderating post wasn't the one to convince you to hold back. There is a fine line between heated debate and something ugly. I interjected, not because I thought this line had already been crossed, but because I wanted to prevent it from doing so.
"Long winded posts" are more than welcome, as long as everybody keeps it cool...
(me included, I hasten to add )

Who cares?

The OP for one and probably all the other users who have posted here so far... ;)

[Maximus]

This is true of all renderers. Performance is based on the architecture and choices made by the user. Please see the Vray docs where they describe the Adaptive Subdivision sampler occasionally performing better than DMC:

(Vray Docs) On average it takes fewer samples (and thus less time) to achieve the same image quality as the other image samplers. However, with detailed textures and/or blurry effects, it can be slower and produce worse results than the other two methods.

So I have said nothing that isn't true regardless of implementation. Hence my comments about "sphere on a plane" indeed being slower with Unified. This matches exerience and documentation in other renderers. Unified and DMC are both designed for complex scenes where control is best left to the core.

Except that for other render engines the documentation is quite explanatory, making users to quickly grasp the concept and achieve the desired result quickly. I am not saying Unified is bad, i'm saying that you need a lot of preview render to understand when its worth using and you throw away a lot of time doing this, especially cause we dont have any interactive preview.

Please visit www.themill.com for quite a few examples of flicker free animation that is consistent. The fact you have never been able to achieve it yourself may be a training or time limitation of some sort. But I would hardly say your experience is universally applicable and is as demonstrated above, a verifiably false assumption.

Yeah i was talking about them, they are friends of my friends of another friend i had. I didnt say its impossible to do, doesnt mean that its an enjoyable thing neither fast.
Mr doesnt have a brute force solution that works out of the box like Brute force in vray and Arnold. Yeah final gather in exact mode? cmon be realistic thats gonna take ages compared to any other BForce solution in the market. Irradiance particles? Please.

My solution to this would be that you should indeed think about two things: 1. linear workflow and color selection choices in the software that avoid tonemapping in 3D. 3D packages are not especially designed to handle this well since they invariably bake-in the solution. Hence the reason they are 3D packages. 2. You should rely more often on Nuke, Photoshop, etc for this. As shown by your Vray passes argument, you do rely on passes at times. To make it even more simple, try using correct additive passes only: diffuse, indirect (and indirect with ao), scattering, emission, reflection, and refraction. In post these are simply plus or add operations. This shrinks your options down to 6 passes. At Sony, the passes output for many shots of Hotel Transylvania were restricted to the passes I just mentioned.

Exactly what i was pointing out, you think that all this crap is something that should keep going nowadays? Did you see what you wrote? IN another render engine you just put a single light with all paramters, no problems linear workflow, no problem passes, no problem export. Now with MR you have to go trough all that thinking and all you posted, and you call this Flexibility? I call this counter-productive, old workflow. waste of time. Why should i rely on a post process package when i'm using a render? Give the user the option maybe, but never force someone to do so, thats not Flexibility, its stubborness by developers.

Very quickly typically. And the control set is much easier in SSS2. I'm not sure where the comment about SSS not designed to work with GI is from. It's consistently used with indirect illumination and calls irradiance in the shader.

The comment come from me, since everytime i used SSS with GI the render times were ridicolously high. Again the fact that it can be used with GI doesnt mean its efficient neither smart to do so, considering the render times compared to even using brute force sss. And the fact that they removed lightmap proves me right. About time.

Vray also uses a cached mechanism for SSS. In Vray it is referred to as the "pre-pass rate" in the shader and is also resolution dependent. In mental ray 3.11 the lightmap has be removed and is now rendered on the fly without any caching or point clouds.

Except Vray doesnt take ages to compute it you barely notice it, also I dont have Mental ray 3.11 so others, so I am still on lightmap. Thank you for the insight on next release. I cant wait to see how it performs.

The physical light, or any light with quadratic decay can be considered "physical". The mia_material, final gathering, irradiance particles, builtin BSDF, and user_ibl are all modern shaders with physical properties by default. You should avoid legacy shaders like phong, blinn, etc. Usually these are kept exposed by Autodesk for fear that legacy customers will have a stroke if hidden or removed.

They are all modern? Final gathering is something you call modern? irradiance particles are modern? Wow, so i wonder Modo render/Arnold/vray/maxwell are all old. Ok!

From a mental ray perspective, the physical light shader can be used with any non-ibl light type. This is perhaps a documentation problem with workflow in XSI. Sadly Autodesk has not put a lot of effort into documenting integrated products. Instead they often opt to include the integration manual which is not the same thing. It's up to the OEM to do this because the mental ray user manual is different in XSI than 3ds Max for example. It's also possible the information on forums is incorrect. I often see the "solution" to a problem is to simply crank everything up. . .

And where is the official mental ray documentation that states to use it with point light? Because if you use it with spot light it produces a lot of weird errors, one in particuar where you have to flip the light emission by 180°.

I feel this is a misunderstanding of both Vray and mental ray. Neither dome lights generate indirect illumination (or "GI") they are simply smart area lights, or direct lighting only. To generate additional bounces you would still need an Irradiance Map or Final Gathering. Vray documentation shows it allows you to store the light in the irradiance map, but you still have to have it "on". In many cases for exterior or set recreation you may not need to add GI for a satisfactory image because the dome effect looks pretty good alone. This is obvious to see from the days before useful GI when lights were placed in a dome pattern to "fake" GI. Both Vray and mental ray (and even Arnold without a diffuse bounce or "GI") operate this way.

Dont worry its not a misunderstanding, MR doesnt have brute force (not counting FG in bf mode since you better just let maxwell go considering the time it takes to clean).

I completely disagree with a few of those points as I talk to many of the developers directly. You do understand quite a few technologies, including rQMC (used in DMC and Arnold at Sony) are in fact mental images patents. However, it's true the documentation and tutorials are very much lacking. Autodesk has not provided enough information on these features. This is one of the reasons the Elementalray blog exists, to fill that gap. I also find (sadly) that many schools and universities are teaching old methods and technology. We find interns that have freshly graduated are using very poor practices they learned in school. (Put down the Blinn and walk away slowly. . )

Still you didnt list a single thing that MR does better than other render engines.

I'm curious if you've tried this combination in earnest or are simply rejecting it by fiat. I suspect your frustration is fueled in part by the poor support XSI seems to have received lately. And if you have in fact found a better alternative then I wonder why you are in this thread. Or for that matter, why anyone devotes the time to troll on forums. As for flickering, any and all interpolated techniques will flicker without user interaction. This is a mathematical nature of interpolation. The alternative is brute force. Which is certainly easier (as evidenced by Arnold) but much more expensive in rendertime (as evidenced by Arnold again).

I am in this thread because i'm in this forum since years now, i'm a softimage user, and i'm a mental ray user. Why are you on this thread? I could say the same to you why are you trolling?
So based on your logic you define someone trolling when they write something against mental ray? I think you should revisit your priorities, maybe get a walk or something.
You are giving me an explanation of why the flicker exist which i already know it depends on interpolation, what MR is uncapable of doing is having a decent GI solution to avoid that, like Modo render engine, Arnold, Vray, FinalRender, Maxwell, and even Cycle. Now the fact that people still manage do it once again DOESNT mean that is smart to use it.

The idea that a single button click solution exists is what permeates a lot of frustrated artists. Unless you use something like iray, there's a good chance you're having to tweak and push and pull. Having used mental ray, vray, and the current project on Arnold, I can say they all have this "problem" of requiring you actually, dare I say it, work some to get the best possible result. For mental ray the challenge can be navigating the integration. Vray it can be eliminating noise. And Arnold the challenge can be slow renders or unsupported features.

And this is the stubborness of developers since years, exactly that, and the fact that a lot of people (keep going) dropped mental ray for Arnold is just another proof of what i'm saying. People nowadays are tired to litterally waste hours on tweaking things. No one asked for a single button click soution, but no one asked for 8 hours setup scene either. Maybe a damn compromise? I used all render engines for quite a while and in none of them i had to tweak stuff so much like in Mental Ray.
But this isnt me vs you, or Mental ray vs other render engine. Anyone who used it knows exactly what i'm talking about, if you want to keep your head under the sand i couldnt care less, go ahead.

I think learning mental ray can be valuable for a few different reasons:

The flexibility many seem to loathe actually prepares you to use multiple techniques from rasterization to raytracing used by other renderers. Just as user_ibl and vray dome lights, etc are very similar, you can learn the processes without spending the money on another product (if people do indeed buy software nowadays. . .) This means you can use Vray, Renderman, or even Arnold type rendering without spending a dime because the concepts overlap.

This sounds like Autodesk bla bla bla marketing. Flexibility? I call that bad workflow, bad optimization (actually unexistant), old amount of shaders stacked on top of each other letting users confused on what and why to use that. Flexibility for you is throwing a gazilions of nodes without a proper documentation on why and when to use them and let the user figure it out? Hey you have everything now use it! This isnt flexibility, its called mess. user_Ibl and dome lights are totally 2 different world, no idea why to you they are very similar.

And this render from Maxwell is perfect every time the first time? You have no client that asks for changes? Last minute changes? Expensive changes? How do you handle jobs where render time is at a premium where you may share a render farm or version 28 needs to be sent to compositing in 10 minutes? Are you using this for animation? In some cases where it's a single freelancer or usually an arch viz studio, these are not considerations and you may indeed find other software better suited. But I simply cannot hire someone and keep them when every frame they send takes greater than an hour a frame on the farm. Ideally we require 25 minutes or less a frame.

Its not perfect every time, but maxwell has a very intuitive/fast way of working, You have material preview instant at almost max quality, you have FIRE which is an instant lighting preview of your scene and you know exactly that at some parameters render will be perfect. Dont compare this workflow against hours of tweaking without an interactive preview because honestly its dumb. Vray has also RT preview, Arnold have RT preview, Cycle have interactive preview, even my grandmother has interactive preview. MR has what? Progressive rendering? OK!

No offense taken, but I do hope after typing all of that you feel better or maybe had a Xanax handy. I really don't think software is worth having a stroke over.

I could say the same about you, i hope after typing all of that you feel better or maybe had a Xanax handy. Typical mental ray fanboy attitude.
I dont write in this manner usually but since you are a flaming person i made an exception just for you, consider it a christmas gift, we are almost there.

Moderator edit (and this is not a warning or anything, just a (sort of) friendly observation):
I understand this to be a heated debate, the "Xanax" remark was, however totally uncalled for IMHO!!! - HB[/quote]

This is a forum (i tought forum is the place where you write and share ideas or am I wrong) and i'm writing my experiences with Mental Ray, being considered fanatic or troll is just stupid, disrespectful and childish.
For the same logic i could define anyone who write anything, fanatic and troll.
Its not a heated debate, i couldnt care less if he uses mental or another engine, i dont work for Mental/chaos/autodesk, i just exposed things that are bad to me based on my experience.
I pointed out my experience and things that dont work with mental ray to an user that asked to. Now if fanboy jumps here calling me fanatic i couldnt care less but its quite stupid frankly.

@Soth3dlife, "On the other hand what Maximus says it's true, until 2013 the IBL system was crappy in Softimage but now it's fixed." Is it? doesnt look to be fixed to me, not remotely.
Native IBL doesnt work in Softimage, you need to use ctrl_ghost to activate it and the preview/render to disk doesnt update everytime you make a change, you have to refresh the render twice in preview/render region mode, thing that is impossible to do when you hit "Render >Render current frame" to disk, because you cant refresh and it renders all black.

Beside the fact that as i said IBL itself is suited for outdoor scenes and nothing that involves bounces since you have to activate final gathering.

"There are several reason why some people, like me, still use Mental Ray." I said the reason why i use it, because I dont have time and patience to learn another engine and spend money on it. But honestly i never saw a single person switchin to antoher engine and coming back to mental ray. If mental ray wasnt bundled in autodesk packages I bet my own balls no one would even think to buy and use it, and Mental images would just bankrupt. Strip it out of autodesk packages and see how many people would go for it

I always find interesting those threads, i have no idea why when someone confront ideas people always jump like "flaming, fanboyism, fanatic". Its a damn forum, no one is offending anyone.
Keep experiences going. I'm interested.

Anyway Nizar i'm more than willing to help you with anything you need, feel free to ask

[Kerro Perro]

I actually also had a long winded post about the things i love about mental ray and the things i hate

I hope my previous moderating post wasn't the one to convince you to hold back. There is a fine line between heated debate and something ugly. I interjected, not because I thought this line had already been crossed, but because I wanted to prevent it from doing so.
"Long winded posts" are more than welcome, as long as everybody keeps it cool...
(me included, I hasten to add )

Who cares?

The OP for one and probably all the other users who have posted here so far... ;)

Nah your post had nothing to do with it, i was still (re)typing mine when you posted it.

And i pretty much stand by what i said, but just in case anyone truly cares:
what i love about mental is all the npr capabilities (if you know me that's no surprise, i go on about that stuff all the time ) and yeah the thing a lot of people seem to hate: the way you can edit and fiddle with nearly every aspect of it.

What i hate about it was pretty much well covered by Max ;) Also yeah Nizar displacement is pretty much a nightmare, but i found that annoying in vray as well..

But i still think my previous answer to Nizar is what matters - find a method that works best for you and refine it! Don't get bogged down in "ooh that other renderer has feature x and mine hasn't!" Unless it's Arnold of course

[Maximus]

Here are other opinions Nizar, go ahead happy reading, and check out the comments at the bottom too:

http://www.chadrikvon.com/mental-rays-s ... ce-in-vfx/

another page where you can get information, mostly the problems you will encounter sooner or later:

http://forum.mentalimages.com/showthrea ... sion/page2

in particular that is one of the "WTF" moments you have in your render tryin to figure out why you have some random noise around in your scene and you have to crank up parameters to the roof. That is a nice reading if you have time.

[Rez007]

Hi Nizar,

Since you are the OP of the thread, I am just going to direct this towards you, I would either take the time to learn Mental Ray, or don't and move on. So far it seems that you like when people give you advance and where to find better information on how to learn Mental Ray, but to me, it feels like you keep holding on/agreeing to the statements of Mental Ray being "bad", basically to reaffirm your own feelings about it. I would like to think you started this thread in hopes to acquire legitimate information on how to produce better results from Metal Ray, because regardless everyone has their own opinion on how to make things work. I am not a "friend of a friend who had a friend" who got flicker-free animations accomplished, I did it myself, and for the scene I was doing - it worked for me and I shared my method. I suggest you read over Elemental Ray if you want to learn Mental Ray better, if not then just stay with the renderer you like. I for one, will use what I pay for when I find no need to move to away from a renderer that has produced many awards, just to go to another one that has probably some other limitations as well.

Obviously, everyone is entitled to their opinion, but for those rather than venting solely on a forum, have some of you actually voiced your opinion directly to Autodesk? Because I have, and I did mention to them some areas I would like to see addressed in regards to Mental Ray. Whether they do it or not is another story, but hey, it is better than just discussing it solely on a forum.

http://usa.autodesk.com/adsk/servlet/in ... id=1109794 (Select Softimage from the list which it finally got added...)

http://area.autodesk.com/blogs/cory/re- ... ing-myself (email to Cory Mogk who is the new Product Manager for Softimage. He shares his email address near the end of the page in one of his posts. It does seem like he is trying hard to do well with Softimage, so it couldn't hurt)

And also, it might be a good idea for those of us who use Softimage and Mental Ray to turn on your CIP, so Autodesk can see that you indeed use it. Maybe then they might address users' concerns more with Mental Ray if they see more people actually using it...

http://area.autodesk.com/blogs/cory/ena ... -softimage

In addition, to whether you should use Adaptive Sampling or Unified Sampling, I personally would use Unified Sampling from here on out. Mental Images, I feel, is setting up for Unified Sampling for now on. I think it would be pointless to stick with the old method, kind of the same logic how some people stay with the legacy shaders - it is there but more efficient possiblities exist.