I won’t be able to test it atm but I will as soon as I get home.
Great, remember to get the updated CUBIN files as explained here.
V6 is unstable right now, so I have had many crashes through the few tests I have ran. But the speed is great and the result is good too. But there is some strong banding going on, and that isn’t nice.
Here is a simple scene with some plastic materials and an illuminating one.
What we need for this to be a competitive product is in my opinion:
- an intensity multiplier for the illuminating materials (@pascal)
- a exposure setting right there on the viewport so we can add lights with out having to lower the environment multiplier, or all the other light values
- a slider to rotate the environment from -180 to 180 degrees.
Bugs:
When clicking the object properties tab I often only see one setting, or I can’t adjust them, so I have to click another tab first, and then go back to the properties tab. Then it works.
The text doesn’t position itself correctly on the gray bottom field, making it hard to read when background is white.
Ground plane material updates on the fly when I alter it, but then when I close the panel it goes back to white. I have to switch to another display mode and back to Raytraced for it to update.
Crashing while Raytraced was running has occured when I copy an object, when I do a boolean operation and when modifying materials while multiple objects were selected. (No bugtracking was done to make it repeatable, sorry, no time for that yet)
Edit:
Another crash when I tried to turn off a layer. (The crashes are similar to those I was used to with Neon, maybe they are related, maybe a coincidence)
'Edit: I just had another crash with another file whiles I tried to hide a few layers while Raytraced dislay mode was running.
I haven’t seen this, nor can I reproduce with latest builds. If that is the case this particular bug may have already been fixed.
https://mcneel.myjetbrains.com/youtrack/issue/RH-35773
This is related to another bug which I’m working on ATM: https://mcneel.myjetbrains.com/youtrack/issue/RH-34922 (@BrianJ, if this is a youtrack item that can be publicly visible, can you change it to be so?)
I’m currently completely reworking the way geometry data is collated, aggregated and pushed to Cycles, I think that work will also fix this error. The layer problem is also a manifestation of the same bug.
Thanks for testing!
/Nathan
1 Like
Done.
1 Like
My pleasure, anything you would like me to stresstest?
Another thing I notices is that when I open a v5 file then starting the Raytraced display mode is quite fast, but if I reset the render settings (by using the 00 icon) then starting Raytraced mode can take quite long to start. I’ll test this out some more to see if I can find a consistancy in that.
Anyway, tested it out on a watch I made with wip to check out edge softening as a sketch tool and I must say that Cycles is really bringing the default render quality to vanilla Rhino that Rhino has always needed. This image has 500 samples and took 1 minute and 10 seconds. (you can of course have the file if you like)
Edit: in comparison: Using the dual 6 core XEON CPU (12 cores 24 threads) to calculate 200 passes took 2 minutes and 3 seconds. 200 passes on the 1070 takes 27 seconds, that’s 4.4x faster…
At the moment I don’t need specific stress testing, as I have two major overhauls in the pipeline: geometry aggregation, and improved materials.
Until that is complete just regular testing should be fine. Make cool renders and make people drool (i.e. make them want Rhino WIP/V6), esp if you can do that with the rhino materials and the material library 
/Nathan
1 Like
Another test with the metal material. Only adjusted the reflection slider and the main color. Groundplane given a dark metal material. Default environment lowered from multiplier 1 to 0.5 and added two rectangular lights, one slightly warm and one slightly cold. 500 passes. Rendertime 48 sec.
Notice the banding, is this due to 16 bit instead of 32 bit?
1 Like
So I did some very brief testing last night and pretty much all I can say at the moment is…Rhino Cycles works on 10x series. Main thing I’ve notice is how long does it take to switch to Raytraced viewport.
On this example top picture is Rhino Cycles bottom one is Thea Render.
What you can see on the Thea picture is a result of about 6 seconds…the time needed for Rhino Cycles to start rendering.
Cycles picture is about 15 seconds on after those initial 6 seconds.
I’ll do some more testing over the weekend but my first impression was “It’s so laggy”
Does your model consist of a polysurface created out of many surfaces? If the answer is yes, then this is the reason for the slow start-up and rendering.
You could verify my statement by running ExtractRenderMesh for your object, delete or hide the original one, assign material and start rendering. Now that only one object exists with no sub-objects rendering should be much faster.
This is the precisely the area I am currently working on: aggregating geometry data into fewer objects while retaining material assignments.
/Nathan
Oh I see, that’s a good news I wasn’t sure what you were saying by geometry aggregation, now I know
Thanks
Small metal test.
2 Likes
Feel free to post your awesome render results in the Gallery category
/Nathan
I’m testing, but with 2x980ti and it’s working very smoooothly, just one crash. I can’t find the settings for it though, I want to set higher than 200 passes.
Edit:
Found it, only as typed command though: RhinoCycles_ChangeSamples in case anyone else is looking
You can also click on the pass count in the lower-right hand on the HUD. Note that the text is drawn wrong, but the edit box is there.
/Nathan
1 Like
Thank you!
I get some white pixels in the renderings when I let them run for a long time.
Sorry I can’t show the whole rendering but there are more like these, it doesnt seem to be the material cause it’s on other materials too.
Did a little bit more testing today and the results are promising…apart from Rhino crash or blue screen every 5min or so when I play with Reflectivity, Transparency, Environment sliders 
2 Likes
That shouldn’t happen of course, maybe some tweaks are needed to glossy filter and related settings. Does your scene have (many) glass or gem materials? Is it a set up you could recreate as a simple scene which gives similar outliers/fireflies?
I have noticed increased instability with (quick) shader changes too. I actually think that the two big changes that I am working on should mimimize if not eliminate such behaviour.
/Nathan
Testing out the focus, but something is wrong with the aperture, it works the opposite way as a physical camera and thus is miss leading. (Should be like: Low number = large opening -> shallow focus)
14 seconds:
Bug: Raytraced mode doesn’t update material changes done to curves with “piping” unless they are moved.
I see that camera sample count and jitter are not applying to cycles way of rendering, as all here is based on passes. Here’s a strong focus blur with 3000 samples. Nice and smooth:
