Render update, when?

[AdamT]

a) why can't you do it in Cinema as it now stands?

b) you know no one can (or will) disclose development in-progress, so why keep asking?

I'm all for improvements to AR, but me thinks you are just :banghead:

[STRAT]

Adam's correct.

we ALL want an AR update, which i'm sure Maxon are fully aware of, but asking in forums and posting masses of pointless posts is just wasting time and space.

As soon as there's any update available, or news of one, Maxon themselves will announce it. keep an eye on their web page :)

[lllab]

there are quite a bunch of users waiting for that...


me too, but i guess i can say for sure we will see it this year;-)

cheers
lllab
stefan

[ThirdEye]

it's ready when it's ready

[fritzman]

...i guess i can say for sure ...



Hi Stefan,

that's cool. Can I use it or is it copyrighted? ;-)


All the best, FRitz

[rendermania]

we ALL want an AR update, which i'm sure Maxon are fully aware of, but asking in forums and posting masses of pointless posts is just wasting time and space.

As soon as there's any update available, or news of one, Maxon themselves will announce it. keep an eye on their web page :)

I'm a bit worried that that AR update will introduce nothing more than better GI sample smoothing or maybe some new memory management trick to keep print/archviz/broadcast graphics people sattisfied. The shader control/creation system really needs to beef up proportionally with other leading renderers like Mental Ray, because shaders can do more work towards achieving photorealism in my opinion than last refuge techniques like HDRI or GI lighting.

Its not all GI, GI that makes for great renders. If you know how to light properly and you have a powerful shader creation system at your fingertips, its possible to go a pretty long way towards rendered realism without incurring the speed penalty that GI lighting brings with it.

And yes, I do check the Maxon website almost daily to see if there is some update but sometimes I wish I could just buy a plugin renderer with the desired capabilities and slot it right in.

[LucentDreams]

cinemas shader system? Thats what I'd consider its strong point. Granted its not a shader tree, that is the shaders current serious flaw, but with our layers and the procedurals built in, we have one very slick (and easy to use which is one thing against shader trees) system.

[rendermania]

I think the difference between the current system and a proper shadertree + a few new illumination models (why do we only have 3?) will make a difference like night and day in complex shaderwork. Nothing against the material manager. Its nice for slapping together materials quickly, but its also inherently limiting in a lot of ways. A shadertree doesn't have to be slow. XSI's is a lot more cumbersome and click intensive than it needs be.

[ThirdEye]

a few new illumination models (why do we only have 3?)

actually you have 6. Lambert (just uncheck specular), blinn, phong, metal (in the specular tab), oren nayar and anysotrophic (lumas/danel), and it's the best anysotrophic material i've ever seen in 3D. Even XSI and Maya don't have these illumination models

[LucentDreams]

not to mention in the luminace channel you can put any illumination model you want. Lumas is a prime example, Vreel's new skin shader is another example, where to use it best, you deacitvate the color and specular channels because it has its own shading models including anistropic built inside.

but myself, if maya would let me scrap lambert form the file, I would in a second, what the heck do you want to use lambert for and why have to have two seperate shaders for lambert ot not? cinema's got a great system for haveing all settings in one shader. I mean have you used max much? Especially with some of the 3rd party renderers?

Having to choose a raytraced material as opposed to a normal one, how stupid is that. if I want raytrace why rebuild anything, I should simply beable to turn on raytraced parameters and they become raytraced. Its notlike raytracing even has a noticable renderhit nowadays, unless of course you use maya and max' default renderers instead of MR.

This is one of the concerns I've had about when maxon finally does get a shader tree, is losing the simplistic logic of their shaders. Having a shading model that limits what settings are available is bad imo. Having one wher you simply turn on and off options you want or not, is good.

[Other3DMaster]

This is one of the concerns I've had about when maxon finally does get a shader tree, is losing the simplistic logic of their shaders. Having a shading model that limits what settings are available is bad imo. Having one wher you simply turn on and off options you want or not, is good.

This is one of my great concerns as well. Maya drives me nuts in that regard.

[lllab]

rendermania,

to be honest i too think, cinemas shader system is one of ist stronger parts. and light it one of the most important things. if you have testes vray you get an idea what you are missing in cinema, there you need a lot less shader tweeking to get highly realistic and nice materials. its all a matter of light.

a good example is also maxwell, it does extremly nice surfaces and bumpmaps, although its shadersystem is more than limited. BUT it takes bumps, reflection etc absolutely realistic.
in this renderengine, material setup is absolutely easy done.

i just wished cinemas shader get some small, but important updates: an option for fast blurry effects. the ability to have anisotropic blur as it is in maxwell, brazil, vray etc, that means i have different settings for u and v coordinates(remotions paranormal reflection shader has this)

a distance to ray falloff! node/shader

a simple shader/ effect that blurs the content of the layer shader, again with u & v control( like you can do with bitmaps.)

for the rest i hope we get some kind of lightmappin a la vray, irradiance maps, a lot faster "photons"(look at brazil), both savable into a file, so that i can save diffrent solutions

distributed rendering from within the imagemanager.

better, more crisp AA, i know most think different here, but i am no big fan of cinemas AA, again brazil and vray do it nicer (crisper, cleaner, faster) cinemas AA is so blurry, its also more than 2 or 3 years old, there are much better algorythms, at least i would like to have some more options.

some color mapping built in(see color mapping plugin)

save renders as hdri as vray does (really nice!)

edit: forgot: improved snapping -visual feedback!

my2cent

stefan

[rendermania]

This is one of the concerns I've had about when maxon finally does get a shader tree, is losing the simplistic logic of their shaders. Having a shading model that limits what settings are available is bad imo. Having one wher you simply turn on and off options you want or not, is good.

I don't share that concern because a more advanced shader creation GUI can be designed in any number of arbitrary ways. It doesn't have to look like a tree or Cinema's current material manager.

Just to give an example I've pulled out of my hat, imagine that you have a number of presets for different material types distinct enough in look to make a category out of them. Say you have:

woods
plastics
glass
metals
skins/organics
etc

Now you make a color coded chart where each of those materials takes up a certain graphical space and there are more specifec presets represented by dots in those spaces, like oak, or PVC, or plexiglass, or tin or nylon or tarmac or cardboard. You could mix influence on the material you are trying to create by sticking graphical markers on that chart and determining how much influence each should have on your shader. You can imagine what you could do if you put one marker near 'dried leather', one near 'PVC plastic' and one near 'velvet' and give them 30-60-10 influence weighting. Things like that. Shader parameter mixing based on a graphical space representing real world material presets in relation to each other.

In other words, even though the final input that goes into the renderer is things like so and so much specularity, so and so much blurry reflection, so and so much diffuse falloff with blah illumniation model or x amount of GI emittance, you can make any number of arbitrary graphical user interfaces that let you mix and match your own stuff in different ways, and drive one parameter with a stack of others, etc

It will only get cumbersome a) if the UI doesn't give sufficient feedback, so you're staring at a spaghetti shader made out of 15 nodes and you don't know whats driving what, or b) if you make it as stupidly click intensive as some of the shadertrees out there, with their mishmash of drag-to-connect, right click for options, pick a node from a 40 item pull-down menu etc. You could probably do a slick tablet interface instead that uses things like mouse/pen gestures or simply lays things out for quicker access.

But what I was talking about wasn't so much the GUI as the shaded look AR delivers, particularly if you're not using GI. It should be possible to approximate the look you get from using GI with carefully placed lights, and I've personally found it impossible to get that look because AR's GI layer seems to be tacked on to the raytracer. Its almost like switching GI on even without the use of HDRI or image based lighting gives you more dynamic range than using straight lights and shaders.

I also have to second what Illab said about the AA quality. I'm no particularly impressed with it, particularly when it comes to rendering large landscapes where far off detail should be antialiased but still sharp enough to see. I frequently have problems getting the basic scale of a scene across with AR, particularly when the 3D models used don't give enough clue as to how large what you are looking at might be in real life. Maybe a system based more on real-world units would do a better job.

[Per-Anders]

i suggest you try doing these things in another render engine and see for yourself how "superior" they are.

everything in rendering is about what you put in to it. if you want something distant to look distant then you have to be observant and work out what it is in an image that makes something look far away. our aa is just fine for distant and close objects, provided you know how to use it. the only thing it apparently lacks is a stronger sharpening filter, which can be done in photoshop/ae beyond that its' down to how you use it yourself.

if you want something to look big you have to work out how to make that thing look big. this is the same as when you make any image in any media. it's not down to the render engine. there's no render engine that makes things look big, or one that makes things look small, there's just a bunch of little observations and tricks you can use to achieve the look.

as for the radiosity/gi comments. you will also find this is the same in every other application. lighting in c4d is probably one of the easiest lighting systems out there, i find it much easier to light in cinema than in maya or xsi or lightwave. gi is a complex thing and to think that you could mimic it exactly with lights only easily is a falsehood. it requires a lot of skill, those with that skill will of course make it look easy, but really it isn't. ar with gi is not some different raytracer, it's not adding anything to the rays except part of the previous sample, same as all radiosity/gi solutions. if you want to make something look like it's lit up with gi but using normal lights it's purely about skill, it's possible in c4d as it's possible in any app, but it is a hard skill.

with regards your "advanced shader creation system" concept. what is different about this from the layers shader? apart from the look? layers shader gives you the control in another visual form (sliders) with a tad more precision and different overlay modes.

a node based editing system really is something rather different, it's a visual programming language. it allows you to create new shaders from component parts and functions (nodes).

rather a lot of misconceptions come from poor quality node based texture editing systems, such as xsi's where you are forced to use nodes to do many simple tasks, when in actual fact cinema could very easily have a node based editing system without changing a single thing as far as the users are concerned currently (apart from a small button on the end of each interface element, rather like the dot for recording keyframes we have currently). that nodes need only be visible when really needed, and that's not the same thing as making it so that basic tasks need nodes.

i honestly haven't seen a great deal of artwork from cinema users that pushes AR to it's fullest yet. AR will no doubt in due course as and when Maxon decide to be pushed further and improved upon. but if you are having such great difficulties with it right now that you cannot produce the necesary result and you believe you have pushed that side of the C4D render engine to it's maximum potential, then the only solution is to try out a different render engine, and see if that is less limited and gives the result required, if that means learning another app, well honestly i think that's a good and healthy thing.

[boxy]

For what its worth I like the look of C4Ds current shader system. I don't own R9 yet but I know from the demo I'll feel right at home building shaders when I do. The only shader tree I've used to date drove me absolutely nuts! No presets, having to add the same nodes everytime you want to build the basics of a shader aargh! And then spending far too many hours trying to suss out what other nodes should be added and how they should be connected to get the look you're after - aaargh x 2 :) Well I can understand people want to create shaders with a tree, but hell I hope Cinema's current system isn't completely rejected if it does happen.
Cheers
Boxy

[raycerx]

well I am STILL trying to get my brain around Expresso [nodes and such], so the whole node material thing scares the crap out of me.... Cinemas materials are why I ditched max years ago. I think its simple and powerful. Can it be improved? sure, there's always room for that. But I really do like the simplicity and power of the material system as it stands..

[kuui]

i second mdme_sadie 100%

the material editing system with a single button to edit special settings nodebased or with a language like RSL would be way cool.

for AR and GI, photonmapping/lightmapping seems to be todays premiere solution for fast and believable GI.

and about competing with vray, yes vray has that "render photorealistic interior button", but material editing is not easy and for all other tasks, vray doesnt seem to do so much better than the most GI renderers out there (in animation, it even has flickering, which e.g. brazil does mot have)

[basilisk]

From the work I have done on film (not yet Hollywood degrees of realism) my criticisms of AR are in other areas.
1. Vector blur - great when it works, but no use at all with any shaders with alpha channnels (virtually every scene i do has alpha channels somewhere)
2. Post effects cancelling each other out - highlights can be quite useful when used with restraint - turn on depth of field and off they go.
3. Would be nice to be able to tweak post effects in real time (on a single frame) to get what you want, before rendering a whole animation.
4. Why can't you copy and paste render settings between setups, or duplicate a render setup, or save a render setup as a preset or something (perhaps I am missing something here)

These might turn a potentially powerful tool into a genuinely powerful one, and save me time faking things in After Effects.

[Fre]

rendermania,

save renders as hdri as vray does (really nice!)


I just had this little idea but don't know at all if it will work.
Hdri is the same photo taken with different exposure settings ...

then wouldn't it be possible to render your scene in multipass & adjust the brightness of your lightpass in photoshop (= ? different exposure settings) & save it as an hdri in photoshop ?
just an idea, I have no hdri background to back up this idea.

[kuui]

i dont think its possible as c4d puts out the usual 24 bit images as tiff, tga or psd in multipass as well as in singlepass.



hdri is defined by having higher dynamic ranges, and to get that feature you need more information. HDRI's do have 64bit or 96bit or whatever, thats the thing.

[DELTAadmin]

I also feel that Cinema's material and shader system is quite good.

Sometimes I feel the need of plugging an already existing channel setting into different channels, here I need to copy and paste channels, but it is not tragical. Sometimes I need to drive noise processing, or pattern processing which I am not able to performe here in Cinema, in these cases I can use darktree.

If I had some suggestions:

Node based shader making is not necesarrily confusing if u can create interfaces to the shaders. This is the case with darktree, Renderman slim, and also with Houdini. In houdini u have a visual interface to create various shaders, u can create custom interfaces to it, and u can save these shader networks as normal Cinama like shaders.

That is>

U create a tree based shader > Build an interface to it > Save this shader network as a usual shader.

I would like to see this same method in Cinema. Houdini also has built in predefined normal shaders (amongst which there are some very brutal (Vex layered shader, vex supermaterial...), so u do not need to resort every time to manage nodes.

[Fre]

i dont think its possible as c4d puts out the usual 24 bit images as tiff, tga or psd in multipass as well as in singlepass.



hdri is defined by having higher dynamic ranges, and to get that feature you need more information. HDRI's do have 64bit or 96bit or whatever, thats the thing.

I read somewhere that most renders render images in 48 bit & then save the images as a 24 bit (or 48 bit if you choose 16 bit/channel)

How hdri's are made : at least three times the same picture but with a different exposure setting. That picture is a normal picture (8 bit / channel).
If you merge the three pictures to an hdri you get a (32 bit / channel) hdri.

I've seen pros take hdri's & they usually just take one RAW image, change the exposure of the Raw & save different versions of it in a normal image format (8bit/channel). Then they merge those images to an hdri.

Making hdri's is possible with Photoshop CS (full HDRI support) I use the bridge CS2 now to browse thru my hdri images.

Now the only question is : Is adjusting the lightpass in a multipass render the same as changing the exposure of an image ?
If it is it would be possible to make hdri files of your renders.

I could be wrong about the whole thing, as I said, haven't read that much about creating hdri's yet.

[CGTalk Moderation]

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.