Just wanted to comment on this one. It should not make a lot of difference how complex the model is. Things like stitches work best if they are meshed and joined into one mesh.
A while back I got a model that contained about 28 thousand of objects. This takes a lot of time to initialize. Once I separated things based on material and extracted their render meshes, joining them into single disjointed meshes, render initialization goes an order of magnitude faster. I suspect that for Octane it is not any different and could make a big difference for your animations.
Not sure how you receive your car files but in my case I always get them as iges files with hundreds to thousands of surfaces. Not only does this bring the render engine to a crawl, but my Rhino viewport as well. Joining into polysurfaces as much as possible helps a great deal.
Btw it should be possible (maybe with a bit help of some scripting) to quickly render out a separate shadow pass. In the whole assignment of rendering a full (car) model this should not add much to the total time. The added benefit is that the shadow is not baked into the rendering and can be tuned better.
single disjoined meshes is a very good trick, I’m going to try this out and see the difference in times, especially in geometry cashing. I’ll report back. Thx.
For our car stuff, we get all kinds of data, funny to see the different skill levels and best practices by car makers. In most cases we mesh the data and toss the nurbs stuff, we just inherit the normal maps. In some others we remesh/retopo. Especially soft parts, because the CAD of it it’s just ‘close enough’, but inaccurate/unrealistic).
@gustojunk Thank for the details about your work. From my side I can tell that I use V-Ray since approx. 14 years now. A long time I used version 2 and skipped version 3, since it was a complete new version and quite limited. Since the last summer I use v4 in GPU mode daily.
I used Octane for rendering product shot animations with a lot of relections/refractions in the past, since it was faster than V-Ray. Last summer I tested Octane and V-Ray 4 GPU rendering and I found that Octane wasn’t faster than V-Ray at my interiors. I decided me to upgrade to V-Ray 4 and GPU. I’m glad I did the step. Quality and speed makes me happy.
I trust that Chaosgroup is a quite stable company and the current plugin continued the next years. Most I render complex train or air plane interiors. For example an image at 4000x3000 needs approx. 15min or less in high final quality. I’m impressed that 11GB GPU memory is more than enough. If necessary NVlink should allow to use 22GB by two 2080ti.
My impression is the Chaosgroup is hard working to make GPU rendering feature complete, stable and faster.
There are a lot of great points made here. I think this probably goes without saying, but the bottom line is that there isn’t one render engine that satisfies every need from every user, same as with CAD.
The majority of my companies output is based on Rhino and Octane. However, sometimes there are items just to tedious to model in Rhino so we have a more organic package as well, 3D Coat. Sometimes some shots take a long time to render in Octane or are more difficult to set up and Vray is faster or easier.
I think out of everything out there for Rhino, the two best options are Octane and Vray. I really don’t think you could go wrong with either. I think the key here is to give both a good go based on the type of work you are doing and see what works best for you. Our work is all over the place, a little bit of everything. Octane has fit our needs the best.
Someone above brought up animations and caching, so let me clarify something as we do animations using Octane frequently – while the caching has gotten faster within Rhino, the key here is that there is an output option within the Rhino/Octane menu that will allow a “one push” export of your entire scene, animation, materials, everything into the Octane Standalone. You open your scene in the Octane Standalone and render our your animation frames. It’s super simple, super quick, requires virtually no knowledge of the standalone and completely eliminates all the caching as it will take into account both camera motion blur and vertex animation (object motion blur). Speeds up your animations x5 and they look better.
Yes, you can!! This animation (while not flashy just an explainer video) was done entirely in Rhino using Bongo and was exported and rendered via the push button export to the Octane Standalone. Saved a TON of time. Notice the camera and object motion blur, which the Octane Standalone allows.
I have a basement that needs some love so I can setup a shop there, so I’m researching the whole finishing basement world, staring with insulation and all the things I didn’t know existed. I found out today what an egress window is (…and how much it cost to put in ).
I got reply from Chaosgroup: shadow catcher works for direct light only. It seems a difficult task to make it work with GI in GPU. So if you are lighting your scenes with HDRI through environment slot, you get no matte shadows. However if you apply the HDRI to a dome light instead, it should give you matte shadows.
Very cool! Yeh, definitely a lot to learn for sure. I do a ton of imagery for the construction industry, I’ve learned so much over the years, i feel like i could almost build my own home at this point.
The Rhino file for this project was right at 100MB. I exported each of the animation sets separately, as each was a different set up with different objects. There were 4 animation sets total for this animation.
I just re-exported the set with the most frames to see how long it took and how large the file was. It took just under 7 minutes and the file size was 24MB. The exported file to bring into the Octane Standalone is a single .abc file.
Thank you! The average time per frame for the far off shots was 15 seconds and for the close up shots was 17 seconds. Please note however, that this scene is highly optimized. It is using the Pathtracing Kernel, but I spent some time (maybe an hour or so?) doing tests to optimize everything for speed.
I usually only do this for animations, the still shots I typically don’t optimize quite so much. But . . . the still shots look much more realistic too . . .
At work we currently have 3 designers all using Rhino 6 and Octane V4 for the last 6 months. You are right, there are very few tutorials for this workflow. You can learn a lot from the Cinema4D tutorials (www.aoktar.com/octane is excellent) but I have found that there is a learning curve to getting a good workflow depending on the type of work you are doing. Using HDRI environment textures for lighting has made a big difference for me, since there are no great lighting rigs in the Rhino plug-in version as opposed to Cinema4D. I do like how fast the rendering in Octane is with the new GeForce RTX 2080 Ti I am using, and the quality of the renders is very good and doesn’t take too long to get there. I have also encountered a lot of bugs, more so than when I used VRay for Rhino, which is pretty frustrating but Octane support usually gets back to me fairly quickly. Ultimately I think Octane is a good value as a service overall. I get really fast, good-looking interior renders, and I can even automate rendering multiple views with the Badger plug-in, which is very helpful. If you haven’t already, do the Octane Bench test and make sure your hardware is appropriate for getting the most out of Octane before committing. I previously had a single NVIDIA Quadro K4200 and it was way too slow for Octane to work well, path tracing was impossible. Octane is only worth it with a good graphics card.