Vray vs. Shaderlight ?
-
After using Vray for awhile, I have a feeling that it's a bit too complicated for my use: too many options, too many choices. I saw another renderer which claims to be more intuitive and user-friendly - Shaderlight on this forum. The trial version has been tested. After some tuning, I can get some images pretty close to what I got from Vray. HOWEVER, it seems that Vray optimizes the use of multi-core processor. For similar results, I think Vray can finish them about 30% faster than Shaderlight...
Anyone know anything about it? Why is Vray faster? Or I have done something wrong there? Thanks!
-
Thanks for taking the time to test Shaderlight – as you rightly say, our focus has been to make the software easy to use whilst still enabling you to achieve great results. Hopefully you have seen the time savings in the setup of your scene with the interactive updates.
Shaderight’s rendering speed relies on two things – processor and RAM. The faster the processor you have the faster you computer can deal with the computations needed to render a scene. However to process the scene it needs to be loaded into RAM, if the full scene is unable to fit in to the RAM then the HDD will start swapping memory and this does slow things down. Shaderlight is a 64-bit application so can use more than 3Gb RAM in a 64bit OS, it is also mutli-threaded so will benefit from faster processors.
More RAM will make a larger scene render faster but wont make much difference to smaller scenes, processor will make a difference all the time. When rendering in the interactive AUTO mode Shaderlight is rendering the entire render frame so will use alot more RAM, so having more RAM will help there too.
Hope this makes sense, if you still feel things aren't quite doing what they need to please email support@artvps.com and the guys will try a few things to help you with the scene.
Martin
-
There is a reason Vray is the number one biased render engine on the market, especially for visualization companies. That is because of it's complicated open framework. The biased market is all about sampling and who can interpolate the fastest with the most accurate results. The developers of Vray decided to give overwhelming control of the engines tunings to the user and allow each aspect of the processed to be tuned and manipulated. So the main difference between the Vray engine and any other biased solution is that it allows the knowledgeable user to also be more adaptive to each scene they are rendering and find the good balance between speed and quality.
Programs like Shaderlight are a little more "closed hood" when it comes to the inner framework. It doesn't make them better or worse necessarily, just better suited for someone who isn't concerned with knowing how it works under the hood and would prefer a one click solution. The argument on which is faster is again not valid as it would vary from scene to scene. For example a scene with a lot of glossy reflections may need an increase in overall sampling from a closed hood engine to get rid of the noise, whereas Vray can be tuned to specifically remedy this area, without compromising the speed in areas where extra sampling is not needed.
We've had a lot of "this vs. that" render threads and it always comes down to the user + quality of modeling skills and scene building abilities. So just find what's comfortable and stick with it until it no longer fits and then look for something else. In the beginning you need easy with good results so you can focus on the important things (modeling, scene development, etc) but once you reach a certain level, you may want more control over your render settings. No need to over think and over compare....just find what resonates with you and leave the rest.
-
The perceive weakness of vray is actually its strength. i have tried lots of software including Shaderlights. Of course when it comes to quyality you have to sacrifice the speed. But since i am looking on both (the balance), I will go with vray definitely. The learning curve is a bit frustrating, but once you conquer that particular stage, there you go.
-
But surely the elephant in the room here is:
-
Any lighting application that approximates light as Red,Green,Blue is not accurate - so these holier than thou unbiased renderer claims are often bunk.
-
If unbiased renderers are exact, then why do they need any controls whatsoever? We don't have to fiddle with our "eye controls" when looking out the window.
-
Having controls to increase / decrease and generally twiddle with "sampling" what is fundamentally a continuous effect (light) shows all renderers for what they are: approximations.
All of them, without exception.
It doesn't make them bad, it doesn't make them of no value. But just don't drink the Kool-aid.
Lastly, always remember this:
The biggest factor by a country mile in the results you get, is the skill of the operator, not the software.
Adam
-
-
I never say that the choices in Vray are weaknesses, but I don't need them at this moment. It gives other developpers their room to grow. How about a Vray basic version with a much lower price to fight the entrance market?
@nomeradona said:
The perceive weakness of vray is actually its strength. i have tried lots of software including Shaderlights. Of course when it comes to quyality you have to sacrifice the speed. But since i am looking on both (the balance), I will go with vray definitely. The learning curve is a bit frustrating, but once you conquer that particular stage, there you go.
-
@unknownuser said:
@unknownuser said:
All of them, without exception.
You and a few others in the biased render business are taking the traditional ray-tracers to task with front-ends that make for intuitive workflow and fast results. Real-time rendering has become a catch-phrase in the last few months because of it's importance. I appreciate that you have put the pressure on these guys to give more thought to their flight controls...I think some are just going to desolve as the audience for complicated biased render studios gets smaller. The tools have to be fast in order to compete now at my level...If your engine is still measuring your render times in hours I think it is time to change...I still hear folks bragging about 3d max and it's front end (where have they been) but they better like the wallpaper in their studio cause that's where you will find them.My point there was simply that ALL renderers - biased / unbiased / whatever make very large simplifying assumptions and approximations.
A casual reader of forums would be forgiven for thinking "unbiased" means "does not use approximations" which is of course is not true. All renderers "cheat" - just sometimes you talk to some where confuse how they would like the world and how it actually is..
thats all.
-
Unless I didn't read the answers properly (have been guilty of that before), none of which directly addresses owgo's question:
@unknownuser said:
........I think Vray can finish them about 30% faster than Shaderlight...Anyone know anything about it? Why is Vray faster? Or I have done something wrong there? Thanks!
I assume that both are biased render applications, so that is not the factor. If the question has a answer, I too would like to know what is the reason one application is faster then the other on the same machine?
-
As I said, Vray is most likely faster because of it's ability to fine tune the interpolation of sampling and intelligently determine where more samples are needed and where the image can get away with less. A lot of biased solutions simply break the rendered image up into quadrants of pixels and sample a predetermined number of pixels from each quadrant and then interpolate the rest. (Brute Force methods)
Think about it like someone doing a soil study on a particular lot. A surveyor will not come in with an excavator and remove all the soil to study it, rather he will take the lot in question and divide it into quadrants from which he will take samples from. He will take a vile of soil from each quadrant and send it to the lab. In the lab they will identify what minerals, organisms or contaminates are in each vile and then based on the quadrant overview, they can interpolate what the overall soil content of the lot is. If the technician determines that the results are too vague, then he will send the surveyor back out and have him subdivide the quadrants into smaller quadrants and take more samples to get a more accurate reading. This will, however, take much longer for the surveyor to collect the samples and longer for the lab to process them.
Biased render engines do a similar process, except it is working with pixels instead of soil. Based on the readings (material properties, sun ray direction, camera settings, etc) assigned to each pixel, it can determine the average result between multiple readings.. This is the interpolation that takes place. The more samples taken from each quadrant, the less guessing that takes place and the more accurate the overall image will become. But, as in the example above, more sampling takes more time.
The strength of Vray is that it doesn't just quadrify and take uniform samples from a flat image, but it will also take into consideration a lot of other factors. It will save samples for areas that are more important to the final composition (foreground items, refractory materials, dark shadows, etc) and use less samples where more interpolation would be less noticeable. This is possible due to the unique means by which Vray calculates and stores the Irradiance Map, by which it creates what is essentially a point cloud of radiance calculation and then uses that information to make more intelligent overall sampling decisions.
Vray also does a really good job at loading and unloading geometry on the fly dynamically, which helps to use the memory where needed most and then unloading it.
Once you can wrap your head around what the Vray settings do, it is not as complicated as it first appears. In fact there are really less than ten settings that you every really need to mess with and out of those, less than five that you will adjust often. By spending an hour or two with a simple scene full of primitive objects, you can quite easily see what does what. Any and all questions can also be answered at http://www.spot3D.com
I'm not sure how exactly Shaderlight works under the hood, but if it appears slower, it may just be a matter of it performing less approximation algorithms and more of a brute force attack. It doesn't mean it is better or worse, just different. Sometimes a program that takes a more brute force approach may take longer on the front end, but they also may get it right on the first try, whereas Vray may take a couple test renders to determine where the sampling tweaks need to be made. Also Vray set to DMC as it's primary will yield different render times and may come in slower but more accurate than Shaderlight on the same scene. It's all relative.
As far as unbiased solutions, there is a different framework, by which each pixels is analyzed in a progressive fashion and continuously refined based on the parameters assigned to that pixel. There is no sampling, interpolation or approximation in the same manner as a biased engine. It is a constant tracing of pixels until the user is satisfied. It is still "rendering" and as such there is still consistent query and decision process being made by the engine, but there is no "averaging between pixels" taking place in the direct manner by which a biased engine does it.
At least this is how I've come to understand it all.
-
Thanks for the detailed explanation.
@unknownuser said:By spending an hour or two with a simple scene full of primitive objects, you can quite easily see what does what.
Good advice getting to know any render application. -
adam has said it all. and in this sampling methods, the user is also involve in telling the machine how much samples you need. Here where you use your creative jusdgement which scene you need how much samplings which one you need never mind. Once you know all these dials. the frustrations is gone and it will turn to pleasure.
-
hi! i need a help please. İn sketchup i just wondering. İs it possible to get a proper render while using photomatch.i mean ,after i match a model on to the photo by photomatch i tried to render it and the photo which is behind the sceen is disappear
Advertisement