So Enlighten really turned out to be a waste of everyone's time.

First, I find it amazing how single devs can release much more efficient alternatives in all of the different lighting departments and approaches:
https://forum.unity3d.com/threads/r...bdivision-one-button-uv-atlas-packing.451970/
https://forum.unity3d.com/threads/r...aced-global-illumination.447861/#post-2905341
https://forum.unity3d.com/threads/segi-fully-dynamic-global-illumination.410310/

With that said, once upon a time, Unity was 'an indie friendly' engine, but not only it takes a huge team of professionals to make something out of Enlighten baking, but it turns out even with Microsoft's backing, the makers of Recore had a ton of lighting challenges and the end results suck and did not look current gen.

This entire Unity cycle was met with complaints about Enlighten ridiculous bake times and in the end, they were never solved.

I am just saying next time plan better your acquisitions. Because not only did Unity waste 2 years trying to build something functional out of the Enlighten codebase that won't be reusable, but now you will have to clean this mess out of the engine veins and every single corner that it infected (dependences).

Brigade3 is of course a step in the right direction. But Unreal is getting that too so it seems like it wasn't really Unity's initiative but OTOYs... while Enlighten feels as if Unity stepped in their door as beggars considering how much more of a better treatment they give Unreal.

 

Am I the only one who likes enlighten? LOL

Don't forget

https://blogs.unity3d.com/2016/09/28/in-development-progressive-lightmapper/

 

Yeah, there's a new lightmapper in development because the Enlighten one is so terrible that it was the only course of action.

 

Only really for bigish open levels. I've had no real issues with my one room game

 

That's only because you still haven't finished the whole room

 

Im guessing you didn't read this 9 part guide to Precomputed Realtime GI which shows how to optimise an example scene's bake time from 7.5 hours to 2.25 minutes?

https://unity3d.com/learn/tutorials...uction-precomputed-realtime-gi?playlist=17102

 

That's an oxymoron Andy.

 

Eh, its what its called in Unity. From the doc I shared:

"When working with Precomputed Realtime GI, a lighting precompute is the process of calculating the bounce of light around the static geometry within a Scene in the Unity Editor and storing this data for use at run time. This process reduces the number of lighting calculations that must be performed at run time, allowing realtime bounced lighting while maintaining interactive framerates."

 

I would say that enlighten NEEDING such guide (and 30 minute long manual scene preparation) is a problem.

 

I'd just like to add to OP's original post.

Link #1 looks decent but very similar to unity's planned progressive lightmapper, probably isn't as great as enlighten for GI, most likely is closer to what 'beast' was <- never used beast mind you.

Link #2 looks like an unbiased render, somewhat similar to cycles in blender. Not really getting the point of that in games other than static shots. Unless I'm otherwise mistaken.

Link #3 still unusable unless the guy can match the R&D of cryengine, plus requires a beefy graphics cards and windoze machine, not great for a mac user

So meh...

 

Clearly you boys haven't seen this

 

I disagree; any and every piece of complex software requires a tutorial to learn and understand it. Id argue that a 30 minute scene preparation guide for Realtime GI bake optimise times (and the wealth of knowledge you'd get with it) is pretty damn amazing!

Anyway, we are working on a Progressive Lightmapper for statically baked Lightmaps; so users can have the best of both worlds.

 

The other lightmapping system was better and faster

 

You are right. It makes me wonder why Unity is wasting their time on this:
https://blogs.unity3d.com/2016/09/28/in-development-progressive-lightmapper/

The links are about how Unity made a deal with Geomerics, the 'leading lighting solution' of the market and it turned out into this mess, while single developers are able to make a functional solution that with a staff a little bigger would be production ready so it should be an even smoother process with Unity staff size and R&D, but instead it is only now that they have addressed and admitted to their lighting issues.
You can thank ReCore for that.

Link #1 wasn't chosen for a particular reason other than showcasing a small dev able to get done a lightmapper in his spare time, while Unity even being fully aware of the current lighting issues [which some still ironically try to claim do not exist] is still taking all this time to address them.
Problem is that if you never used Beast, you won't know that it was a much better lightmapper that was only limited by Unity 32bit RAM limitations and to this day we have people not moving to Unity5 and still using 4 because Lighting in Unity5 is a no go. Unity5 would have been perfect. BUT ENLIGHTEN RUINED IT.
When I call it A MESS I am not trying to be annoying. They ruined MANY people plans and vision and hold them back for a whole cycle.
Beast with 64bit support would have been PERFECT. Zero complaints!
Their GI+baked and GI could have been a separated experimental builds or asset store asset or anything they wanted.
... But guess things were looking too good to be real.

Link #2 eventual goal is realtime. The same as Brigade3 from OTOY which Unity is also getting (go watch the presentation and the press release). And we will get there.

Link #3 point is that with Unity's staff size and R&D, you could have that exceptional looking GI baked into lightmaps for the whole production cycle and only go do the full baking at release, which would be insane. Lightmaps update in a second!

What Unity needs to solder into their skulls is that once they get their lighting, Unreal Engine is DEAD.
Of course I like competition, but I want people to get off their heads that mess of an engine is production ready.

Unity has always been production ready. It is always been the lighting and they know that it is the lighting hence most of the Unity5 showcase being around lighting.
It just didn't work! The main feature of Unity5 was its most disappointing one. It did not deliver.

/not-a-rant

 

Highly opinionated ranting. Enlighten is so much faster than the "solutions" you mention at run time when it matters. For example on console, mobile or any normal desktop that didn't cost many thousands.

Assuming you want "realtime" GI and not just baked lightmaps which Enlighten again isn't the problem, but the shadows and UV packing.

Enlighten isn't the problem here, but integrating it in an evolving engine is.

 

Enlighten works amazingly for Precomputed Realtime GI. Progressive Lightmapper will work amazingly for Baked Lightmaps. Different tools for different setups. You just need to research and learn how the systems work before utilizing them; same with any software, tool or craft.

Quoting that blog post you shared:

"It is not a replacement for realtime GI. It only works for baking in the Unity Editor, so you cannot bake from within your game. This rules out baking GI for procedural scenes in-game, or the ability to change the lighting setup for time of day and baking on load. We want to nail the baking workflow in the Editor before we take it further.

Also, it is not necessarily faster than baking with Enlighten. If you setup your scene well for Enlighten it can be very fast, so we cannot guarantee that all scenes bake faster with the Progressive Lightmapper. However, the time between starting a bake and getting some visual feedback will be significantly faster with the Progressive Lightmapper."

And about the two systems working together, from the comments section of that blog post:

Both lightmappers will work seamlessly with the new mixed lighting system. This will offer a mode using baked indirect + AO in conjunction with realtime lights and shadows or realtime lights with baked shadows.

 

You are only corroborating the point I made:

Unity brought Enlighten to the table as an improvement over Beast lightmapping but it ended up as a downgrade in the lightmapping department while bringing Precomputed Realtime GI as a new feature rather than a replacement one because there wasn't Realtime GI before Enlighten.

That means the ideal course of action would have been as it is being done now: to make both lightmapping and realtime GI separate. But it has taken you long to admit that. I put it in Enlighten never being conceived as a tool for baking lighting, not on Unity's end. Problem is that Unity wanted a convenient hybrid that simply does not exist and as shown by 'Also, it is not necessarily faster than baking with Enlighten. If you setup your scene well for Enlighten it can be very fast, so we cannot guarantee that all scenes bake faster with the Progressive Lightmapper.' still do not admit to it.

Either the Progressive Lightmapper is bad or they do not know the difference between having zero feedback if Enlighten has locked up or is still going

Efficient UV packing is the main challenge behind lightmapping, not the lighting equations sir

 

And I wouldn't agree with this point of view. If following the guide speeds things up by factor of 200 hundred it means that unity comes with horribly misconfigured default settings. There's really no other way to say this.

It is important to realize to that people do not need PERFECT results. The user should be able to press a button on a default and get a reasonable result in reasonable amount of time. Those who want more need those guides. 7.5 hours on default settings is not a reasonable amount of time.

In unreal engine lightmapper "just works" out of the box. If user wants to tweak, they can do, but Unreal's lightmass is faster on maximum settings than unity is on default.

Besides, WHY should I need to fine-tune positions of individual light probes manually? This stuff should be done automatically. Also, where is GPU acceleration and network swarm support? Where are translucent surfaces affecting light that passes through them? What about situation when people don't need gi and just want a lightmapper?

There was a very big thread about it, actually:
https://forum.unity3d.com/threads/u...ghting-a-big-step-backward-from-beast.343025/

 

Correct me if I'm wrong, but I'm pretty sure there's never been a proper easy solution to bringing outside lightmaps into Enlighten. Up until Unity 4 we could lightmap in a separate program then bring it into Unity. Not so with 5. That's the way it was when we first started working with 5, at least. So, we had to use Enlighten instead, and it took forever, and it cramped our style. If there's a decently easy way to import external light maps (as simple as it was with previous Unity versions), can someone point me in the right direction?

 

I believe "Bring beast back" thread mentioned some interface with possibility of connecting to external lightmapper, but if I remember correctly it was incredibly arcane. So the best idea for lightmaps would be probably either a legacy shader or something like plugging existing lightmaps into emissive channel (and possibly writing shaders for making this happen).

 

As far as I understand it, the big disparity between people happy with enlighten and people calling it trainwreck comes from the fact it works, for some use cases, pretty well.

AFAIK where it falls flat on its face is big outside scenes, ESPECIALLY terrains which has been a big problem even in Beast. Trying to fit the lightmap for a big 4x4km terrain into a single 2048x2048 pixel image is not going to end well. Even if 4096 pixel lightmaps would work as advertised, still nowhere near the resolution needed.
Sure, you can split the terrain into smaller chunks, but then we get into the many, many troubles with that unless you use some clever 3rd party assets. Even then, its kind of a hassle.


Fact is terrains are still the unloved step child in Unity, and I guess until the new Terrain System, which Unity started working on in 2014 and have gone disturbingly quiet about in the meantime, come to light, Unity will most probably not improve the lightmap baking for large terrains. Or the way to work with chunked terrains in stock unity.

 

Many mobile developers were making heavy use of Beast, had no interest at all in real-time GI when 5.x launched, and still don't today. For now, tech like this (and deferred shading) is still out of reach when targeting mid-spec devices and making actual games rather than tech demos.

One of the biggest failings of the move to Enlighten was the inability to import Unity 4 lightmaps+probes. That and the fact that lightprobes were broken in various ways for many months. Some mobile developers had big projects with loads of carefully-baked scenes, and found moving to Unity 5 extremely problematic, as the upgrade trashed the lighting completely.

And then it was made unnecessarily difficult to implement custom lightmapping solutions (...such as a 4.x->5.x import process) - the lighting data is locked away in LightingData.asset - a file that only Enlighten appears to be allowed to modify - unless I've missed something? (Any changes applied in editor scripts to lightmapping settings - such as per-object lightmap index/offset/scale or lightprobe coefficients can't be saved. Although you can use ugly hacks to store the data elsewhere, and reapply it on scene load...)

I hope that future lighting system updates won't be so painful. But I'm aware that directional specular lightmapping is being removed soon - and have recently worked on a project that relied on that...

 

I think people simply want TWO different things:
1. Fast lightmapper that produces alright results and can be used on mobile platforms (something similar to Unreal's Lightmass, pretty much)
2. Pretty -looking Global Illumination system that can be tweaked and may require high-end machine.

So the problem is that previous Beast was #1, and Enlgihten is #2. The mistake was replacing "Fast Lightmapper" with "Global Illumination" instead of making them both available.

 

Yep. Usually they keep legacy systems around for a while but Beast got ripped out pretty much straight away. I'm thinking maybe a licensing issue or something similar?

 

It could also be that having support for both bloated the engine considerably. Just a possibility.

 

IMO, this could've been 3 minutes shorter and does not cover even 10% of information from the "9 part guide".

 

Thankfully, a 'Fast Lightmapper' and 'Global Illumination' will both be in Unity when the Progressive Lightmapper is released. As I said before, best of both worlds!

 

hate to sound cynical but, how long after is it released will it be working?

 

And how long until the lightprobes also work? (Is it worth even asking if there's any hope for additive loading of probes?...)

 

the upside of this thread: I found that enlighten tutorial... enlightening.

It would be nice if it were a bit condensed though, and maybe had a 'quick reference guide'

 

I like very simple and easy to use lightmappers, i didn't know two of those was available, thanks for the post.

 

Is it time to start a thread on recursive prefabs yet?

 

I don't know why everyone is hating on Enlighten to be honest. I agree it was quite bad in 5.0, but now it has actually matured into a really cool piece of software. If you use it correctly and have a decent PC, light mapping really doesn't take all that long – especially when you compare how long a comparable bake would have taken with Beast. And then the effects you get from having a dynamic lightmap are just breathtaking. The ability to have a full day-night cycle with realistic lighting is mind blowing.

And really, you're comparing those lighting engines? Do you even understand how lighting works? That first lightmapper looks to be doing monte-carlo mapping which is known to be one of the slowest, least physically accurate method. The other two are doing it dynamically. And with that, they both have a fair number of restrictions.

That raytracer actually is unusable for realtime as it takes a fair number of seconds to compute a finished image. It also doesn't compute an image with much global illumination. Then Segi is great, but that's not without restrictions on its use (high end computer, small scene).

Basically, enlighten is amazing stuff, no asset on the asset store provides what Enlighten does. Don't hate on something you haven't done better with yourself.

 

So do I need to write my own lightmapper to dislike enlighten?... I never liked that logic.

But I agree, the precompute stuff are cool and baking is actually doable now.

There are issues with contrast and light fall offs and gamma being super weird but they are not exactly enlighten issues.

 

^^This...

Except the last bit about being able to write a lightmapper yourself.

 

There are several people on forums that would be able to write an alternative. The problem is it would take few months.

So, are you paying for this?

 

I was just trying to suggest that you have a less legitimate stance as a complainer if you don't actually know what it would take to fix the problem at hand.

To be honest, I'm highly doubtful of this. A light mapper better than Enlighten would be worth a ton in the industry. If a person really could do this, they already would have.

 

I don't quite agree, sure you could say, I think I know what it takes because I've tried writing one before but it doesn't negate your point if you haven't.

Same like a game player giving critique on a game even though they may never have tried to write a game before. They probably have valid points and arguments. You can't just tell them to go sling their hook.

I'm inclined to agree with this. That one man light mapper in the OP is probably quite simple, and just looking at the cost to license enlighten tells me that price must have had a ton of RD a few forum members couldn't simply recreate or clone.

 

Okay...

 

You don't have to be a chicken in order to recognize a rotten egg.

Basically, "make an alternative first!" is an excuse frequently made by people in order to dismiss opinions of others instead of listening to them.

That's not the right kind of thinking. It is like saying "heaver than air machine will never fly" a day before Wright's Brother's first flight. The issue here is that it takes time, so someone needs to pay upfront to fund the development.

A lightmapper better than enlighten already exists in unreal 4. The issue is that Enlighten is geared towards being GI, and as lightmapper it doesn't really work well. Another issue is that there was a different lightmapper that was removed, and some people actually liked the old lightmapper more. Then there are issues with transparent surfaces not being handled, manual light probe placement, requirement to read long manual and manually optimize scene for 30 minutes in order to have faster build time, lack of GPU acceleration support, lack of networking support, huge memory requirements, etc, etc, etc.

Those are FLAWS. They need to be addressed. Being defensive about your favorite product is not helping anybody.

Also, just because something has a high price tag, it does not mean it cannot be recreated.

 

haha, fair enough!

 

We used enlightens realtime GI for our game 'The Station' and are very pleased with the results.

http://thestationgame.com/

 

Not sure if this is a bug or feature, but found Enlighten is not baking correct information to LightProbes. Top image is the Progressive Lightmapper, bottom image is Enlighten. Tried numerous scene setups and Light panel setups...

 

Enlighten lightmap baking is garbage, sorry for the harsh words.
Their direct shadows baking is the worst ive seen. It’s just a couple of taps with a simple filter on top.

 

...hence the development of PLM, and it's awesome incoming GPU version.

 

Was about time, it have been like that since v5.

 

The acceleration of PLM is going to be a balm on our development and soothe our prickled indie hides.