Thanks for your thoughts and welcome to our forum! I think you are right. We have noticed that C4D users love own C4D workflow a lot.
Our engine has already quite good speed and can improve in this area. And we currently need to add other graphics features like hair&fur, motion blur and particles, we want a lot of them. And materials...
I can say that we have almost no performance related settings. No #samples for different scene regions and etc. There are only "max iteration count" for render stop criterion. We will also add time and noise based render stop criterions. There are also GI clamp params to take noise under some limits. That's all. All other settings are for creativity. After 0.45 version which I hope to publish today (maybe after midnight) there should probably no need for current IPR priority param too, as the GUI lags are probably in the past (but we need more user tests to confirm).
However we have a lot of settings in material. It is very fat Uber Material. And this is not over. There will be a lot more shaders/texture types with ultimate flexibility.
Current drawbacks is that we need to support native C4D things, especially TextureTag. The most tricky thing now is to support layering materials on top of each other using these tags. However selection tags are working.
As for shaders/textures we currently have own selection menu and currently support only own shaders. See the screenshots. The reason is that C4D textures/layers/channels were not yet implemented. Also the converter from C4D native material is too limited for now.
We haven't used native Cinema texture slot because I hope it's not right to put user into confusion in the material menu telling him what is supported or not from the very long list of C4D textures menu wherein our supported shaders are hidden deeply in 2-nd level.
We have built own menu of shaders just to tell what is supported for sure. Our own shaders can be copy/pasted among different materials, aslo can drag&drop.
It is possible to build a node-based workflow on top of that. Many things are ready, just a suitable GUI choice is needed. It can be similar in some sense to the current Octane or Arnold node-based shader network.
But there is also a plan to support native C4D materials and shaders. This task can probably be done partially, not sure yet how far we can go, e.g. with glow. But we will try to do it. It depends on how we can extract their visual behavior into maths to code them inside GPU.
We have released CentiLeo 0.45 for 3ds Max 2014-2017.
Download link:
Shaderball scene:
Please, read this changeloog. There are serious changes!
Change Log: cntlmax 0.452 alpha (2016 Nov 30)
[core] Fixed numeric bug which caused render convergence slowdown for HDR lighting
with very bright region.
Change Log: cntlmax 0.451 alpha (2016 Nov 29)
[cntlmax] fixed potential crash problems in Centileo material, IPR and Uber light source script data.
Change Log: cntlmax 0.45 alpha (2016 Nov 27)
[core] Significantly better performance when there are many light sources in the scenes. You may notice better noise cleaning when there are more than 10 light sources. E.g. when there are 500 light sources in the scene then 0.45 may render 50x faster than 0.44.
[core] Special image AOV called “Blend light passes”. Recall that light passes can unite several light sources of the scene. When allocated in AOV rollout together with other light pass AOVs then you can change the brightness and color tint for each AOV in post without re-rendering. Currently these settings are only in render settings.
[core] Same post-process feature as above but with “Blend beauty passes”. When this AOV is enabled then you can mix beauty passes Environment, Reflections, Base, etc.
[core] The simple Visibility settings of light sources (with on/off state only) were changed to more flexible spinners with range 0..infinity and with additional color tint per each Visible material layer by individual light source.
[core] Shadow catcher as shader. Now you can attach the color of background to any color slot of material (for diffuse or say reflection) and it will work correctly. It also supports rendering to alpha channel of shadows from HDRI which are casted to the object with shadow catcher.
[core] Alpha channel with automatic black background rendering. To render beauty and light passes with alpha channel select corresponding checkbox in AOV rollout. When selected the alpha channel will be included and the background from non-background passes will be black (this avoids artifacts on the border of opaque object and background).
[core] Translucency slot for CentiLeo standard material (placed together with diffuse).
[core] IOR slot for each reflection layer of CentiLeo standard material.
[core] Fixed confusion when color doesn’t change in case of changing the weight of some material layer (of this is the only enabled layer of material).
[core] Advanced mixing mode for CentiLeo Multi-Material. Now each layer has the checkbox “is coat”. When it is selected then the current material coats on top of materials under it (it is old mode of CentiLeo 0.44). When this checkbox is not selected then the current material is evenly mixed with the underlying materials.
[core] Added multipliers for Direct Lighting, Direct Specular and Indirect Lighting components of the rendered image. They force re-rendering and affect all beauty and light pass AOVs.
[core] Fixed weird bump map and normal map bugs when they are scaled using cntlMath/Binary or cntlColorCorrection.
[IPR] Added 3 modes for the Region Render (play = start drawing render region, pause = fix current render region, stop = continue render the full image).
[IPR] Fixed bug with appearing strange black borders when Region Render is changed to full image render.
[cntlmax] Less lags in 3ds Max GUI when IPR is running (needs more tests).
Note: the amount of lags was reduced so well that we recommend to up the IPR priority to 1 (in IPR settings rollout). This may increase samples/sec productivity of the IPR rendering. [cntlmax] Changed Standard CentiLeo Material GUI. Now the const values of the properties are tuned by spinners on the left side of material. The shader slots are on the right. Between them there are sliders which blend between the corresponding const value and shader evaluated value. This solution simply brings more functionality into the same GUI area.
[cntlmax] cntlNoise shader can now be rendered and manipulated in the 3ds Max viewport.
[cntlmax] Added button “Save current IPR image” to render settings. When IPR is running and is producing many AOVs at one time then this button may help saving all these images to the specified location at once.
[cntlmax] Fixed few other small bugs.
Known issues to be solved soon :)
- No yet multi-GPU
- May not initialize if AMD GPU is in your computer together with NVIDIA GPU.
- Very limited scene converter tool from other renderers (only diffuse property and textures from Standard material and Vray material)
- UVW mapping change of parameters may not be accounted by IPR scene auto update mode. To be solved.
- Need to finish implementing integration with ForestPack/RailClone/MultiScatter
Will post few comments and demos about this new update when will publish the update for CentiLeo to Cinema 4D (will be out soon, hope today).
One image demonstrates a 3 minute render performance with 136 light sources distributed randomly among planes which may block the light. The planes have glossy-diffuse materials in a dark environment.
Another image is shadow catcher which is implemented as a simple shader which is usefull to be attached to color of diffuse layer and even reflection layers. So reflections are supported.
Will wrap up current developed things for 0.45 and publish it. Other features go to further versions.
It currently the same history with other lights too like Omni and Spot. Currently when you change SizeX/SizeY you change the area of the light source, but the intensity (with multiplier) is specified for the entire light area. So the same intensity is evenly distributed on entire light area (if area become very large then intensity of small light part may become less bright). So when you change the SizeX/SizeY the emssion doesn't change but the softness of shadows changes.
I remember that few other renderers have a checkbox to "normalize intensity over area" which basically tells that intensity is specified for unit area (not for the whole area). Should I do similar option too?
Default environment color - it is used to illiminate the scene only when Sky with cntlEnvMapTag (HDRI) is not specified. It works per each visibility set (available in this tag). There are 5 of them: visible for camera (it is a background), for reflections, base and transmission. It is possible to setup diffent HDRI for each visibility set. But if some visibilit set (eg. camera) has no enabled checkbox (eg, "visible to camera") then default env color is used.
This is true that to disable default env color you make it black. Can add checkbox enabled/disabled so that color is not needed to change.
For those curious in status of development of CentiLeo 0.45 we can report:
Done features:
1) "Light pass blender" and "Beauty pass blender" - these are 2 new AOVs wherein you can change the lighting based on scaling the light passes realtime without re-rendering. Similar to LightMixer of Corona and Multi-Light of Maxwell.
2) Visibility settings per light source now have not only checboxes (enabled/disabled light) but also the scale and color factors for each material layer. It simply gives more freedom.
3) Few improvements for GI solver. That our adaptive GI solver is now a little bit more clever attacking most noisy areas even earlier.
4) Each render iteration is around 10% faster.
5) Several bugfixes. Most recent is fixing scaling and combination issues inside the shaders for bump and normal maps.
6) Imrpoved the situation with lagging GUI when IPR is running.
7) Material GUI has now dedicated Reflection IOR (default 1.4)
8) Only CentiLeo for Max: changed material GUI adding Mix params between const and texture.
Yet to be finished:
1) Improve GI solver when there are a lot of light sources.
2) Want to automate this manual trick:
3) Shadow catcher.
4) Add that small few additions/fixes to plugins for 3ds Max and Cinema that were requested in respective 0.44 threads.
Working very hard to publish it soon. #3 and #4 are easy, but #1 and #2 may requre some effort.
Also want to say that adding support for Multi GPU and interactive ForestPack/RailClone/Multiscatter are postponed again and will go separate to this upcoming release because these features may change stability on the first release with them.
[cntlmax] Fixed crash of map preview (the reason was cntlRandom shader)
[cntlmax] Fixed some GUI lags when IPR is running (needs testing)
Download here:
See small manual trick which may help increase perfomrance 2x:
Initially in 0.44 CentiLeo grabs all the free memory of GPU. Instead a better way is first to allocate less memory for caches, then push Reset. And then change settings to high cache sizes and push Apply. After this manual mumbo-jumbo it may become 2x faster (see samples/sec). Will try to automate this process.
0.45 is still under preparation, it has a lot of changes
That's good! Just wanted to write you about it. Because since 0.43 we have moved to CUDA 8 and it requires the latest display drivers.
Since 0.43 CentiLeo became much faster.
Recentily we have found few interesting things.
For example
1) if in Hardware Settings you set rather small GPU cache sizes (below 1GB for geo and 500MB for tex)
2) then start IPR.
3) And after that while IPR is running you may increase the cache size to previously set maximum.
If make steps 1-3 the renderer may perform faster (sometimes much faster for small scenes) compared to default behavior where max cache sizes are enabled and with which you start IPR.
For next version we will try to make this process automated. It's quite intersting thing with the roots from out of CUDA driver.
Welcome to our board! What GPUs are there in the system? It worked with NVIDIA GPUs with at least 2GB memory.
Have you updated the NVIDIA display drivers to the latest one?
Have you any one AMD GPU in the system?
Can you start Cinema 4D with copied plugin?
Or CentiLeo just does'nt start inside loaded Cinema?
Tiago, I have checked: well by saying "directional" light I was meaning "Infinite" light (of Cinema). To tune it's parameters you need to add our cntlTagLight. Spot, Area, Omni, Infinite - they work.
In our cntlLightTag there are 2 parameters: "scale area" and "scale angle". The first one scales the area of light source surface (this is valid for Omni (becomes a sphere), Spot and Area light). The second parameter changes the directional angle of light (valid for Spot and Directional light). Both params have effect only in renderer output (not yet in viewport).
As for Parallel, Parallel Spot, Square Parallel Spot, IES - they don't work yet. IES is important, so need to make it.
And please don't forget to post your WIPs here so we can visually see where to change the engine to make better performance and integration.
Well, mesh light will come slightly later, however I like this feature a lot. But next few versions will contain multi-GPU, faster multi-lights, light pass blender (aka light mixer) and some improvements for displacement tesselator. We simply come very close to these features, they are partially done. I hope, this is not too long to wait for them.
And then there will be mesh lights, physical sun & sky. Mesh lights should be done via texturing, so that some part of polygon is emissive and another part is reflective.
Btw, currently the native Cinema Sun tag is translated and it determines the directional light based on the used time.
We have released CentiLeo 0.44 for 3ds Max 2014-2017.
Download link:
Shaderball scene:
Change Log: cntlmax 0.441 alpha (2016 Nov 16)
[cntlmax] Fixed crash of map preview (the reason was cntlRandom shader)
[cntlmax] Fixed some GUI lags when IPR is running (needs testing)
See here:
0.45 still under preparation, it has a lot of changes
This is a list of changes 0.44
[cntlmax] Fixed bug when old viewport wasn't locked in IPR when viewports are
switching (actually view angle was not locked).
[cntlmax] Fixed bug when environment map disappear from one of the slots of CentiLeo
Renderer settings. But now when open old scene the reset of render settings may be
required (not always): if something is strange please change selected renderer in main
settings to scanline and then back to CentiLeo again and then recreate Environment map
slots again.
[cntlmax] UV tiling of cntlTexture is now mapped same way as standard Bitmap which is
also supported
[cntlmax] Now cntlTexture supports all those image formats that are supported by 3ds
Max itself (FreeImage lib dependency is removed).
[cntlmax] 3rd party light sources are converted to CentiLeo uber-light and convert old
light instancing (this feature accelerates manual conversion process for light sources).
[cntlmax] cntlBlend has const values together with shader slots
[cntlmax] Resource collector for CentiLeo assets connected to the scene (currently
cntlTexture and cntlProxy)
[cntlmax] Checkbox “reflection as coat” for Reflections in cntlStdMat. If enabled then
reflection layer become more important especially at grazing angles when cntlFalloff is
attached
[cntlmax] cntlRandom shader (at input 10 shader slots that are selected randomly with
some random color variation)
[IPR] enable/disable render status info (mouse right-click)
Known issues to be solved soon
- No yet multi-GPU
- No yet supported camera projection shifts for vertical lines tuning.
- May not initialize if AMD GPU is in your computer together with NVIDIA GPU.
- Very limited scene converter tool from other renderers (only diffuse property and
textures from Standard material and Vray material)
- UVW mapping change of parameters may not be accounted by IPR scene auto update
mode. To be solved.
- Need to accelerate first-time scene load and full reload
- Unfortunatelly benchmark scene is not yet available.
As I already told in C4D branch we now work on some important improvements of the core:
1) improvements for scenes with many light sources,
2) multi-GPU support,
3) and I think that soon there will be also some post-processing multi-pass very nice features just in renderer.
Btw, Ali Kavyani suggested to add one more setting to Reflections (this is IOR) and we are going to implement it
We are looking for your tests and suggestions which will help us improve.
We have originally designed the system in such a way that it uses 3 buffers (see bottom line of status info in IPR window):
1) buffer for rays - it uses between 512 and 1024 MB which is computed automatically based on available GPU memory (this is needed for out-of-core geometry engine and powerfull pipeline processing. This thing has a lot of advantages (out-of-core geometry, more flexible and efficient mapping of graphics features to GPU, it has huge potential to be extended with very advanced future features). The dissadvantage is that it takes this up to 1GB of mem (but no more) which is quite influential on small GPUs with 2GB only. However the trend is that new generation low-end GPUs will have 4GBs (1050ti) and 2GB (for just 1050).
2) geo cache GPU - this can be tuned by user in HW settings. And in multi-GPU setup the first iteration of implementation will have one setting for all selected GPUs. And later we will probably make it a setting for each GPU if this will not complicate the usage. If the scene is large and doesn't fit to GPU memory then the more you allocate here the faster the renderer is (slightly). However, the influence is not very pathalogical here. It should not deaccelerate too much for very out-of-core scenes. Your scenes just have to fit in main RAM.
3) tex cache GPU - same history for textures. They are implemented even more efficient than geometry. I think that something less than 1GB is ok here almost always. There is also a cache size setting for RAM memory which is uses for textures. For instance you may have 32GB RAM and 60 GB of textures in total. All textures are stored on disc in unpacked format. A convenient thing is that you need just few GB of cache for textures in RAM and don't to pollute the whole RAM. But the disc is polluted with all that *.mml files. I believe, that discs are much larger and cheaper than RAM anyway
So, as you see we have designed the system focusing on big scenes with stress situation.
However, when people start testing it with basic scenes the questions emerge. For example, if there is only single GPU in the system then it can be not nice to grab all memory from it. But we take only allocatable memory and don't exceed the memory limits.
Anyway thanks for pointing to this thing. I think it is possible to make some changes in the future, so that geometry/texture caches will be auto-extended only on demand by the amount that is necessary for scene up to specified limits in HW settings. Right now it's all controlled by manual input.
Btw, this new feature of inter-material copy-paste for textures creates the foundation for node-based material editor. What we need is to map things to GUI. It is planned to be done later after core improvements.
