- This topic has 17 replies, 4 voices, and was last updated 2 years, 11 months ago by Eric.
October 19, 2019 at 2:04 am #2398Jeremy Hill
This began as a simple test to see how difficult it would be to reproduce a material I ran across, Transparent Candy Purple Powder Coating Paint:
Something like this should be applied over aluminum, chrome, or a silver basecoat, so I used an aluminum complex IOR substrate, and added a tinted (using transmittance) layer:
I think I can improve on this (maybe some subtle thin film, scattering, some imperfections), but for a first try, I am pretty pleased with how the layer simulation naturally exhibits some of the effects seen in the reference.October 19, 2019 at 4:17 pm #2406Jeremy Hill
Another test, with some scattering added to the layer, and some work with levels to try and mimic whatever might’ve been done to the photo in the reference:
(also note, if the color looks off, maybe try another browser, as I find firefox appears to be inconsistent with proper application of color space)October 20, 2019 at 9:09 pm #2409Tella
Very nice Jeremy! Those effects are not easy to achieve in other render engines.October 20, 2019 at 9:54 pm #2410Jeremy Hill
Thanks man, it has been a fun test so far. 🙂 Props to Albert & Oscar for coming up with this material model.October 21, 2019 at 9:18 pm #2412Eric
I had a go, I am using 19.5. I am seeing an issue (that 45 line in the IPR) when I raise substrate roughness (anisotropy is high too). Have had a crash when exiting app, and a crash with IPR on when I set layer opacity to zero.
My machine blue screened today every time I run IPR for more than 20 secs, didn’t happen before with 19.3 (not many tests and simpler material before) also my machine is really old so I don’t trust the hardware.
JD do you have time to share that scene, also curious how much our materials differ.
PS. In this earlier test, the central cap geo on the right torus appears raised, I didn’t move it?October 21, 2019 at 10:20 pm #2413Jeremy Hill
I have not seen any similar crashes, but I’ll try to reproduce using your notes, as soon as I finish something I’m on at the moment. Meanwhile, it could be interesting to see if you are able to observe similar, when restricting IPR to 1 or 2 threads, and not enabling the mat preview IPR (that uses max threads -1), just to try and see if it is maybe thermal-related.
Regarding the scene, let me clean it up and I’ll post back here with a link to download.October 21, 2019 at 11:23 pm #2414Eric
Hi, I haven’t had anymore crashes, I tried IPR 2 threads, got boring to wait for long though.
Does your scene use a hdr from HAVEN, because I may already have that.
About the material, I am learning to use a low ND for the layer to keep good saturation which is entirley logical 🙂 , also the hue affects the brightness around blues so much, which make this material a bit fiddly. Don’t really get the scattering yet, I do find the term anisotropy being used in multiple places a bit confusing.
ND1.23 This material looks just like christmas baubles.October 22, 2019 at 1:22 am #2415Jeremy Hill
Sorry for the delay, I’m pretending to be an archviz person this afternoon, and it’s slow going. :p
Here are links for two scenes:
You already basically have the second one, but I cleaned it a bit for Oscar earlier (this version has only one wheel), so I link it, whether for you or others.October 22, 2019 at 1:52 am #2416Eric
Thanks for that, one question, how can i import the material from my saved scene to the new one?October 22, 2019 at 2:02 am #2417Jeremy Hill
Sorry to say, you cannot yet do that.October 22, 2019 at 2:25 am #2418Eric
I can rebuild it easily enough this time.
Just looking at old maxwell materials which achieve a similar look by brute force additive blending and ‘eyeballing’. The realism is lost with the old materials, now I’m getting excited about Bella.
We will still need to eyeball a lot of materials though, so it would help to simplify things further.
Btw. our candy coats are very different setups,? There should be a correct setup for this material, it fits the model.
One more issue, when i go to beautypass/denoiser and click yes/no it triggers the denoise process which freezes my machine.
Maybe you didn’t see this bit about the bug, that line fades in as roughness goes up.
October 22, 2019 at 2:57 am #2419Jeremy Hill
Well, this is why I have basically been “exploring” the material system with some of these tests — the model is something the guys wrote from first principles, and like any such model, it is not reality itself, and is rather just one particular way of trying to represent reality, using a very finite set of properties.
In this particular case, I ran across the reference and thought, sounds like a layer with transmittance, over polished aluminum .. and indeed, it was nice to see some similar effects emerge in the render. On the other hand, it’s not like we have a measured IOR for this particular powdercoat, so beyond that basic intuition, it is inherently a guessing game.
So, I was playing with the sss to basically try and kill some of the internal bouncing of light in the layer — you can see this most markedly near the highlights, where my first render has much more of the bright magenta gradient.October 22, 2019 at 3:03 am #2420Jeremy Hill
Missed some of your additions — the denoiser is CPU-heavy, but it should print progress in the status bar (near top of the render area). If not, I guess you are using win7? Not sure that I have tested it on my older machine, and it’s conceivable Intel could be using some instructions that aren’t available, so I’ll check it out.
On the complex IOR material screenshot, I’ll take a look.October 22, 2019 at 4:35 am #2421Eric
The progress bar works, I didn’t mean a full freeze it was just unresponsive, I ended up deleting the whole beautypass node, that fixed it.
After I got my xmas bauble material onto the hub scene it looked all wrong 🙁 Maybe that paint has a special ingredient.
I’ll try again tomo, bit too tired now.October 22, 2019 at 12:27 pm #2422Albert Martinez
Hi Eric, we have detected recently some issues that shows albedo in scattering potentially generates wrong results, as I have checked in your material that is happening, and will be fixed asap. Forcing albedo 1 ( default value ) should be ok for now. If you need less scattering you can decrease the scattering rgb energy ( or even play with transmittance ) and if you need more transparency you can lower surface opacity while inheriting it from medium. Not same because you will lose some reflections. Somehow you can compensate it by increasing surface ior, not best solution but kind of workaround in the meantime.
- You must be logged in to reply to this topic.