Chaos Corona Forum
Chaos Corona for 3ds Max => [Max] I need help! => Topic started by: vossiewulf on 2015-09-02, 10:32:14
-
I hope the answer here is that I don't have Corona/environment set up correctly, which is fairly likely since I have roughly one total hour of experience fiddling around with it.
I decided to use a very low poly (300) game asset I did with a single texture and material for simplicity's sake when experimenting with a new renderer. And I'm pleased with the results for just a little fiddling time, and Corona appears to be massively less complex to configure than mental ray or Octane or pretty much anything else. I'm using pure HDRI lighting here.
Question is WRT the reflections - it looks to me like Corona is using the smoothing group info on the diffuse channel but not the reflection channel, at least here it looks like every quad is being treated as a partially reflective flat mirror surface. So is that the way it is and you pretty much have to use high poly models to get continuous reflections, or is there a setting I'm missing that will fix that artifact?
I'm hoping there's a setting as I've been doing low poly models but also would like to do some nice still renders on occasion without having to modify the geometry. But I'd also assume that's an edge case, and it's going to be pretty rare that someone wants a nice rendering of a low poly model at close ranges.
-
Hi,
Screens? Renders?
Which version of Corona? Max?
:)
-
Hi,
Screens? Renders?
Which version of Corona? Max?
:)
Yeah I just did captainpicardfacepalm.jpg. Smooth, very smooth.
Max is 2015, Corona is the latest - 1.2 hotfix 1, I'm doing the 45 day trial. It's doing pretty well on day 1.
Below is render. You can see what I'm talking about very clearly in the reflections on the cab. Thanks for looking into it ;-)
(http://i.imgur.com/h65CafS.jpg)
-
Nope, it's not Corona's fault. That's how raytracing works. You'll get same result in MR, VRay, etc. You can see same distortion even in game engines.
-
Yet I was not able to recreate this. Any tips?
-
Nope, it's not Corona's fault. That's how raytracing works. You'll get same result in MR, VRay, etc. You can see same distortion even in game engines.
Odd that never registered with me, I've been an inconsistent modeler but certainly have made use of many raytracers. But then again, I don't recall ever doing so with such a low-poly model, whenever I've gone for render quality the objects have been reasonably high poly counts. Thanks, guess I'll live with it, as mentioned to turn the geometry into something nice wouldn't be worth it.
Yet I was not able to recreate this. Any tips?
I think you'll see it if you rotate your object 90 degrees.
-
hmmm, I am probably doing something wrong, can you help? ( no irony, I am serious! :D )
-
Most probably it's the smoothing grops causing this isue try to asign them manualy
-
Agree, smoothing groups or some problem in the mesh I think.
Cheers.
-
Sorry for delay getting back here, this is hobby for me now, not work. And work decided to impinge with a 20 hour day.
Ok, first there is nothing wrong with the mesh that I'm aware of. I learned modeling in 3DSr4 and have basically always done mesh modeling. There are zero open edges, zero multiple edges, zero isolated vertices, and there isn't a single face or vertex that doesn't absolutely have to be. Not that there can't be something I'm not aware of, but I'm pretty confident it's not a mesh problem. See below:
(http://i.imgur.com/ldKDG9L.jpg)
So next question is smoothing group problem. First I cleared all SGs and rendered again in Corona:
(http://i.imgur.com/2yoRyS2.jpg)
Then I did it as a single smoothing group for the whole object with a high value of 80 degrees (I think it was 75 before) to get everything rounded that I wanted to be with this low-poly model. Here it is with the scanline renderer, take notice of the trailer side as it wraps around to the trailer front, a continuous curve there. Also on the trailer overhang as it wraps underneath, again a smooth curve.
(http://i.imgur.com/PqEVRdP.jpg)
And with no other changes, now back to Corona. Notice first that Corona's smoothing interpretation is quite different from the scanline/QS (didn't try mr). Second, HDRI environment map reflection artifact still there. Can someone explain the smoothing differences? And also what's next suggestion for figuring out why Corona is doing this? First guy said hey that's just raytracing, other people seem to think this isn't correct Corona behavior.
(http://i.imgur.com/7cr1Cjq.jpg)
-
I would rather not compare it to scanline but to real raytracer, like Mental Ray or Vray.
In my experience low poly mesh really can cause various problems with reflections / refractions despite the fact that diffuse shading looks butter smooth. I can show you model that has much denser topology than your example and yet it breaks reflections quite noticeably and it is rendered in Mental Ray.
My suggestion: render it with MR under similar conditions and see if results will differ much.
-
My suggestion: render it with MR under similar conditions and see if results will differ much.
Continuing on our magical mystery tour of rendering engines, after of course replacing the lights and materials again, the answer is... yes but differently and not nearly so visibly. Look at the far side of the cab, you see MR go from one mode of reflection to another, with the second one appearing to be faceted. And I have no idea what is going on with the cab top on the far side of the trailer overhang, that's not making any visual sense to me. And the trailer front texture is misaligned with this skin, grumble.
So Romullus, since you are clearly ahead on points in this thread, what is going on there with MR shifting from what looks relatively accurate in terms of reflection to something faceted? If it's raytracing, why would only the falloff side be getting a pure raytrace?
(http://i.imgur.com/AD7Lkca.jpg)
Oh, and also if anyone can tell me how I managed to do this, let me know - I haven't used MR that much. At the time I just said "holy shit" and deleted that material and started over. I started with a glossy template arch & design MR material, and then wired MR physical sky to the reflection color channel, and then put the HDR image I started with as an environment map in the background image slot of the MR sky. I took a screenshot because it was amusing but then deleted the material and started over with nothing in that crazy-time reflection color channel. Now I wish I had stuck with it for a minute, that could be fun to use at some point.
Only light in the scene is a daylight system with MR Sun and MR Sky.
(http://i.imgur.com/rVSqKK6.jpg)
-
hmmm, I am probably doing something wrong, can you help? ( no irony, I am serious! :D )
Can we go back to whatever Maru is doing? If it's wrong it's working out right. Your model is as low-poly as mine, but all your reflections are lining up.
-
Ok Rommullus, not sure we can say it's just raytracing. Since I just found a way around it for MR, at least.
I decided to try HDRI lighting, but instead of putting it in the environment map slot I created a standard skylight and then used the HDRI in the skylight map slot. For some reason MR didn't seem to want to render reflections from that, so I set a .jpg version in the environment map slot, and that reflects fine. I'll put two versions, one more or less realistic after fiddling, another with the material reflectance cranked up to like .8. I also had to use blur and blur offset in the .jpg's mapping parameters, they were unrealistically crisp.
This is the "nice" version. Even unenhanced, you can see no faceting in the reflections of the .jpg environment map.
(http://i.imgur.com/ZnwrHBY.jpg)
This is with reflectance cranked up to .9. Very weird how MR is handling the transition from ground plane reflection to environment reflection, and the sky reflection is more saturated than my eye says is right. Also very weird reflection on rear of trailer.
(http://i.imgur.com/pIvKr87.jpg)
In both cases, no faceting on the reflection, so it can't be just raytracing. Actually now I'm wondering if it's HDRI images in the environment slot being reflected, not a raytracing issue. Going to try that now. In any case, if I can get around it on MR that suggests we should be able to find a way around it on Corona. Maru, how much you want to bet you weren't using an HDRI image in the environment slot?
-
Can you share this model? Just a mesh without textures.
-
Can you share this model? Just a mesh without textures.
Attached. I just spent some more time with it in Corona and can't find a way around that faceting despite my success in MR. It facets everything, HDRI or no HDRI, including solid objects in the scene. I have no idea how Maru is getting the smooth reflections he's getting.
-
Hehe.
(http://s10.postimg.org/wcves3b09/T1_raw.jpg)
Yours geometry
(http://s10.postimg.org/kc9yrd3ll/T2_raw.jpg)
My geometry
(http://s21.postimg.org/brgbf5j47/T3_retopology.jpg)
New geometry with Toorbosmooth
And the corrected file
I just aded a couple of connections on the bad looking edges
I will make a coupe more tests may be something more decent will come around
UPD _ updated file in attacment
-
That was one of the original questions, whether Corona simply doesn't like low poly counts, I've always been sure I could tessellate or subdivide the mesh and get perfectly fine results. The question has been whether there's any way to get rid of the faceting on very low poly models without raising the poly count. And actually it's a hardly critical question, I'm just curious at this point and it's providing a good excuse to poke around with MR and Corona and learn something.
-
That was one of the original questions, whether Corona simply doesn't like low poly counts, I've always been sure I could tessellate or subdivide the mesh and get perfectly fine results. The question has been whether there's any way to get rid of the faceting on very low poly models without raising the poly count. And actually it's a hardly critical question, I'm just curious at this point and it's providing a good excuse to poke around with MR and Corona and learn something.
It is not about corona. It is about 3ds max and smoothing groups.
(http://s15.postimg.org/64r16h6wb/T5_LP.jpg)
-
How can it not be about Corona when you see MR images above where this faceting artifact doesn't exist and there are smooth environment reflections all the way across the mesh with no changes to geometry or smoothing groups?
Looking at your file, you changed it to editable poly and broke it into separate objects. These game assets can't be either of those and the question at the beginning was can I get better results than this faceting WITHOUT changing anything about the model because if I did have to make changes, it would defeat the purpose, which was can I do nice renders of a bunch of game assets without doing surgery to each of them.
I can think of several ways to fix this problem if I'm willing to make changes to the geometry. Also, clearly it's an issue that's only going to be noticeable on a low-poly editable mesh, which makes it an edge case to begin with.
As I said, I'm mostly just curious to see if there's a way around that problem with no changes to the model just because I'm curious at this point and it's a good excuse to poke at Corona, trying to solve a problem is a good and quick way to play with everything and see how it works.
Anyway, Romullus and Maru, unless you're also curious about it, I don't think there's enough value to keep pursuing it - very few people are ever going to run into this issue. Corona is a high quality rendering engine for high quality models, not insanely low poly editable meshes designed for very old game engines.
-
Ok... Sorry but I just can't imagine how this could work in any renderer.
Also, are you getting crashes when drag-and-dropping this file into viewport? Tried with max15 and 16.
-
Yes, converting to editable poly breaks normals immediately. That's no-go of course. Checked your file and it looks that MR treats those normals a lot better than Corona. Curious how Vray would stand?
-
Ok... Sorry but I just can't imagine how this could work in any renderer.
Also, are you getting crashes when drag-and-dropping this file into viewport? Tried with max15 and 16.
I just tried it, no crash. I always work with a single maxed viewport, if that makes a difference.
And I did get a decent result in MR. Overall I like Corona's version much better, but I can probably fiddle with the MR scene and then import the asset meshes and switch out a single bitmap for each skin. But it did give me an opportunity to really fiddle around with Corona, and I like it. Strongly considering getting a license.
Yes, converting to editable poly breaks normals immediately. That's no-go of course. Checked your file and it looks that MR treats those normals a lot better than Corona. Curious how Vray would stand?
I don't know, don't have VRay :) Beginning to think Romullus is using this as a survey of current low-poly rendering capability ;-) Ok, what about iRay? Ok, maybe we should try it in Final Render. OK, but what does it look like in Octane? I did grab the Octane demo but the output limitation is so small and the images so obscured with watermarking that I'm not sure what the point is. Corona's demo method is the way for companies to go.