Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - pokoy

Pages: 1 2 [3] 4 5 ... 124
31
[Max] I need help! / Re: Windows reflections exterior
« on: 2024-01-18, 13:02:11 »
Sorry for the confusion, the map slot names seem to have changed in more recent vserions, I'm still on v9, thanks for helping out Avi.

32
[Max] I need help! / Re: Windows reflections exterior
« on: 2024-01-18, 12:20:07 »
In the material properties, there's a long list of map slots available on the bottom, that's where you'll find these two slots.

33
[Max] I need help! / Re: Windows reflections exterior
« on: 2024-01-18, 11:10:29 »
Yes, you can use the HDRI you want to be reflected in the material's 'Reflect BG Override' slot that is used on the window objects. You might have to use the same map in 'Refract BG override' too in some situations, just try whatever works for you.

34
[Max] General Discussion / Re: Price Increase
« on: 2024-01-11, 17:06:02 »
Just had an email from Chaos informing of another 10% price increase to an annual Corona Premium Subscription, this coming after last years 100%+ increase.  Thoughts people?
Just read the email. If I remember correctly the last price increase came with a comment along the lines of "this will allow us to keep the price stable for the next years".

35
[C4D] I need help! / Re: Proper color mixing
« on: 2024-01-05, 17:29:17 »
The amount of complete or partial misinformation in this thread hurts my tiny brain. Don't want to come across as a smartass but we just can't let this pass, can we? :D
@OP - this is possible, with some limitations which I'll outline below.

First - what's the difference between the additive and subtractive mode?

Think of it that way: what's your starting condition? If your starting condition is, for example, a black screen, then you'll only be able to display colors by emitting light (adding energy). By using red, green and blue as your primary colors these ADD up in varying intensities to produce a certain color. The primary colors used in this model are R, G, B, hence the RGB model name.

If your starting condition is a white sheet of paper, then your starting condition is (ideally) white - from which you now will need to subtract colors by using either a dye or filters. This however needs different primaries than with the additive model - cyan, magenta and yellow. Each of these dyes/filters will now SUBTRACT (absorb) energy with varying densities, producing a certain color. The primary colors used in this model are C, M, Y, hence the CMY model name.

We'll now use these color definitions throughout the examples:

Red (R) = RGB 1/0/0
Green (G) = RGB 0/1/0
Blue (B) = RGB 0/0/1

Cyan (C) = RGB 0/1/1
Magenta (M) = RGB 1/0/1
Yellow (Y) = RGB 1/1/0

Here's a render for the additive RGB model, where 3 lights use red, green and blue filters and produce a white color where their light 'cones' overlap on a white plane:



The setup in Max:



Note how the secondary colors produced by red, green and blue filters are cyan, magenta and yellow > the primaries used in the subtractive model. White is the result where they overlap.

Now for the subtractive model, here's a render of that one. White light passes through cyan, magenta, yellow filters (they're offset a bit to avoid coplanar faces) and finally reach a white plane:



The setup in Max:



Note how the secondary colors produced by cyan, magenta and yellow filters produce red, green and blue > the primaries used in the additive model. Now the funny thing is that where they overlap the color should - in theory - be black, but it isn't. It's a brown muddy color instead, and this is also what happens in reality. Corona produces a slightly lighter color here than what happens in reality but it's still pretty accurate. This is a limitation of the CMY model and also the reason why in printing there's an additional black color, named 'K', and the resulting color model is CMYK and is used in printing (offset, inkjet, laser printers all use these dyes/colors). This is the limitation of the CMY model which I've mentioned at the beginning of the post.

Now for your orange color. Let's fire up PS (or any other app supporting CMYK color mixing) and open the color mixer. A typical orange would be CMYK 0/60/100/0, which would be RGB 239/124/0. This means for CMY you only need the magenta and yellow component, in RGB only red and green.

Let's produce and orange color for the additive model:
Set up two materials for the filters to be placed in front of the lights, so that for the red filter you define a RGB 239/0/0 color, for the green filter you define a RGB 0/124/0. Place each of these in front of a white light respectively, the color produced on the white plane should now be orange. (Note - technically, you should be able to just use these colors on two lights with the same intensities, they *should* mix up to orange... for some reason they don't and I'm not sure white. The result is yellow and only turns orange if I really decrease the exposure a lot...)

Result:




For the subtractive model the setup is a bit different:
For the magenta filter, we need it at 60 perc density > use a mix node with a white color as the base and a magenta layer on top - now set the opacity of the magenta color to 60. White is the base because white is the starting condition for the subtractive model, and adding a 100% magenta color with an opacity of 60 results in 60% magenta as needed by the orange color we defined above.
For the yellow filter, use the yellow color defined above, it needs 100% density so we don't need to change it.

Result:



For the filter material, you can use the filter colors in thin absorption mode, however this will not account for object thickness which might not be looking too good if your object's aren't just a thin plane. For anything more realistic, use the filter colors in the volumetric absorption channel (and scattering if needed) of the material and set absorption distance to a meaningful value based on the thickness of the geometry the light is passing through.

If you want to produce other CMY colors, here's a handy online converter to convert from RGB to CMY - http://colormine.org/convert/rgb-to-cmy
The CMY results are in a 0-1 range. If a color has 3 components, you will need all 3 filters - cyan, magenta and yellow. This is different from the orange tone we produced, this one needed only 2.

The theory behind all that is much more complicated. Visible light is not just a sum of R, G and B components, that's just a model that our body/brain happens to use to see and represent colors. Both the RGB and CMY model are just *models* but since they're part of the same spectrum each model can be used to represent the other to some degree - we've used what's technically a strictly additive RGB model (emitting screen with software that works with RGB primaries) to somewhat accurately represent what happens in a CMY model. Since each of the models produces the other model when you mix its primary colors, they are theoretically interchangeable. However, in reality each color represents only a narrow band within the visible spectrum, leaving some inaccuracy on its tails - that's why it's really hard to mix these three colors in reality and get a clean saturated result.

That's why you won't be able to reproduce all colors in CMY. Darker colors will always tend to get a brownish muddy tint and when they overlap, they still won't absorb light completely. There's a ton of technical stuff on the internet for further reading if you're interested in why that is.

EDIT
Not sure why the forum won't display images hosted on my site, giving up now. You can open them with a right click in a new tab/window, I also attached them to the post.

36
Yeah wondered about this too. The built-in volume object from Arnold is much faster (and supports the new volume rendering in viewports in newer max versions, too...)

37
[Max] Bug Reporting / Re: NaN pixels on CoronaSun disc
« on: 2023-12-21, 11:19:27 »
Afaik this can happen with certain filter types and/or sharpening, for example in Terragen this happens with catmull rom used as the image reconstruction filter. Of course I can be totally wrong but it's worth a try to go with other filtering types and disabling sharpening if it's being used.

38
[Max] I need help! / Re: Slow DR Parsing etc
« on: 2023-12-15, 13:24:12 »
This seems to be the right place so I'll add my issue which sounds very similar.
In my case I'm rendering through backburner on a 128GB node. During preprocessing RAM consumption shoots up to 125GB (this is already with some elements disabled to fit it in RAM), only to decrease to 52GB - 55GB within minutes once the render process begins.

I can't render the same scene on a 64GB node because of the RAM spike, even though it looks like Corona actually needs way less than that.
I'd happily trade less RAM consumption for longer preprocessing/parsing times if it means I don't have to split a render into 2-4 jobs only to get my elements/masks.

This is with Corona 9hf1 (I know I'm waaay behind) but this issue has been there for quite some time already.

39
Thank you Tom. In my case (tested on stills only so far), splotches become visible much more clearly when re-rendering the same image a few times, the cache seems to 'dance' a bit each time. Still, when comparing it to UHD, UHD looked worse most of the time.

40
As a note, it is safe to use until it isn't ;) That is, it works and is better than the UHD Cache in most cases, but in some niche cases it can cause artefacts. This is why it is not the default at the moment - but that does mean you can enable it, and if you have no artefacts you can safely use it. You could make it your own personal default, and have returning to UHD Cache as the first step if you run into a scene with artefacts. We don't want to make something default in Corona itself until known issues are resolved, though.
Something I'd be very interested in is learning what these artifacts are that you mention here. I've asked in a few places and never got a clear answer :D

41
For me, it's consistently better than UHD in difficult interior tests, it's less splotchy in general. but it isn't super stable, renderings the same scene 3 times will result in minor differences in small areas of the render, and sometimes it struggles with some areas still being noisy (for example in areas where PT takes a long time to resolve to noise-free, too). Still better overall looking than UHD but there is zero documentation or official tests/examples so it's a bit of a black box.

I'd also love to know where it's going.

42
I ran into this issue a while ago when working with sinks. The reflection colors reflect upon themselves. The solution is to use a rayswitch material with desaturated reflections. Attaching a simple setup with ComplexFresnel.
Workable workaround definitely, thanks for the suggestion - updated example render attached.

Though I still have my doubts as the default behavior leads to too saturated reflections if no workaround is used and has probably implications in general, maybe even for refractions? Plus, depending on the scene and material count it might be cumbersome to set all the material overrides needed. Maybe a general solution or a parameter to desaturate the Mtl's reflected color would be good to have... Still, a usable workaround for now.

43
 I guess OP has a point, colored reflections *are* heavily saturated and they are less pronounced with increasing roughness of the material but when glossy, are pretty much impossible to get rid of. With the gold mtl preset, the coating with greenish absorption helps a bit but not really.

Tried to replicate the example posted above, and while it's not a 100% match, you can clearly see the strong red tint in reflections which is not as pronounced in the example. Image attached.

44
IIRC there was a thread/post here recently reporting something similar, only with mask visibility through refraction, where it would be missing for a couple of frames but ONLY on legacy materials, Corona Physical would render fine. Maybe a thing to look out for... sorry if this is misleading, just thought it might be related.

45
First off, if you expect this to render in 2 minutes (seems to be the time after you stopped the render) - that's probably not going to be enough.

I'd set everything back to defaults as a first step. For example, setting 'Clamp Highlights' to 500 will already make a scene with complex lighting like yours slower to render than needed.

If you are using self-illumination on the materials that are supposed to generate ligh, switch to LightMtl wherever possible or make sure to use only LightMtl for all bright visible materials.
Self-Illumination is calculated through GI so it takes way more passes to get rid of noise compared to LightMtl.

You seem to have many rough-reflective/semi-glossy materials/metals. This will mostly render slower and produce some fireflies. Don't overuse or dial it down where possible.

Also, using 4K cache instead of UHD is a better choice most of the time, try this one.

Don't have much advice other than that, but I'd definitely allow it to reach more passes.

EDIT - I realized that Highlight Clamping set to 500 isn't much of a problem (the default is '0' which means no clamping at all), and using lower values can actually help to prevent fireflies so for anyone new to Corona and reading this at a later stage, just dismiss what I wrote above, it's wrong.

Pages: 1 2 [3] 4 5 ... 124