Which Renderer Is the Best?

I am going to hopefully be doing a movie in the future using blender. I was wondering for rendering out seens what is the fastest, free for commercial use, and able to do soft shadows and caustics from the light source and not a seperate photon light, renderer out there? and can also render out spherical panoramas because I like to do VRs as well. Lately I have been trying Povray because I could not get Yafray to do what I was wanting. Any help is appreciated.

u can try out the new sunflow renderer… i think it needs a JVM(java virtual machine). its supposed to b good but i never tried it. search for sunflow renderer in goolge n u will get the link. . ;-)…

I think sunflow uses photons.

You aren’t gonna be able to render anything movie length in yafray or povray. They are just too slow.

For the things you want (other than speed) try indigo.

There is no “best renderer”, it all depends on what you need.

It might be quickest to learn to fake it and do in the blender internal.


i agree with IanC, it is almost always a good idea to try to fake the effects you need if you want render times down.
Render times with yafray are just too sluggish for anything longer than a short. That is unless you happen to have a render farm tucked away in your pocket:)

Speed, Quality, Cost - pick TWO

Your critera are Speed and Cost… guess what? You don’t get Quality.

On a home computer, and I’m talking top-of-the-line, bleeding-edge dual core or HT with Gigs of RAM and an SATAII array… if you want to render even just a few minutes of animation at broadcast resolution with a high quality, I hope you have another machine to use for the next year or so; while you’re waiting you can come back here and bitch about how slow raytracing is. :smiley:

Let me demonstrate.

Let’s say your machine can raytrace an average frame in your movie in 15 minutes. That’s pretty low for a complex CCIR601 resolution scene with reflections and refractions and procedural texturing… So:

0:15:00 minutes per frame, also known as 900 seconds per frame. Animation is 1 minute long @ 30 frames per second x 60 seconds = 1800 frames.

Rendering time: 900 seconds x 1800 frames = 1,620,000 seconds rendering time. Also known as 18 days, 18 hours. For one minute of animation. O-N-E minute. Chugging along, rendering 24/7, 100% full time doing nothing else at all on this machine.

Welcome to our world. You stick a few dedicated rendering boxes in the corner, you outsource to commercial rendering services like ResPower, or you fake it.

Raytracing is not fast. You really want fast? Fake it. :cool:

Oh plus env maps often look better. You can much, MUCH, nicer chrome this way.


hey guys wait for a week … right now i am downloading jdk for linux… let me get it up an i will post som images for u guys to decide… but giv me a week or two

The question of the “best” renderer is one of the most common questions. For those who are starting in 3D, the common assumption is that if it’s not raytraced, it will be an ugly render. Behind this assumption is that because raytracing is somewhat mimicking the physics of light, it will be more realistic, and therefore look better.

In practice, there are a few problems with this.

  1. There isn’t any renderer implementation that is 100 percent accurate in mimicking light-physics. Too some degree, things are still being faked. The use of a BRDF (bidirectional reflectance distribution function) or something similar is designed to mimic how light behaves when interacting with surfaces and materials in reality, which is often on a microscopic or even atomic level (in the case of real materials). For example, your model doesn’t really have microscopic grooves when rendering using an anisotropic shader in a raytracer, it’s still faked.

  2. Your eyes and brain rarely can judge between physically correct and “faked” anyway. The brain just expects a certain level of complexity to be judged as realistic. Accuracy isn’t anywhere near as important as expected complexity. Think about it. Do you ever stand a point of view where you instantly know all the angles of obscured but reflected objects bouncing light around the room or how light is supposed to refract in a complex-shaped vase? Personally, I have a hard time judging between 5 feet or meters away from me and 5.5 meters. Artistically, Hollywood has often blatantly defied physical laws for artistic effect. For example, a 3D character looking into one of those funhouse mirrors makes them look small or big according to the plot or mood, and is often exactly opposite of physics.

  3. Take a cue from Renderman. Most Renderman renderers are REYES-based, which is SCANLINE. Only on occasion will they use a raytrace for one or two refelections. The rest is enivronment maps or something similar. Would you say Pixar makes poor quality renders, etc.? You get extremely good quality from using the Renderman Shading Language, which is like C/C++ for 3D materials. Most scanline/zbuffer renderers I’ve seen make smoother images out-of-the-box simply because they make quality approximations, whereas a lot of raytracers have a harsher overly-crisp look due to using rays. You usually have to shoot even more rays into the scene for softer shadows, etc. when you could have just faked it in the first place.

  4. Have you noticed how 10 years ago (or longer), they still had photorealistic imagery in Hollywood films, and yet the processing power is less than today’s machines? You don’t necessarily have to have a huge renderfarm for quality images (of course it helps). Compositing in the sense of rendering in layers has always been used as a way to push the limits of a renderer in terms of speed and resources, and also capability. Rendering in layers/passes allows the renderer to deal with less at a time, then reassemble the final image. You can even blur highlights separately, etc. to make an effect that your renderer may not have had built-in.

One of the most common examples on the net. Trust me, Blender can do this stuff too: http://www.3drender.com/jbirn/ea/Ant.html

NOTE: With the exeception of someone like @ndy, there are going to be very few people who are even close to outgrowing the Blender internal renderer (definitely myself included, since I’m a semi-noob). Especially with the power of material and compositing nodes in the new version, it’s going to take me yet another 100 hundred years to figure out the potential of what Blender’s internal renderer can do.

Just my opinion, but I like POVRay for couple of reasons:

  1. The documentation is awesome. I don’t believe there’s a single facet of the renderer that isn’t documented.
  2. The POVRay community is very active and helpful.
  3. I find it more intuitive to create textures and macros in POVRay.

The fact that I’be been using POVRay for around 10 years may have something to do with as well…

One aspect that you may be interested in is the abiltiy to set the rendering quality to a very low level, say , for long animation sequences. This allows you to tweak the animations without the long render times. Another plus is that you can distribute POV images to numerous computers using http://www.imagico.de/povany/index.html or http://www.it-berater.org/smpov.htm

I’m setting up a render farm at work that uses SMPOV and 120 P4 XP pro machines to do my renders after hours.


Look what Ume brothers use:

and peep this: http://www.nvidia.com/page/gelato.html


I’m a beginning user of Blender. Do you use blend2pov when rendering with POV? I can’t seem to figure out how to use that darned conversion tool.


Sounds like a chat-up line: “Hey, is that a render-farm in your pocket or are you just pleased to see me?” :wink:

As for the question, I personally use YafRay, but it’s true about it being slow. I’ve never used any of the other externals, but if you tweak the Blender Internal (using AO and stuff), then you might be able to get a similar effect to yafray but with quicker render times (of course it won’t be as good, but it may suffice).

You could always pay for a renderfarm, too. There’s some website that’ll do it for you; I can’t remember the name though. Anyone want to point me in the right direction?

Edit/ Read above- ResPower is what I’m thinking orf.


Do you use blend2pov when rendering with POV?

I’m just starting to. I’m a Blender noob (I used Moray for a long time) so I’m not familiar with the interface. Sorry I can’t help you there.

However, I will eventually be using Blender2POV in an animation project. I’ll post more when I’m further along.

Right now I am using blend2pov version 6a, and it seems to work better for me. I just need to get the texturing to work.