Any plans to fix RAYTRACED mode or is it now where its gonna stay?

wTf?

I can’t believe I just posted that, better go wash my mouth out with some Dr Bronner’s Pure Castile Soap!

Seriously, how has this forum been such a refuge from the stuff being spouted in this thread? I can suggest a couple of forums where it would fit right in if needed.

Guess it is matter of taste then. Won’t comment further on all caps etc.

I addressed my recent issues/bug within Cycles in these two threads:

Thanks for your contribution to the second one!

Hoping to help fix bugs and help get cycles ready for prime time. I need it occasionally, but when that happens, I need it to work reliable.

I give you credit, @cfee , you’re a man of principle!

Or is that PRINCIPLE?

Rose colored glasses ? Maybe. RENDER-mode: 0.0 (zero point zero) seconds per frame. NEON-mode ~30 seconds per frame with CONSISTENTLY EXCELLENT results on EXISTING video cards. RAYTRACED-mode 30 - 55 MINUTES per frame with consistently NOISY results right up to the last on a ~$1000 video card and the “free” Real-Time-Render ( :=)) ! ) option, with improvements “promised” some time in the “future”, even tho this has been debated since EARLY - WIP. Rose colored glasses ? Maybe.

Can we see some of these files you are testing? Like an actual 3dm? Cause your results don’t seem like you are comparing apples to apples, or have something grossly misconfigured.

9V Torch 3d.3dm (14.6 MB)

Here is the file

And in V5 all you’re using is Neon with the basic Rhino renderer, no Brazil?

'Cause yeah, in the case of this and the original rendering at the start of this topic, we’re talking about setups that are so basic you’re not actually seeing anything of what Neon or Cycles can actually do. Cycles in pure CPU mode is certainly slower than Neon, you would expect that since that’s what it was made for, but those numbers would indictate that you’re using the CPU instead of the GPU, and/or the GPU is not suitable for this task. What’s the hardware?

@ec2638 -

Yeah, its DEFINITELY “PRINCIPLE” ;=) !

The numbers really work tho. Seriously -

I was on a project for Chevron contracted to them by MY parent company Foster Wheeler. Both BIG names in their respective industries, and I was the CAD Lead for the Chevron Hurricane Restoration Project. THE_CAD_Lead , for the PROJECT . Don’t get me wrong, we had a STRONG team, and I was genuinely very fortunate ! We started with RH-3, migrated forward to -4, and topped it off with -5. SHORTLY (I mean SHORTLY) after adopting-5 and getting everyone up & running in addition to my duties off-shore, I was hit with throat cancer. As a non-smoker we were all taken by surprise, but working offshore, … ! I came through the treatments - I’d chosen the most vigorous and brutal combination of therapies … but that’s a different story for another time - but the point here is that I was blessed to survive (almost 9 years now clear and CLEAN !) and was able to return to work.

SO- focused as I was on returning to WORK, I was able to do ONE MORE rotation offshore but instead of releasing me back to Foster Wheeler, they chose to set me up in an office at one of their Houston Locations. There, one Monday morning I phone-conferenced in to a weekly meeting where they discussed a BRAND NEW project initiative (Look up the Versabar VB-10,000 on Youtube). I KNEW I could get them something that would solve their communication needs, and BEGAN work on my first BONGO Animation. We had BO-1, had NOT upgraded THAT, so I had to run it on -5/32. A week later, I had my first full-run experimental animation I sent to the Chevron V.P. over the project, JUST as a “concept” of what we "could do if … " . He ended up using this silly little 3-1/2 minute animation to communicate the “Concept” of the initiative, and boy, we were OFF to the races ! I was able, using every trick I could squeeze out, to get them Animations on Thursday of things we had discussed on Monday, FULL-BLOWN and READY TO ROLL !

Needless to say, then, I’ve got some pretty strong opinions when it comes to SPEED of Production MARRYING it to QUALITY as RHINO can DELIVER ! So when I say ZERO-POINT-ZERO seconds per frame, I’m NOT just whistling in a hurricane, even IF the Hurricane is of my own making ( ;=) ! ) . THEN to get the QUALITY boost that NEON provides with only 30-seconds-per to a GREAT frame-image-quality, well, you can spot RIGHT AWAY why I’m PREJUDICED (PRINCIPALLED ? ) against 30-minutes to 55 minutes / frame and NOISE still in the FINAL resulting image, and we’ve seen this stalled at this point even tho we hear all about “Cycles” and PROMISES for development … since early in WIP-Land !

Requiring $350 to a THOUSAND dollars for an upgraded video card to run a “free” plugin (“Raytraced”) is NOT the answer. By today’s standards, my CURRENT video card (about lower-mid-range) STILL gives me ZERO-POINT-ZERO seconds per frame, ~30-seconds with NEON, and I’M_NOT the one reporting 55-MINUTES per frame, someone ELSE here is reporting THAT. However, I just give up at 5 - 10 minutes per-frame, and even FULLY RENDERED examples shared elsewhere in this thread STILL even WITH “upgraded” video cards, report what they are willing to admit are excessive render times, (without disclosing the numbers) and their samples STILL show WAY too much noise, and THOSE are at the END-CYCLE capture on “UPGRADED” ($$$) video Cards.

Moral of the STORY ?
If I need QUALITY of NEON, I just drop back to my -5, and my CLIENTS LOVE THE RESULTS ! How do I KNOW ? They PAY THEIR INVOICES ! Too many “CAPS” ? Sorry ( ;=) ! ) !

Sincerely, ABSOLUTELY No animosity ( beyond some basic frustration) is intended, but I won’t brook any explain-it-away silliness, either . I am GENUINELY GRATEFUL for a WONDERFUL product, and and am SINCERELY GRATEFUL to the ENTIRE McNeel Team for an outstanding effort - and their WONDERFUL product ! I ALSO want to thank those here who genuinely try to help each other. This may be ONE OF RHINO’s best features ( ;=) ! ) .

Thanks, and ALL my best to ALL here and at McNeel -

-C.

2 Likes

Just out of curiosity - have you tried timing the difference between your CPU and your GPU? It may very well by that your CPU outperforms your GPU (then again, it may not be the case, in which case I’m upfront sorry for wasting your time).

Thanks, I get some odd results with the materials so @nathan can you look at the black material on the battery and figure out why reflection goes to 100% mirror even if the slider is sat at 5%?

Anyway, I removed all lights and materials and made new and here is a quick test to check speed and noice:
This took 2.46 minutes on a GTX 970 card:
500 samples, aporox the same size as your images.

If you could attach this great file of yours, sure.

Doh… It isn’t my file, it is the one from a few post up.

Try controlling the reflection amount on the black material. It is mirrorlike and I can’t figure out why.
(maybe a V5 → v6 conversion thing)

@cfee

Ok cfee, I can be quite the wise arse at times, and I was initially tempted, but, I’ve decided you are OK in my book. Still, if we are being honest, you could do me a favor, and try to be frugal with the CAPS. Not because there is anything wrong with emphasis, rather, I’m not that smart and I’m lazy. Why do you want to make it harder on me to read and respond? It’s not like I’m getting paid for my free crappy advice…:wink:

Back to your problem:

Ain’t no one going out in the Gulf and poking no horizontal holes without spending a few nickels, right? I mean, it’s a bummer, but times change and you might need to reach into your pocket, in future, too.

That said, perhaps there is/will be, a way to tweak up Raytraced to get you where you need to be. I’m no 6 expert yet, so can’t help, though perhaps someone else here can. @nathanletwory might be on the right path regarding CPU, with your system.

Real-time (or close to it) is the future via the GPU. In future, an investment might be necessary, unfortunately. Stuff changes.

I failed reading.

The material is missing Fresnel Reflectivity making it metal-like. Turning on Fresnel Reflectivity has the material behave wrt reflectivity amount.

But a black material with 5% reflection shall not act like 100% mirror.
It shall act like 95% a black material and 5% a mirror. Am I not right???

Frensell is just the feature that it reflects more at a different angle. Which is an addition, not a replacement. A feature I think McNeel got wrong regarding materials. Not being able to adjust the three values for Fresnell is a bottle neck too. (Angle, max and min reflection)

Yes you are right, I was merely suggest out of current behavior. I have somewhere a YT item to look at fresnel vs non-fresnel. I have a conversion bug possibly there. I think assuming metallic workflow and using wrong data.

1 Like

Here is a simple comparison of V6 vs V5 default Neon (with skylight turned on)

I sat both to white viewport background and added a default gray envrionment to both too, to compare “apples to apples”. All renderers are stopped at 1 minute sharp. And uses the same resolution.

V5: (On the old i7 CPU)
19 passes!

V6: (On a 970GTX)
259 passes!

V6: (On the old i7 CPU)
35 passes! (It took the GPU only 10 seconds to reach 35 passes)

So you can see that a good GPU (not top notch) is that much faster.
Of course you can argue that Neon has less noice with the same amount of CPU rendering time, and that is correct.

And maybe “Cycles”(Raytraced) is sat a bit too bright, or maybe V5 is too dark. In V6 Rendered OpenGL mode this is how it looks: ( Which in appearance is closer to V5 Neon than V6 Raytraced.)

V6 Rendered (OpenGL at 60FPS)


And it gives a testmaxspeed of 60FPS (which is limited by the screen syncronisation)

Another thing, this is a must to turn on for Raytraced (Cycles):

It makes big pixles for the first frame, like Neon does, and makes large files much faster to navigate. V6 should have this turned on by default, and @nathan I think you should add that to the next SR. On small files it does not matter, becase the large pixles are gone in a flash, but on large files it makes the scene so much easier to navigate, so please fix that.

2 Likes

Best way to “fix” raytraced is to test it out and give proper feedbacks.

It’s improved much already from when it was WIP. No reason it wouldn’t improve further.

As for having to spend money on GPUs… for me it’s fun to search for new devices and buy to experience something better. But that’s just me…currently very interested in giving up my current dell laptop to new gigabyte aero 15X with 6core and a 1070gtx.
Raytraced and anyothers will be quite fast.

I personally like to have also RhinoCycles.UseFastDraw enabled. It is yet a bit faster than just RhinoCycles.UseStartResolution. Note that the graininess of the progressive refine can be managed with RhinoCycles.StartResolution.

Also regarding the brightness of Raytraced you can always opt to set the gamma to something smaller 1.5 or 1.8 look nice with a bit more pronounced shadows.

1 Like