Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.Daz 3D is part of
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Thread hijack warning, since this is Luxuscore unbiased GI, but still there is something to learn since only the Path surface integrator is GPU accelerated (Bidir is supposed to be better for interiors, but on CPU it also is slower (0,7Ms/s)).
Using the Carrara preset Indoor/Day as posted in the OP (Indoor light study 1), using OpenCL/Path settings:
Details:
Single sunsky2 luxcore light at the same position of the original Sunlight (all other lights deleted), manual shader update of chrome and matte to luxcore (Multiple shading domains do not translate with the current luxuscore version).
Speed:
1Ms/s on CPU (intel 2700k quadcore/octathread at 4,6GHz)
3Ms/s on GPU (AMD R9 280x stock)
9Ms/s on 3xGPU
Result after 6 minutes on 3 GPUs and 17 minutes @ 10 000 samples::
I'd comment on this, but honestly I've read it about 3 or 4 times and I can figure out what you're talking about.
Could you explain?
This is fascinating, really fascinating. For me the result looks a lot more like Carrara's native GI than the Octane version appeared.
In theory, these integrated plug-ins should make it very easy to compare Octane and LuxCore as renderers with the benefit of camera placement and other systems being consistent. Several of the surfaces arent quite the same in this render, and I am curious if it is lost in translation or if the plug-ins are not interpreting. One example is the orange colored tray on the table.
Pretty soon I was planning on starting a thread about critical comparisons of the two engines, as these plug-ins should enable us to compare the various unbiased renderers simply by enabling their plug-ins.
One thing I notice with OR4C is that colors in the mid gray range translate as much closer to white in Octane. Notice how much different those surfaces appear in the original post.
Based on the look of your test, the LuxCore plug-in seems to treat mid gray similarly to Carrara.
However, it might just be up to the fact that the sunlight in your example isn't quite as strong as in the original scene, so that could be why more of the surfaces aren't brighter, especially the outside.
I'd comment on this, but honestly I've read it about 3 or 4 times and I can figure out what you're talking about.
Could you explain?
Mostly I've read about gamma correction and the linear workflow in relation to images used as texture and bump maps, etc. If they're already gamma corrected it can cause issues with how the final image displays. Lighting cheats are often used to compensate for some of these issues. Perhaps I'm assuming that some of the cheats I use aren't correct for a truly linear workflow as I have seen some of them mentioned.
Mostly I've read about gamma correction and the linear workflow in relation to images used as texture and bump maps, etc. If they're already gamma corrected it can cause issues with how the final image displays. Lighting cheats are often used to compensate for some of these issues. Perhaps I'm assuming that some of the cheats I use aren't correct for a truly linear workflow as I have seen some of them mentioned.
That seems completely plausible!
FYI, is there any way I might get a hold of the Dells for myself? And today finally, I will get to play around with the scene you link me to. Fun fun fun!
Mostly I've read about gamma correction and the linear workflow in relation to images used as texture and bump maps, etc. If they're already gamma corrected it can cause issues with how the final image displays. Lighting cheats are often used to compensate for some of these issues. Perhaps I'm assuming that some of the cheats I use aren't correct for a truly linear workflow as I have seen some of them mentioned.
That seems completely plausible!
FYI, is there any way I might get a hold of the Dells for myself? And today finally, I will get to play around with the scene you link me to. Fun fun fun!
I've made some changes so I have to upload the new iteration. I'll see what I can do in the next 1.5 hours or so. I have to get supper ready, and I'm also expecting a hay delivery for the beefers at any time.
Rashad, I sent you a PM.
Rashad, I sent you a PM.
Thnx!
Luxcore ignores empty shaders on Multi shader domains, so for the orange box I made a new Luxcore material quickly. There may be a better way, but I was testing new Luxcore shaders too.
Sunlight was just increased slightly, but still have not learned its parameters, other than testing once on another scene.
Think there are some "test scenes" for Luxrender the developers use to verify each build. They probably need to be translated to Carrara to check both Octane and Luxrender... Another new project.
Gotten a little lost with the whole Iray thing. Testing testing testing. Will be back to posting in this thread any moment.
Here are a couple YouTube videos I bookmarked long ago showing how to fairly quickly setup Fake GI in Carrara for an outdoor scene.
https://youtu.be/K_m95FqGM5A
https://youtu.be/pR45MP66i5s
They were posted 5 years ago and he was using an old Gateway computer. Render time was under a minute! Imagine how fast it would render on today's faster PCs.
Gamma set on 2.2 can make some DAZ Studio renders look much better.
Those techniques seem to work very well! Thanks for including those. My own thoughts were similar.
Rendering architecture is slightly different from rendering vegetative landscapes. While cities feature tall buildings that can obstruct a lot of skylight, vegetation filled scenes have the issue of the overhead canopy. How do we get the proper amount of light in areas beneath the canopy? Using light domes with shallow shadow depth creates a washed out somewhat low contrast look within the leaves of the trees.
With full landscapes there is also the issue of grass. One can easily tell the lights to ignore the terrain so that the upward light can penetrate into the scene, but I've also found that if the terrain is covered in dense grass the grass can block almost all the light from entering the scene as well. I usually have to add a separate set of lights that illuminate just the ground plane and the grass
If we start with a photo or an unbiased render as a reference, we can more easily tell how truly realistic our fake GI rig might be.
So for me fake GI needs to do two things, it needs to be fast rendering for sure, but it also needs to look at least similar to the full GI. I've found that few more lights are usually necessary to get a look more in line with a full GI. We will discuss options more as we go along. It's very good to have you join the conversation.
Rashad, I remember the old days in Vue where artists would add many lights and even negative lights to scenes to get it to look like GI. I actually first saw it done with an outdoor scene in a French village or such and they had separate lights under plants, in the grass, etc.
When you are back, try latest Luxcore, it now supports multi domain shaders and the standard scene now renders directly...
So glad this came up, since I was planning some indoor and walled courtyard projects.
1. Rashad, I loaded your indoor scene, turned off your fake GI stuff and rendered at half size using your render quality settings, Sky and Indirect Light as you had them set. Picture 1... took 37:32 on my I7 (8 cores). Decent looking.
2. Picture 2... took 7:30 sec... 1/5th the time
So close in appearance, I'm not sure I care about the difference.
What's the trick?
The idea is to fake GI using :) GI itself !! You add very very low intensity bulbs along the windows (Picture 3) so they barely illuminate anything BUT they add new light rays that help guide the calculations. Apparently because you added more light rays you don't need high settings in the Photon and Light Quality settings any more. (Picture 4) More rays, less blotches. So I kept lowing the Light Quality settings till they hit practically rock bottom... that got me Picture 2 in 1/5th the time.
Hope this helps. Play around with it, its late and I hope I got this right.
Project file Here: http://1drv.ms/1BL11eO
Don
Wow! This is incredible cleverness. Will explore. Thanks so much for your additions and please do continue to post your experiences. Fun fun!!!
Don, that is so cool. Actually I think the fake GI + GI render looks even better.
I've also been using the Fake GI method (replicated bulb lights) combined with real GI in a similar fashion recently, using PhilW's portrait studio setup and adding a fake GI plane of bulb lights. It absolutely increases the render quality from the standard GI to get the nice specular effect on character's skin, but I hadn't even thought to time the renders and compare, to see if possibly the extra rays added the scene might actually speed up the render time, I'll have to check.
That's a very cool find
Thanks Rashad and Jonstark... I worked on this late last night so please verify my findings. Maybe we got something interesting here, maybe a game changer :)
- Don
that's a wonderful outcome Don, so good I thought I should bump it
Roygee, just saw your Ot response to my pirate - sorry about that ;) !