Luxrender and Poser
jorge dorlando
Posts: 1,157
Excuse me, but I did not read every page of this topic, because I have to translate every message to the Portuguese of Brazil (but I read a few pages)
The question is:
how LuxRender renders poser animations, this is too slow?
my video card is a gtx560se
Comments
I think each frame would have to be rendered by itself, then the frames combined into video afterward. Not positive though; it isn't something I'd consider doing due to the time involved in rendering each frame.
I think each frame would have to be rendered by itself, then the frames combined into video afterward. Not positive though; it isn't something I'd consider doing due to the time involved in rendering each frame.
well,
I contacted the responsible for the reality3, and he said yes, LuxRender renders poser animations.
I bought reality3 for poser (and there there is a button to render animations in reality plugin).
I did not tested yet animations, still images only, but all I got so far is a lot of noise in renders (the camera and the lighting is a big issue)
On the other hand, I downloaded a demo version of Octane plugin, and this renders the same way as poser camera sees the scene without adjustments.
I liked the octane campaign: fast rendering
but the price of the combo octane (Euro 330.00), when converted to my currency exceeds the sum of R$1,000.00 (Brazilian currency)
and I would have to buy a new video card with at least 1500cuda core ...
Split into it's own thread, as this conversation did not belong in a thread about DS and Luxus.
Oh,
you created a topic for me! Thank you!
Okay, some important things to know when working with Luxrender.
Firstly, by default Luxrender uses software rendering and NOT hardware. It does have SmallLuxGPU which is a cut-down version purely for graphics-card only rendering, but as a result the textures it can manage have to be bite-sized and it cannot use many of them. This is a hardware restriction and NOT a restriction of the software. Octane also has these limitations since it's a pure GPU render engine, which is why they recommend very beefy graphics cards. Luxrender does however have a hybrid mode which allows you to use the GPU as well as the CPU in a render, theoretically giving you faster render times. However, even with recent optimizations, there are still renders which are slower with GPU enabled than they would be as a pure software render.
Secondly, animations can be made but you need to render them as individual frames and piece them together. Luxrender doesn't create movies by itself and can only render static frames which you can queue up so it will render them back-to-back. In order to do this you need to tell Luxrender how many samples it should take of each given scene before moving onto the next frame. Depending on your system and scene, this can take up to 1000 samples per frame so be prepared for a long wait. The advantage of rendering it as individual frames however is that you can re-render any frames you're not happy with without having to redo the entire animation from scratch.
Reality will automatically divide the scene into individual frames for you, so you don't need to render them one at a time. Just let Reality send it to Lux and it will automatically queue up the frames for you, but be sure to tell Lux how many samples it needs to stop rendering each frame or it will carry on rendering forever without manual intervention!
Hmmm, thanks for the explanation!
Now ... This amount of noise in the LuxRender renders, is hair pulling.
I thought that was going to get those images (renders) resemble Hollywood animations in no time ...
I would like to see a render his lux
I'm also a demo of octane (with reduced yield, and watermarking).
Octane seems to be better, but otherwise ... How octane would be able to handle a scene with more than 10 characters, scenarios etc. with a videocard only 4gb of ram?
Even though there are 1500 cuda core, but with little memory on the video will charge off memory, right?
I'm also a demo of octane (with reduced yield, and watermarking).
Octane seems to be better, but otherwise ... How octane would be able to handle a scene with more than 10 characters, scenarios etc. with a videocard only 4gb of ram?
Even though there are 1500 cuda core, but with little memory on the video will charge off memory, right?
I don't know if you could fit a scene with 10 or more characters into 4Gb, and the texture limits of Octane. By using Octane procedural shaders instead of texture maps where possible, and re-using the same texture maps when possible (for example using the same texture maps for 2-3 characters, and making changes in the shader setup to make them look different), it may be possible.
For general reference, the attached image took about 1.7Gb of VRAM when rendering in Octane, but the scene consumed 3.5 Gb in DS. Like the DS plugin, the Poser plugin does a lot of scene optimization to reduce the amount of ram used by the video card. You can also reduce the size of texture maps when the extra detail/resolution isn't needed, to reduce the amount of ram needed. 10 characters would definitely be difficult, but I don't know if it could be done.
I only have 3Gb on a Fermi based card (64 color texture map max), so I can't really try it out for you. The speed and quality of the renders from Octane is pretty amazing!
LuxRender isn't "noisy", it just takes some images a long time to clear up. The noise eventually goes away the longer you let it render.
A couple of my Poser/LuxRender images are below:
I don't know if you could fit a scene with 10 or more characters into 4Gb, and the texture limits of Octane. By using Octane procedural shaders instead of texture maps where possible, and re-using the same texture maps when possible (for example using the same texture maps for 2-3 characters, and making changes in the shader setup to make them look different), it may be possible.
For general reference, the attached image took about 1.7Gb of VRAM when rendering in Octane, but the scene consumed 3.5 Gb in DS. Like the DS plugin, the Poser plugin does a lot of scene optimization to reduce the amount of ram used by the video card. You can also reduce the size of texture maps when the extra detail/resolution isn't needed, to reduce the amount of ram needed. 10 characters would definitely be difficult, but I don't know if it could be done.
I only have 3Gb on a Fermi based card (64 color texture map max), so I can't really try it out for you. The speed and quality of the renders from Octane is pretty amazing!
Hello,
Wow, That's good to know! Incidentally days ago I sent message to the contact email of octane, making this same question, but they did not respond me to this day, still no response. Now, you gave me the answer.
My computer has 32gb of ram ... But the problem: octane is purely gpu ... then while LuxRender is cpu, I'll be better off with him ... I think
the problem with LuxRender is that after the scene is ready to export via reality, then whenever the camera and lighting are now out of place ... and work there and so far not hit a render in lux.
I don't know if you could fit a scene with 10 or more characters into 4Gb, and the texture limits of Octane. By using Octane procedural shaders instead of texture maps where possible, and re-using the same texture maps when possible (for example using the same texture maps for 2-3 characters, and making changes in the shader setup to make them look different), it may be possible.
For general reference, the attached image took about 1.7Gb of VRAM when rendering in Octane, but the scene consumed 3.5 Gb in DS. Like the DS plugin, the Poser plugin does a lot of scene optimization to reduce the amount of ram used by the video card. You can also reduce the size of texture maps when the extra detail/resolution isn't needed, to reduce the amount of ram needed. 10 characters would definitely be difficult, but I don't know if it could be done.
I only have 3Gb on a Fermi based card (64 color texture map max), so I can't really try it out for you. The speed and quality of the renders from Octane is pretty amazing!
Hello,
Wow, That's good to know! Incidentally days ago I sent message to the contact email of octane, making this same question, but they did not respond me to this day, still no response. Now, you gave me the answer.
My computer has 32gb of ram ... But the problem: octane is purely gpu ... then while LuxRender is cpu, I'll be better off with him ... I think
the problem with LuxRender is that after the scene is ready to export via reality, then whenever the camera and lighting are now out of place ... and work there and so far not hit a render in lux.
Glad it helped! I get the impression that the people at Otoy are very busy right now, plus they probably don't use Poser figures that much.
With Reality/Lux, it's usually best to set up lighting specifically for rendering in LUX. Reality does a pretty good job of converting Poser lights to Lux lights, but you typically get better results if you design your lighting like a photo studio, or in your case, a movie set, using mesh lights. Also, fewer lights usually help to increase render speed.
You might give the "Hybrid GPU" renderer a try to see if you get faster renders. It uses both the CPU and GPU to render the image. I get a 1.5x-3x speed increase with the Hybrid renderer.
There is also a lot of good information in the Reality Users Guide that will help you get good results faster - if you can translate it.
Good luck!! You have an awesome render rig!
Another thing that Lux does very well...network rendering.
It is very easy to set up a network render in Lux. You start Luxrender in console mode as a server on each machine you are wanting to connect to. Then you start your render on your machine. Then, in the LuxGUI, go to the network tab and add, by IP (I think you can use name too) all the machines on your network.
You need to be running the same version of Luxrender on each machine...but every additional core you can add will help. I've run renders on my wife's laptop, daughter's laptop, son's gaming computer, home theater computer and my own, for a total about a dozen CPU cores, at the same time...and even though some of the machines on my network aren't very powerful, all of them together were much faster than any of them alone.
http://www.luxrender.net/wiki/Network_rendering
There should be info on setting up a network render (renderfarm) in the Reality user guide, too.
LuxRender isn't "noisy", it just takes some images a long time to clear up. The noise eventually goes away the longer you let it render.
A couple of my Poser/LuxRender images are below:
Hi,
these renders her ... They are beautiful, the scene is thought, the composition ... but still seem to internal rendering poser and daz.
his first render looks rendered internally in poser, and the second seems rendering Daz studio
But as I'm saying "just looks", I believe are rendered in lux as you mention.
I think maybe not worth all the delay in rendering, if the result does not greatly exceed, in comparison to the internal renderer poser / daz.
I'll look at what dustrider said, about "using mesh lights" ... I think not installed this component, no.'ll Demand it here.
Below are 4 pictures, including 2 rendered in poser pro2012 without postwork, and then the same, however now postworked in image editors.
is somewhat what I expect to find in lux ... these effects in images ...
Since Luxrender is a physically based render engine, it excels best when used to make realistic scenes. It can do this easier and more convincingly than 3Delight because, as mentioned before, it will calculate light rays as they would work in the real world. With that in mind, cartoon style renderings are not Lux's strong point. It's clear you're fairly skilled in using the Firefly render engine, and I would argue that with the same level of skill in Luxrender you could not only bring render times down, but recreate similar scenes using it.
Both engines are capable of the same level of art, but it really is down to how you work with it which determines how quickly it can render.
As this started off as a question about animation, I will also throw out that animation doesn't necessarily need you to render to the same number of samples that you would for a static frame. You can afford small bits of noise when a frame is only up for 1/24th of a second. What is more important is that the noise is consistent. Hence why it is recommended to use the low discrepancy sampler in Lux when rendering animations (Reality does this for you automatically when you click the render animation button).
So the long render times of Lux might be less of an issue when you're only rendering each frame to a couple hundred samples per pixel.