Short answer is yes I think out of the box, Octane is a lot more photo realistic than most render engines. With the right team and settings you can make any render engine look great, but for me, octane handles this better by default because it is an unbiased spectral render engine, meaning it’s using real world physics to simulate light and such
Oi… this is tempting for sure. I should probably finish this video before asking, but with a scene running geoscatter addon, animated VDBs, etc… I’m guessing this could still handle it? I’m debating on building out an 8-10gpu render server and running a local node of blender or doing something like this. 😅
You should definitely try this first! Cause it won’t cost you much at all to test it out, and you can scale however and whenever you need. The short answer is yes, if you can export an orbx, the render network should be able to handle it! One thing I didn’t mention in this video is that the size of the orbx does dictate how many nodes you get, so if you do have tons and tons of scatter elements and simulations, you may only be able to pick up a few nodes and if they are 100gb files, your internet speed and the node operators internet speed may be a concern, but I would still imagine it would be faster than 10 local machines, even on the extreme ends.
@@AlexPearce3d ooof. Yeah. Good to know. I’ve gotten the scene managed down to apx 6-12 minutes per frame with a dual 3090 which for a single 4090 would be about 4.5-9 minutes per. Worth checking out still. I guess for me it’s just making sure the cycles data translates to octane seamlessly. It’s been years since I’ve tried it.
@@AlexPearce3d 4K render with a lot of geo scatter and moving parts. Barely fits in 24gb of vram but works out. Yeah. Cutting corners where possible 😮💨
@@cgooch_ the geo scatter, if you can do that with octane scatter instead, I would bet it’s a lot more efficient and smaller vram footprint. We’re building a biome addon using octane scatter at the moment. Octane scatter is somewhat limited, but a lot more efficient then most other ways.
Thanks a lot for making this video, I was wondering for quite a while now about how the render network worked. Just a couple of questions. Did you suggest using the standalone version of octane just because of the long animation export time from blender octane ? What I mean is, can I just export orbx from my free blender for octane directly to RNDR ? My second question is, what should I realistically expect price wise for a real render ? Say 1min 1080p product render ? I know that you can't possibly give me a price, but just a ballpark estimation. Thank you again for the video, I really appreciate it.
You can certainly just file export directly from blender, and not join or test on octane standalone. And with simpler scenes this probably works with no issues at all. In general it’s good practice to test you file on octane standalone and make sure it’s good, and for me, I actually prefer doing a lot of the final tweaks with lighting/materials/render aovs etc in Octane Standalone, I find it faster, easier and more intuitive, and especially with big scenes (but that may be just personal preference). If I’m rendering an animation locally, I also prefer to render it from octane standalone because I feel it’s faster and more stable. I can’t really give a ballpark because it depends on so many factors, but if it’s a simple scene, but if you use the scene from this video, 20 frames of 1080 cost .77 cents. 1440 frames in one minute/20 frames *.77 = $55
Hello. Very interested here too. Would you mind to do a video (or get in touch directly) about how to package a scene properly ? I have been trying with a very easy packaged scene (one cube and one point light), but I cannot get my head around the correct settings. The RCWeb is working, but some of the exposed parameters just won't change for some reason.
There are many things that just don’t work when you package in the web remote, and I don’t know why :( you might be able to take touch osc or some other protocol to change those parameters. I don’t think i have the bandwidth to do anything on this at the moment.
What kind of sphere is that? I tried to put it in Shapes / sphere, and I gave it a chrome material. But it doesn't act the same as yours. Well, when I rotate the angle of the Cubemap, ( Source Cubemap Angle ) it doesn't rotate in my sphere. What sphere did you put and why does it reflect like chrome?
Is there a way to make it only render one camera at a time? Also, how do you stop it from rendering other than ending the task in windows task manager if it's going in the background?
Wow this is incredible, thanks for sharing! Though the NVLink limitation means though that for consumer GPUs, the top-of-the-line configuration we can use is two 3090s, that should add up to a a decent 48 GB VRAM. Sad that Nvidia dropped support for NVLink starting with the 40 series, so this won't work for the 4090 unfortunately 😟😟
At first, real-time rendering worked, but later on, only the phrase "waiting for image" appears and real-time rendering does not work. I tried updating the graphics card and all the methods listed on Google, but it didn't work. Is there a way?
Hi！ I successfully set up two A6000 GPU according to your tutorial and performed path tracking rendering for the project in UE5. The rendering speed doubled and was fast, but after only two or three hours, the rendering crashed.But when I Used only one GPU, everything ok. How should I set the parameters?😢
Hello! I really liked your tutorial. But I have a doubt. I'm using terescopicplayer to play stereo in a room with a 3D projector. But, the test I did left the stereo in and I wanted to make the screen jump to give it greater impact and increase immersion when people put on the anaglyph glasses. Would you help me? I'm searching the internet, but without success in finding an unrteal solution.
Hi there, very very useful tut. But I want to select just one of the hierarchy of parented objects (a child for example). then the script to do the next: select all children and the parent be the active one. then hide them from view, and next Disable them in Render. Would you help... thanks...
Thanks for the introduction, it helped a lot! Do i understand correctly, that camera position is static? Because when i try to move through my 3D interior, i can’t really change the camera position with scroll wheel in the render viewport, i only change the scale of view. Repositioning the camera through Position node works very unexpectedly, and i am unable to put it where i need.
Do you have an animated camera that you have imported? If so you can’t really change that. If not, use the camera icon then select a piece of geometry you’d like to orbit around
Does it work on AMD GPUs? I tried in my setup and it kept saying : Training Process: RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver
Omg thank you! That cache was killing my C:drive feels like I've cured cancer Bro not only has my fps increased but my computer even runs quieter now. This should have been in the default engine setting menu in the launcher. I'm sure they did this so people will go out and buy a larger c:drive ssd and another windows key while they are at it
Thank you, Alex. I was wondering about the two rendered images. How can I combine them to create a 360 3D image for VR headsets? You mentioned Assimilate, but I'm not sure where to start.
Assimilate.gitbook.io is a link to the documentation I’ve started for Assimilate. Stereo workflows are not in the docs yet and are very specific, but you can email support @ assimilateinc.com and ask for help and you can mention me as well, they can loop me in as needed. I can try to make a tutorial on it soon, I’m just slammed at the moment :/
Is the other plugin you talk about at the beginning the Camera 360 by Ivan Elizarov? I consider bying it but it is kinda expensive if it does not work properly. Nowadays the sequencer also has a panoramic mode but my renders come out extremely overexposed. Thanks for this rig system in any case.
This is great- and I have used it a couple times already- so thank you for that! A quick question, though... is it possible to disable the overlay in-game? So far I have just been clearing the post-process material for rendering but on a current project that means removing and replacing this at least a couple hundred times.. any tips?
Even this is nowadays outdated, I'm still grateful, as it tells me about specific UE plugins for my pico, which surely will become useful when I start to dive into Unreal Engine. As a personal feedback to the video... stick to the relevant, especially in a screen filled up with many information. A webcam isn't related to the topic, at best it's unless, at worst it's distracting or hiding relevant information.
Thanks for the review! Hi Alex! I I'm interested on acquiring the Zeiss Cincraft Scenario, but it is taking so long for them to make it available! Is the REtracker Bliss a viable option for commercial work in the meantime?
I remember good ole' 3D Studio VIZ R2 from 2003 having a camera matching function where you could mark common spots in the backround picture and the model. Needless to say in UE I have been a very lost at matching civil engineering models with photos so thank you for this!
