A while ago I'd looked at Octane for UE - to potentially use on my various local machines here, but didn't really delve into it much yet. So - I didn't really watch this full video. But my quick questions are - what kind of total file sizes were there for the 2,000 frames of 12K you mention at the beginning? And what was the approximate cost of that? And can this sort of cloud rendering workflow be used with UE? (BTW, was the OWL 360 MRQ of any value in your workflow?? (though I think you just got the alpha thus far))
Hey tom! I used owl and it works pretty well all things considered. If you’re using unreal, I think it makes more sense to just use the owl plugin. Octane does have an unreal plugin and you can technically use it and save out the files needed to render in the cloud but I think it probably makes sense only on smaller projects and such.
@@AlexPearce3d I'm still dabbling in 3D for VR (though I do have someone who wants to shoot an actual s3D project with VP!), so just looking at tools and such. But meanwhile, quicker high quality rendering is always a good thing ... although with my bandwidth here, I'd be limited on what I could realistically upload to a cloud.
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
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.