Messages

91

[Max] Resolved Feature Requests / Re: Noise level readout in framebuffer to read more decimal places

« on: 2017-01-18, 21:04:38 »

Also, if you are using render regions, only the render region area(s) should apply to the noise indicator.

It would also be nice if the number of passes got reset when regions got disabled to what it was before any regions were defined (but this may or may not be useful to others), since it increases very rapidly when small regions are defined and becomes very misleading when they are removed (and the entire scene is being rendered).

92

[Max] Resolved Feature Requests / Re: Persistent render regions that gets saved with scene file

« on: 2017-01-18, 20:55:54 »

Speaking of regions, some usability related improvements:

1. Perhaps tangentially related, but the ability to create/edit render regions numerically would be nice also:

e.g., 0,0-100,100 (Start at 0,0 for 100 pixel width, 100 pixel height)

2. Also, maybe allow them to be saved/restored to/from a config or XML file, so that you can save, hand edit, and later load a set of region definitions for areas of interest in the scene you are trying to tweak at any point in time. I think this would be a big time saver.

3. Finally, the ability to toggle on/off the visibility of the red region indicator rectangles and X would be a very big thing. For small regions, the region marking rectangle gets in the way of visibility sometimes.

93

[Max] I need help! / Re: What are these jagged edges? Anti aliasing problem?

« on: 2017-01-17, 20:25:40 »

So, the final advice in this video is to use highlight clamping to reduce "the jaggies." My question is: Is it not preferable to do highlight clamping in post processing (e.g., Photoshop) or is there a disadvantage to doing it this way (i.e., does Corona have more information available to it during the render that it can somehow use to produce a better clamped scene)?

There should be no difference, so no disadvantage either.

There is definitely an advantage to doing this in post-processing outside of Corona - namely that it can be done selectively (via masks to include only certain parts of the image), with more granularity (via curves or whatever to do the clamping/reduction), and non-destructively (to the original rendered EXR HDRI image).

94

[Max] I need help! / Re: What are these jagged edges? Anti aliasing problem?

« on: 2017-01-17, 05:20:59 »

The advice in this video may help...
[video link]
(Also, since the video was made, it's also possible to use a small amount of bloom & glare to smooth jagged edges).

So, the final advice in this video is to use highlight clamping to reduce "the jaggies." My question is: Is it not preferable to do highlight clamping in post processing (e.g., Photoshop) or is there a disadvantage to doing it this way (i.e., does Corona have more information available to it during the render that it can somehow use to produce a better clamped scene)?

Because, if there is no disadvantage to highlight clamping in post-proc, that seems like the best solution since you defer the choice of whether to do it or not to a later time and you can do so in a lossless and selective manner in post-processing. If you highlight clamp during the render itself, the clamped highlight information is irrecoverably lost.

95

[Max] I need help! / Re: Does System/Image Filter/Type do anything in 1.5 or is it completely inactive?

« on: 2017-01-16, 22:09:15 »

Image filtering definitely is working, but don't expect that it will help you with pixels that has value >1. Adjust exposure and or highlight compression until pixels are <1 and then do your test.

OK, I tried what you said and once I got the output exposure levels to values less than one, the filter types did in fact have an effect on anti-aliasing. Here is the order that I came up with from least AA to most AA:

(None ?? Blackmann Harris) < Symmetric Tent < Hann < Tent < Parabolic < Box

This is a pictorial diagram that I created to help myself and others make correct future choices from the filter types (since they seem to be undocumented - I wish a diagram like this was in the official Corona manual). The image shown is of an octagonal light source (with minimal rounding of corners) reflecting in the center of a highly-reflective large sphere (there is very little to no spherical distortion of the light source present). I made sure to keep all output RGB values < 1.0 (approx. 0.97 max or so). You can click on the image to expand it and scroll it to see the detail along the edges of all modes:



In my opinion, Symmetric Tent is best for minimal AA (but still a reasonable amount) and Box is best for maximal AA, at the expense of some loss in high-contrast sharpness around edges. Blackmann-Harris was skipped, because it is overly biased towards anti-aliasing horizontal lines and doing next to nothing to vertical lines, to be useful for all scenes. It may be useful for those scenes that have large amounts of aliasing around horizontal edges, however (e.g., sky/land/horizon transitions).

96

[Max] I need help! / System/Image Filter/Type, how does it affect the render?

« on: 2017-01-16, 20:39:19 »

I tried setting it to None, Tent, Hann, and Blackmann-Harris and rendering very high contrast thick diagonal reflections of lights to see what these various anti-aliasing settings do. I get exactly the same results, even after hundreds of passes, as if this setting affects absolutely nothing. Highlight clamping was set to 4, but highlights were still blown out to show very bright and aliased highlights on a dark surface. Anti-aliasing was clearly visible in the resultant images which were rendered at 3840x2160 BTW, especially at 2x magnification of the pixels, but it was constant and independent of the filter type setting. It seems that the AA scheme chosen is fixed (to what?) and cannot be changed/affected by the Filter/Type setting.

Is this one of those completely useless legacy settings that did something at one point, but now does not effect the render at all or does this setting actually do something and if so, what?

97

[Max] Resolved Feature Requests / Re: Noise level readout in framebuffer to read more decimal places

« on: 2017-01-13, 23:25:51 »

It should always be displayed with at least three decimal places, if you ask me.

98

[Max] General Discussion / Re: Corona Interactive vs Vray RT (CPU)

« on: 2017-01-09, 18:10:37 »

So what is the state of caustics, reflective and refractive in the latest version of Corona for 3ds max? Are they available in the progressive production renderer or not?

99

[C4D] General Discussion / Re: Is there a material converter VRay -> Corona for the C4D version?

« on: 2016-12-30, 21:46:44 »

usually i go the long way by creating a new corona material and copy/pasting every shader in the correct slot.. on bump/reflection/refraction you usually have to do quite some adjustments but at least it is working in the end..

I've considered this onerous, non-exact, and error prone process and was hoping that there was some automated way.

Btw, how good is Corona with matching native C4D materials without any sort of conversion or adjustments?

100

[C4D] General Discussion / Re: Corona for C4D A6 (final) released!

« on: 2016-12-30, 18:00:13 »

Where on this web site can we download a material pack that is compatible with this (C4D) version and not the 3dsmax version?

The entire resources section seems to be 3dsmax only:

https://corona-renderer.com/resources/materials/

How do we import these .mat files into the C4D version of Corona?

101

[C4D] General Discussion / Getting 3dsmax Corona materials ported over to C4D version

« on: 2016-12-30, 17:57:13 »

I see many 3dsmax compatible corona materials available on this site for download. How do we get these ported/converted over to work with the C4D version of the Corona Renderer?

102

[C4D] General Discussion / Is there a material converter VRay -> Corona for the C4D version?

« on: 2016-12-30, 17:14:17 »

I am wondering if there is a script or some other type of converter that allows for the conversion of VRay materials to Corona materials (from for example .mat files), but for C4D rather than 3dsmax? Or, perhaps let me ask the question this way: How are you converting/using VRay materials with Corona in C4D?

103

[C4D] Resolved Bugs / Issues with A6 final (AA related)

« on: 2016-12-30, 17:00:27 »

I am test driving the A6 final and am having some issues specifically related to aliasing (and a lack of good (and/or documented) anti-aliasing options):

1. In the documentation and on the web site there is mention of an internal resolution ("Internal res.") setting which by default is set to 1 or 2, depending on which web page on this site you believe. This setting does not seem to be viewable or changeable in the C4D version. It is not listed in the Image Filter settings at all. I see the image filter type, width, etc..., but the Internal Res. setting is completely absent from the available settings. Was it purposely removed and if so what has replaced it?

Update:

I just found the answer to this at: https://coronarenderer.freshdesk.com/support/solutions/articles/5000515615

The setting has been removed, leaving only highlight compression and clamping as fallbacks. I already tried highlight compression and this severely reduces/distorts the highlights and speculars of the image and greatly reduces contrast. I will take a look at highlight clamping, but based on the sample images at https://coronarenderer.freshdesk.com/support/solutions/articles/12000006462 it seems to either: 1) Not do much at higher settings 2) Reduce brightness and contrast to unacceptable levels, just like highlight compression, at lower levels.

So, what does all of this mean? If we want to get the former effect of internal resolution, we have to render at a higher resolution and then scale it down ourselves in post-processing - a major PITA, if you ask me. Also, Corona can do this much more intelligently in its engine. Instead of having the artist render the scene using a 2x resolution, resulting in four times as many pixels, to later scale it down (an effective method of manual aliasing reduction), the engine could potentially limit this process to only neighboring high contrast pixels, of which there most likely aren't very many. To do the same process manually in a reasonable amount of (rendering) time, one would have to create multiple renders: A baseline render at the desired final resolution (base-res) and several selective area (high-res) renders of problem areas with jaggies. These would then be scaled down to (base-res) in order to reduce and/or eliminate the jaggies and composited on the original (base-res) render. This of course would require quite a bit of manual labor.

2. With regard to the image filter type and width. By default, these are set to "Tent" and 2. This is producing heavy aliasing of some high contrast elements in my scene - fairly thin diagonal metal cylinders reflecting direct lighting along a portion of their curved area. Note that this is after hundreds of passes at a resolution of 1920x1080. Highlight clamping is at zero (I will try raising it) and compression is 1.0 (I tried increasing the highlight compression, but it just makes highlights and speculars in particular darker and in my opinion leaves the aliasing just as bad, if slightly less noticeable because of the lowered contrast) . I keep reading about a "Gaussian" filter type from other posts which is supposed to produce better anti-aliasing at the expense of some image sharpness, but this is not available from the filter type drop down in the A6 final. Also, I cannot find any documentation whatsoever about the various filter types (other than Tent or None, both of which produce very aliased results at a width of 2, anyways). Additionally, there does not appear to be any documentation whatsoever on how the filter width affects anti-aliasing. I would like to know what a good setting is to smooth out very high contrast direct reflecting diagonal highlights. A setting of 2 px doesn't cut it for my scenes.

I am loving the quality of the render overall, but these alias jaggies are just killing me. They remind me of ray tracing back in the 90s, but even then we had control over AA settings to combat this. It seems that Corona's idea of anti-aliasing is diametrically opposite to that of the native C4D physical renderer which by default takes a very heavy handed AA approach often blurring high-contrast details in the image as a result. I would like to achieve a middle of the road result - just enough AA to mostly remove the jaggies, without significantly blurring high-contrast details.