I’ve done some tests with a GraphicAll build of Blender 2.5. I did not pay attention to render times, I just wanted to make a few comparisons.
Here is a render from Blender 2.5 build 27614
Here are some tests done in 3DSMax R9 SP2.
MentalRay w/ Final Gather
Scanline Renderer w/ LightTracer
While I did not time any of these renderers, on all tests, rendering times were mostly normal and acceptable, except when I didn’t use a cache.
Here are the files for you to do your own tests.
GraphicaAll build of Blender 27614-render25win32
Scene files are below. You will need to rename the Max file to *.max. The forum did not allow me to upload that format.
cornellbox.blend (341 KB)cornellbox.max.not.blend (268 KB)
Hello
One thing you need to keep in mind when working with the Inverse square fall off of light and indirect lighting is that scale is very, very important when trying to get accurate results.
Try again, but re-scale your model down to 2 Blender units in height. ( ie: 2 meters tall) -
http://www.pasteall.org/pic/show.php?id=2008
Another thing,… I believe there is also a serious bug with shadows and the render branch as well - In the latest Render branch builds, moving the light outside the Cornell box give wacky results.
http://www.pasteall.org/pic/show.php?id=2009
Take care
Ooops, almost completely forgot…
Color Management is also VERY important when trying to achieve accurate results. Be sure to always have “Color Management” checked in the Render tab. and then re-select your color values in the material section…
http://www.pasteall.org/pic/show.php?id=2010
take care
Hmm, thanks. That was very informative - however - the scale of the model should be up to the artist, not the program. I should not have to worry how big my model is except its size in relation to other objects (and my own personal preferences). That said, I recognize this program isn’t even in Beta yet.
Have you tried cranking up the power of the primary light, indirect light, and increase the distance that the indirect light travels?
Well, you can use what ever scale works for you, however you have to keep in mind that math is math - light falls off at a very specific, physical rate, meaning that photo-realistic renders should conform pretty closely to the laws of physics (add artistic license where desired ) .
There are, of course, other light fall-off options such as “constant” or “user-defined” that can help you create the look you’re after.
good luck
Here’s the cornell box rendered in an old Lightcuts build.
cornellbox-lightcuts.blend (186 KB)
I did some tests, and found that you are right. In Blender 2.49, the default mode is Inverse Linear. According to my tests with that, the scale of the scene did not matter at all provided the light Distance was proportional to the scene. If the scene were scaled by 10, so long as the Light Distance was multiplied by 10, it would look the same.
However, with Inverse Square, you are correct. I was just unaware that this was the default selection in Blender 2.5. I hope they change that, because it could confuse a lot of people.
Hey Hauzer, I took your Max rendered cornell box as a challenge to see if the same quality can be done in the render25 branch and I believe I’ve gotton pretty close.
In this new thread (so it doesn’t get lost in the tests in this one)
http://blenderartists.org/forum/showthread.php?p=1593228#post1593228
Heh, ok.
Here’s an improvement on the first one. I think it looks a lot better now. Color Management was turned on, Indirect Lighting Factor was changed to 0.5, and the light was changed to Inverse Linear.
cornellbox2.blend (349 KB)
