Shadow Feathering in Dynamic Lighting

While it could increase processor power, it would be nice if the edges of sight and shadow could be feathered (blurred). The harsh lines look really weird, especially when you have a small object casting a huge long thin shadow or bar of no-sight. Example: This could be a simple static radial blur from corners, or the radius of blur could be defined by the light source or sight token width, depending on how processor intensive you wish.

That would be awesome. Might help if you can provide a resource for html5 that can do this already.

I think this would actually be really helpful, especially if it took into consideration what I like to call "light spill". Basically, the light that would flow into a room but not be direct light. Essentially the way that they currently have the light where you can set the full strength radius and then the dim radius. It would be nice if the dim radius "spilled" through open doors so that the dim light effect filled the room up to it's radius away from the direct light. So instead of this: It would be more like this:

This is a mock up I made to show a similar thing, maybe has less impact on what is made visible while keeping the aesthetic idea. <a href="https://dl.dropboxusercontent.com/u/5306136/lighting%20suggestion.gif" rel="nofollow">https://dl.dropboxusercontent.com/u/5306136/lighting%20suggestion.gif</a>

+1 to bring this back from the dead. I'd still like this.&nbsp;

+1

+1 This would greatly improve the look of the game.&nbsp;

+1 Absolutely. This would make a big difference when it comes to atmosphere.

I was just going to make a post asking about adding a "dim" setting for the edges of light (say a flashlight with a 90 degree arc) instead of just being sharp light/dark (even if a dim range is set for the distance) but this seems like it would accomplish the same thing. :)

+1 This is what I'm looking for.

I believe the problem stems from the fact that the roll20 calculates lighting from only the center of the token. If there were multiple points of origin inside the token, then we would see the shadow feathering effect we want. I illustrated it below: This is what we currently have, just a single origin of light from the center of the token. And this is what we could have, if we had 9 light origins spread out instead of just one in the center.

Wouldn't this essentially let players see further around corners than they normally could, though? I use dynamic lighting as a sort of fog of war, so it would make a difference there.

Well, the solution with multiple origins (say 9 instead of 1 currently) seems to give sound results, even with respect to normal vision around corners.

How were you able to implement 9 origins? Through the API?

The method of have 9 dim light origins following character tokens around could be done with the API. For true shadow feathering, the VTT's WebGL shaders would need to be modified to support it.

Right. Hopefully, this thread gets seen by one of the roll20 guys.

+1

Any chance to see that coming, with the new canvas expert?

Would still like to have.

I suppose this will never happen, then?

It doesn't have the 200 votes to give it the chance at direct Roll20 comment, but it may still come about.&nbsp; We'll have to see what the new system looks like.&nbsp;

+1

This is the single biggest thing that bugs me about Roll20. The dynamic lighting is so harsh, it ruins the mood I'm trying to set for my games.

+1

It has more to do with Roll20 using vector lines instead of raster, hence the sharp contrasts.

With UDL, this is now much more likely to be implemented.

Found this on my first dynamic lighting experiment, feels very much like a missing feature. Would to to see.&nbsp;&nbsp;

+1 it's not so much to "see around corner" as much as "my brain can tell there is the rest of a car or body hidden in that shadow". It could be alleviated by just darkening the tokens in shadows a bit less, so we can tell there is a form but not what it is. Our fantastic human brains will do the rest.

Love to have this. It would add a realistic style. Although it shouldn't use up to much performance. Maybe configurable for each player to take care of people with weak hardware. +1

+1

+1

+1 We were talking about how cool this would be, even though my PC is a potato &lt;3

Hello all. Thanks for contributing to the suggestion. One question that I have is about where the feathering starts and stops. For many game systems, the visibility is determined with a hard line. At 39 feet and 11 inches a character can see everything, but at 40 feet and 1 inch the is pure dark. We would likely carry over the style that Dim Light currently has, which is to feather before the end of the light distance. The trade off for the aesthetic would be a small reduction in what ends up as visible to the Player.

Based on player feedback from the existing dim lighting, please err on the side of more player visibility rather than less. If dim light goes to 20 feet, make sure the player can usefully see 20 feet; feathering a couple feet after that would be good, in my opinion.

For my own opinion, I would feather to the edge of player visibility or just barely past it. - As a player I am ok with the edges of my vision being darker as long as the last 5ft square is still somewhat visible. - As a GM I would not want it to be visible past the player vision at all.

Really, just making a small 5mm blend would go a far way. The lines simply always look rough and unrealistic. The suggestion by Kenton seems reasonable. I doubt anyone would mind a slight drop in visibility if it would mean it is far more pleasing on the eye.

+1. "Can I see that or not", "I can't understand what this room is supposed to be", "is that a column or a long wall" are the most common complaints in our games.

+1

+1

+1&nbsp; (and sorry for the long post) I think to have a proper discussion we need to clarify which "lines" we're talking about, because some users are talking about one type, and some another. First, when there's bright light coming from several different light sources on the walls of a huge room, and the light from none of them reaches the center of the room, the result is this big black "pillar" in the center of the room which doesn't actually exist.&nbsp; The players however see it as a "pillar" because it more-or-less looks identical to a dynamic light barrier.&nbsp; That's the main problem I think, telling a wall apart from from a light edge. This is fixable by adding a "low light" radius to your light source, or just replacing the bright light with low-light altogether.&nbsp; So in this case, we're talking about the "light radius" edge.&nbsp; (fyi: I've noticed that occasionally, even when you add low-light, the sharp radius remains... but I think its just something to do with refreshing the screen because it fixes itself if you move the light source token around a bit to update everything) The second is a&nbsp;"line-of-sight" edge. There's an straight black line formed when a token moves behind a corner or wall or other dynamic light barrier.&nbsp; This is what OP was talking about. I think those should stay relatively sharp because they represent a solid object blocking your view... the fact that we see them on screen as a straight black line is...maybe a bit ugly...but it can't be helped because we're basically forcing 1st person rules of sight onto a 3rd-person overhead view of an area. Those straight black lines are just the result of the disparity in perspective and should be kept mostly the same.&nbsp; A bit of feathering may be in order, but not too much. Secondary would be if the feathering should start before or after the edge of the light, and it looks like people are divided on this.&nbsp; My vote is for feathering before the edge, since you can still kind of "see" what is in the feathered area to some degree.&nbsp; In my mind, the vision limits set by game rules on characters represent the absolute MAXIMUM distance that the character could possibly see, which means that an object just on the edge would be visible... but just barely.&nbsp; The character would be straining their eyes, squinting, that sort of thing, but would still recognize that there was something there.&nbsp; I don't see why the light emitted from objects should work any differently-- the number just sets the absolute MAXIMUM that could be lit up, that doesn't necessarily mean that all of it is lit up WELL.&nbsp; Just my opinion. Last, if you wanted to get really crazy and realistic, you might look into some rudimentary form of ray tracing for bounced light, using dynamic light barriers as the surfaces to bounce excess light off of...&nbsp; this would be more like the real world where light does indeed strike a surface and illuminate things around it (even around corners).&nbsp; This would have to be an optional setting though because I'm sure it would be pretty resource heavy.&nbsp; TL;DR I think the edge formed where a light source ends should be well-feathered (on the inside), but the edges created by line-of-sight should not.&nbsp; Also, implementing basic ray-tracing for light sources as an option would be cool.

Jen said: Wouldn't this essentially let players see further around corners than they normally could, though? I use dynamic lighting as a sort of fog of war, so it would make a difference there. I tend to agree. The hard line is aesthetically displeasing but it does represent that which you cannot see. However, your sight itself is not a hard line. Since you could be anywhere in the square you are occupying you should have the ability to see a little around corners.&nbsp; Right now everything is calculated from the middle of your square. Let us now suppose that light/vision were not calculated in a binary visible/not-visible fashion. Currently, a line is drawn from the center of a token square to the edge of a light-blocking obstruction. Let us instead draw two lines. One line represent 100% unobscured vision and the other 0%. That means that the hard line we currently have would represent the halfway point between totally and unobscured.

The second image shows a concept where the token is fully visible but the obscured portion is ghosted.

Akakemushi &nbsp;said: +1&nbsp; (and sorry for the long post) I think to have a proper discussion we need to clarify which "lines" we're talking about, because some users are talking about one type, and some another. First, when there's bright light coming from several different light sources on the walls of a huge room, and the light from&nbsp; none &nbsp;of them reaches the center of the room, the result is this big black "pillar" in the center of the room which doesn't actually exist.&nbsp; The players however see it as a "pillar" because it more-or-less looks identical to a dynamic light barrier.&nbsp; That's the main problem I think, telling a wall apart from from a light edge. [...] The second is a&nbsp;"line-of-sight" edge. There's an straight black line formed when a token moves behind a corner or wall or other dynamic light barrier.&nbsp; This is what OP was talking about. I think those should stay relatively sharp because they represent a solid object blocking your view... the fact that we see them on screen as a straight black line is...maybe a bit ugly...but it can't be helped because we're basically forcing&nbsp; 1st person rules of sight &nbsp;onto a&nbsp; 3rd-person overhead view &nbsp;of an area. &nbsp;Those &nbsp;straight black lines are just the result of the disparity in perspective and should be kept mostly the same.&nbsp; A bit of feathering may be in order, but not&nbsp; too &nbsp;much. There's a third edge, which is what I think the OP is talking about: the edge caused by the light from a light source being blocked. Consider this situation: Nissia (the monochrome lady) is in a room, and carries a light source. Yasha is in a corridor right outside, and does not. Yasha's vision is sharply limited by the light being blocked by the corners of the door, and that doesn't feel right. Light will feather around corners, bounce of walls, and such, to provide some amount of dim light even when there is no direct line of sight to the light source. In other words, I don't have a problem with Yasha not seeing the parts marked off by the aqua dots, because walls block LOS. I would appreciate the vision being somewhat more permissive (perhaps by having vision calculated from each corner of Yasha's square or something?), but that's a minor thing. I do have a problem with Yasha not seeing the area blocked off by the green dots, because the light from Nissia's everburning torch should illuminate a bit beyond the direct LOS to the torch.

This change is on the Dev Server, and Pro users have early access.

I came onto this thread to suggest this, then found this! I have tried to look on the dev server, to provide feedback, but the lighting looks identical to the production instance.

This was released in late July 2021.&nbsp;