Search Unity

SEGI (Fully Dynamic Global Illumination)

Discussion in 'Assets and Asset Store' started by sonicether, Jun 10, 2016.

  1. SteveB

    SteveB

    Joined:
    Jan 17, 2009
    Posts:
    1,448
    I'd participate in all of those, especially the letters...
     
  2. vonpixel

    vonpixel

    Joined:
    Oct 2, 2012
    Posts:
    31
    Right? I'd buy it again just make sure it keeps getting dev time. I know he has some issues to deal with personally (I hope everything is good with you SE- take all the time you need)- but I would do a patreon. Just sayin.
     
  3. andywatts

    andywatts

    Joined:
    Sep 19, 2015
    Posts:
    42
    +1 for light bleed fix without needing a blocker mesh and material.

    I'd also do a patreon.
     
    RB_lashman likes this.
  4. Flurgle

    Flurgle

    Joined:
    May 16, 2016
    Posts:
    368
    I'd do a patreon
     
    RB_lashman likes this.
  5. elbows

    elbows

    Joined:
    Nov 28, 2009
    Posts:
    2,299
    And me.
     
    RB_lashman likes this.
  6. dadude123

    dadude123

    Joined:
    Feb 26, 2014
    Posts:
    787
    Wow! That video where he's playing around with pro builder is so good it seems almost fake.
    The lighting is so fantastic, especially the part where the "infinite bounces" thing was shown.
    I have a hard time believing that this works as well as in that video...
    Somehow it looks too good to be true.
     
    RB_lashman likes this.
  7. Mauri

    Mauri

    Joined:
    Dec 9, 2010
    Posts:
    1,522
    It works just like that, but as it has been stated multiple times: SEGI (in its current form) is super hardware hungry and not ready for usage in actual games. You'd need a decent computer to reach stable/playable fps. The announced cascades update might improve things along those lines further.
     
    Last edited: Mar 19, 2017
  8. rainisto

    rainisto

    Joined:
    Jul 19, 2015
    Posts:
    41
    Yeah, it would need to be at least 4 times faster to work in VR with a reasonable scene... And I have a 2 x GTX1080 rig here so that's likely an optimistic estimate. Still impressive nonetheless.
     
    RB_lashman likes this.
  9. Flurgle

    Flurgle

    Joined:
    May 16, 2016
    Posts:
    368
    @rainisto I was under the impression that SLI GPU cards don't work well in Unity (though there's an nvidia guide). What is your experience?
     
    RB_lashman likes this.
  10. castor76

    castor76

    Joined:
    Dec 5, 2011
    Posts:
    1,416
    can any one explain this periodic spikes in the profiler. It is supposed to be Gfx.WaitforPresent. It is caused by using SEGI.

    upload_2017-3-22_23-36-4.png
     
    arnoob and RB_lashman like this.
  11. SteveB

    SteveB

    Joined:
    Jan 17, 2009
    Posts:
    1,448
    Hmmm I thought WaitForPresent was a Unity Editor thing. Profile a Standalone?
     
    RB_lashman likes this.
  12. castor76

    castor76

    Joined:
    Dec 5, 2011
    Posts:
    1,416
    The same thing happens. but at every frame now. I guess, it is just the performance hit with the SEGI. Running on GTX 660 and I can't run it at 60fps on high setting. It doesn't matter what other options I change.

    I really wish that the voxel volume shape can be more flexible so they can be optimized for the different types of game. If you create a say RTS game, you don't really have a lot of "depth" under the ground or even for above ground. At the moment it seems like the bounding size is just a cube so maybe a lots of performance is lost even if they are not needed.
     
    RB_lashman likes this.
  13. LaneFox

    LaneFox

    Joined:
    Jun 29, 2011
    Posts:
    6,435
    Gfx.WaitForPresent means that the CPU is ran out of work and is waiting for the GPU to finish drawing so that it can begin processing the next frame. This shows as a CPU spike because the cpu literally cannot do anything else until the GPU finishes.

    It means you are GPU bound, at least at certain times.
     
  14. mykillk

    mykillk

    Joined:
    Feb 13, 2017
    Posts:
    52
    I'm very interested in purchasing this asset, but I've been holding off until the new version releases. Hate to be a little rude, but the author should hear this honest feedback from an interested customer, that for an $80 asset in such an early development state, updates need to be more regular...Now I'm no longer sure I want to invest in this asset at all without confidence it will continue to be developed, especially given that the progressive lightmapper has eroded one of the main advantages of SEGI and that is the quick iteration of lighting.
     
    Last edited: Mar 23, 2017
    RB_lashman likes this.
  15. SteveB

    SteveB

    Joined:
    Jan 17, 2009
    Posts:
    1,448
    Fair point as we've all been eager to see this get developer further...

    ...that aside, Progressive Lightmapper is not realtime, so if the advantage to you is just faster light baking then that would be your best bet, agreed.

    Otherwise the 'Main Advantage' of SEGI is that it's 100% realtime and just so happens to be very fast...
     
    RB_lashman likes this.
  16. yohami

    yohami

    Joined:
    Apr 19, 2009
    Posts:
    124
    Is this still being developed?
     
    RB_lashman likes this.
  17. Mauri

    Mauri

    Joined:
    Dec 9, 2010
    Posts:
    1,522
    Well... sonicether was last seen on March 17. We all know that one of his family members had some health issues of unknown type (not our business anyway). Maybe something has worsen again.
     
  18. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,232
    It's the global illumination curse, anyone who dare to follow this path are defying the gods' will, no real time GI solution is left untouch by the wrath of gods.
     
    Baldinoboy, RB_lashman and SteveB like this.
  19. SteveB

    SteveB

    Joined:
    Jan 17, 2009
    Posts:
    1,448
    ^I have to say there really HAS to be something to this! :D

    Seriously SEGI has come the absolute closest to performant, gorgeous realtime GI and yet it still feels like it's circling the drain. I'm not going to start issuing requests to Cody for updates like I did last time, for as much as I love SEGI, how long will this continue and am I to hold onto hope? I've been willing to do my part financially even beyond the cost of the asset to help because I believe in it, but even THAT appears to get zero traction.

    This asset is just far too valuable to let die, but I'm my gut is telling me that in the end it may not fully be realized. Pains me to say it but yea as I've said SO many times in the past, the lack of developer communication for ANY asset is the deal breaker for many people. You'd be surprised how far a little "I'm still here guys!" post could do to keep hope alive, yet apparently that's not even reasonable to ask for.

    What do I know though right? I've had to continue working my IRL Job even through hardships, and over the many years I've had to learn to communicate as effectively as possible with my clients to ensure they feel secure I'm doing my work (it's all off-site contractor work in the games industry), and feel that their project, and my small part of it, is in competent hands, and that they will get what they're paying me to do.

    I'm sorry to lay it out there, but myself as an AdultAssMan.tm with real responsibilities (many that I've made for myself) in the real world, that's how you breed success for yourself and realize what it is you want to achieve. I see it here from so many asset developers, but equally so MANY more do not respect that. It's a shame and it breaks my heart in this specific case...
     
  20. elbows

    elbows

    Joined:
    Nov 28, 2009
    Posts:
    2,299
    How many updates to the package have we actually got in the approx. 9 months that this asset has been out? I cant remember exactly but I know it's fallen well short of my expectations and I hold assets that are marketed as being in beta with heavy development planned to a different standard to assets that are released in a mature state. I regret my purchase.
     
    arnoob and RB_lashman like this.
  21. castor76

    castor76

    Joined:
    Dec 5, 2011
    Posts:
    1,416
    @sonicether I know SEGI is supposed to be only work on Windows at the moment, but when I tried it on XboxOne platform. Most of the compute shader didn't compile with "unknown" errors. I wonder if sonic can come alive here so I can try and work with him to see if I can make it to work on that platform.
    :(
     
    RB_lashman likes this.
  22. rainisto

    rainisto

    Joined:
    Jul 19, 2015
    Posts:
    41
    I've had no SLI specific problems. I'm running dual GTX-1080 with a HB SLI bridge (they're a bit hard to come by but I guess worth it). Now I haven't done lots of measurements in turning SLI on and off and calculating what the performance gain is, but my experiences, nothing to complain. ymmv of course.
     
    RB_lashman likes this.
  23. mykillk

    mykillk

    Joined:
    Feb 13, 2017
    Posts:
    52

    I'm pretty sure someone mentioned getting it to work on PS4. Even if it could work though, I wouldn't be expecting performance to be all that great. Scorpio however should have the power for it to be doable. Can you post some screenshots of the shader errors?
     
    RB_lashman likes this.
  24. yohami

    yohami

    Joined:
    Apr 19, 2009
    Posts:
    124
    Can we decompile / retro engineer this thing if everything else fails? Im willing to do a goFundMe and put a few thousand usd of my own pocket to get this done or get someone else to do it. @sonicether come back!
     
    arnoob and RB_lashman like this.
  25. Mauri

    Mauri

    Joined:
    Dec 9, 2010
    Posts:
    1,522
    Decompile? SEGI has no .dll's. Everything comes in .cs, .shader and some .compute files ^^
     
  26. sonicether

    sonicether

    Joined:
    Jan 12, 2013
    Posts:
    265
    Hi everyone. I'm very sorry I haven't been around. I know there are people who would handle the situation I'm in much better than I have. I truly hate being at the mercy of this, so clearly I need to find a better way to handle it.

    That being said, I have worked a little more on cascades. It seems that for some scenes (the viking village, for example), the cascades provide a great benefit, and in other scenes that demand more robust occlusion and light spreading (the courtyard scene, for example), image quality suffers a bit.

    Regarding the project at its core, as I've mentioned before, it's been really difficult to find improvements in performance and functionality that don't compromise image quality or speed too much, since, SEGI as it stands right now, is pretty much teetering at the edge of efficiency and visual robustness (meaning, a step towards efficiency breaks visual robustness, and a step towards visual robustness breaks efficiency). However, I do feel that cascades is a step in the right direction, and I feel that it's time to get feedback on it before I proceed, so I'll be putting out an update that includes an optional .unitypackage to enable cascade functionality soon. The update will also include blue noise for both cascaded and non-cascaded versions.

    Something else that's been on my mind is that there may be other graphics programmers out there that could help SEGI improve in ways that I haven't thought of yet. Perhaps this project is growing bigger than me. Maybe it would even be best to make SEGI free and open-source. It would be really helpful to get your opinions on this.
     
  27. chiapet1021

    chiapet1021

    Joined:
    Jun 5, 2013
    Posts:
    604
    I think you'll find overwhelming support for open-sourcing SEGI. Even if you kept the source closed, but allowed a select set of folks assist in its development, that alone would be a leap forward in confidence that it will live on and get the attention it needs. I certainly hope you'll go down one of those paths.

    Good to see you here, btw, Sonic. Wishing the best for you and your loved ones, and hoping that you'll be able to balance taking care of your family and this community so that everyone is attended to. :)
     
    Abuthar, arnoob and RB_lashman like this.
  28. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,232
    While I haven't enough money, there is people begging to send money your way so start taking donation from them, it might help a bit?
     
    RB_lashman likes this.
  29. chiapet1021

    chiapet1021

    Joined:
    Jun 5, 2013
    Posts:
    604
    I don't think it's about money.
     
    arnoob and RB_lashman like this.
  30. AlteredPlanet

    AlteredPlanet

    Joined:
    Aug 12, 2013
    Posts:
    446
    I just want linux support, other than that I like SEGI as it stands
     
    arnoob and RB_lashman like this.
  31. Shinyclef

    Shinyclef

    Joined:
    Nov 20, 2013
    Posts:
    364
    @sonicether, SEGI is your creation so I want to see you succeed and benefit from it if it helps other devs succeed.
    That being said, I can't say I understand the situation you're in with family, and the time you're able to put into SEGI.
    Maybe you could find a partner or two to carry on the torch a bit and keep taking a cut? It would be nice to see development speed up again whilst still allowing you to see some benefits...

    As a SEGI customer, I am of course in full support of anything which will help SEGI's development.
     
    RB_lashman likes this.
  32. punk

    punk

    Joined:
    Jun 28, 2013
    Posts:
    381
    @sonicether Although i bought SEGI honestly I think making it free and open sourcing it would be the best thing you could do. You've made a great start, but it needs to come a long way before it's useable for most games, if you open source it, it will take the pressure off you and may give it the attention it deserves
     
    Last edited: Mar 26, 2017
    TerraUnity, arnoob and RB_lashman like this.
  33. BruceBarratt

    BruceBarratt

    Joined:
    Jun 20, 2015
    Posts:
    46
    @sonicether Thanks for the update. I personally think that people are overreacting a bit to the small absences. I think it's just because the asset is potentially vital for so many projects. So many things are coming together for dynamic worlds and more accessible procedural generation and this asset is such an important part of it. It's really nice to hear you are thinking about making it free and open source. Perphaps it would be worth waiting until you reach SEGI 1.* at least so that there is a clear direction for the development. Also there could be a lot more sales then and you can use the funds from that to work on other projects. I say don't short change yourself. In the further future it would be nice to see this become open source and free as a new baseline that everyone can use and learn from.

    I'm very excited about having cascades to play with. Thanks for your vision and hard work and good luck with the cascades problems.
     
    RB_lashman likes this.
  34. elbows

    elbows

    Joined:
    Nov 28, 2009
    Posts:
    2,299
    It's not about short or even long absences for me, its just about overall progress and whether this asset will ever reach its full potential. I also recognise that its not necessarily for the lack of effort, there may be problems that are proving tricky to solve at all whilst retaining the quality that the asset established from the start.

    I said I regretted my purchase because I just decided to be honest, and because nothing positive that people were trying to come up with such as alternative revenue streams for sonicether ever seem to make any difference. It's hardly the only purchase that hasnt worked out the way I'd hoped, and if it never gets much further than present state my overwhelming emotion will be sadness rather than anger.

    The opensource idea is very interesting and its welcome news that sonicether would even consider this. There are ways to do it so that it would still be possible for sonicether to profit from it, but I suppose my main doubt would be whether there are actually any graphics programmers out there with the necessary skills who are interested in working on this project. I would hope there is, but I don't get much sense that there is from the history of this thread - but opensource possibility was not mentioned before so I should make a judgement using the past only - if there are any suitable candidates out there then now would be a great time to post on this thread!

    Anyway sorry to sonicether for being negative in my last post, I do understand the realities and complexities of these things and my moaning is only because I care. Cheers.
     
    BruceBarratt and RB_lashman like this.
  35. BruceBarratt

    BruceBarratt

    Joined:
    Jun 20, 2015
    Posts:
    46
    @elbows Fair enough and well said. I've certainly worried about progress before. I agree about the open source thing. It would be really nice, but it's typical of someone doing something awesome to think "surely someone else is doing this or could be doing this", and then find only a handfull of people can understand how it works and even fewer also have the creativity and drive to evolve it. Still there must be someone somewhere.
     
    RB_lashman likes this.
  36. Martin_H

    Martin_H

    Joined:
    Jul 11, 2015
    Posts:
    3,846
    Did you talk to Unity staff about a possibility for them to buy and opensource SEGI? Maybe even hire you too? They've done similar things in the past and they certainly have enough money to compensate you for your invested work. But (imho) most importantly they've got some good people too to work on and guide an open source project. The post fx stack turned out very well for example. I generally think open source would be a good idea, but there's always the possibility of the project falling apart because everyone who wants to work on it wants to pull it into a different direction and ultimately no real progress is made and everyone leaves in frustration. I'd hope this would be somewhat mitigated when some kind of leadership remains and it was unclear to me whether you'd even want to or can fill that position yourself.
    Either way, good luck with everything!
     
    Cascho01, vonpixel, arnoob and 4 others like this.
  37. punk

    punk

    Joined:
    Jun 28, 2013
    Posts:
    381
    I agree ;)
     
    RB_lashman likes this.
  38. AlteredPlanet

    AlteredPlanet

    Joined:
    Aug 12, 2013
    Posts:
    446

    You should foucs on more platforms :)
     
    RB_lashman likes this.
  39. dadude123

    dadude123

    Joined:
    Feb 26, 2014
    Posts:
    787
    Dude...
    This whole page and the previous one are about him not knowing where exactly to get more performance out, where this project is going, etc...

    The last thing he will focus on is cross platform support when the core of the product isn't even solid.
     
    arnoob and RB_lashman like this.
  40. davide445

    davide445

    Joined:
    Mar 25, 2015
    Posts:
    138
    @sonicether as other pointed out as far as I know will be difficult to find some experienced Compute Shaders developer. It's a hot topic and probably all will be already busy in some project.
    I think you have these options: open source but for such a specialized topic and for the above reasons I doubt will be possible to achieve a solid evolution plan, find a development partner and maybe crowdsurce the project so to have the resources but all the risks will be your, or sell the asset to a bigger organization with a way remaining engaged for development and benefit from revenue stream.
    As business designer it's my work to find also suppliers or partners, will be a pleasure to help you.
     
    Flurgle, punk and RB_lashman like this.
  41. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,232
    He started se shadows because he needed money though, you always need money, and with money option open too, like delegating, hiring and contracting.
     
    RB_lashman likes this.
  42. chiapet1021

    chiapet1021

    Joined:
    Jun 5, 2013
    Posts:
    604
    Sure, everyone can use more money, but this particular stretch of development delays is not related to SE lacking monetary resources to continue to develop SEGI. This is about where his time and focus needs to be. You can't delegate taking care of your family. Not completely, anyway.

    We're getting off topic anyway. Again, I'm in support of either open source or expanded closed source. There are pros and cons to both approaches, as has been pointed out by others here already.
     
    RB_lashman likes this.
  43. Mauri

    Mauri

    Joined:
    Dec 9, 2010
    Posts:
    1,522
    Selling it to a bigger organization would be not really good, though, because you never know a) how they will handle existing customers and b) where they'll go roadmap wise. Some of us invested money into this asset, because we trust sonicether and have faith in him. It's his baby and it should remain to him.

    If everything falls, I think the best thing would be Unity Tech integrating SEGI natively inside Unity (without hiding it behind a Plus and/or Pro plan).
     
    arnoob likes this.
  44. davide445

    davide445

    Joined:
    Mar 25, 2015
    Posts:
    138
    Sorry but I have to disagree, this is also a business not only a baby. Sonic created an unique asset, and the right company can valorize it and give the right resources for tech and commercial development. M&A it's happening all the times and there are normally clause to manage the ownership transition btw old owner customers and the new one.
    Also I didn't find probable (even if will be convenient) that Unity will integrate SEGI as is doing today with other assets with Unity Prp, since it's an unfinished product with a narrow usage (being Unity a game engine focused on cross platform usage). An acquisition and internal development can make sense if they believe his usability will be extended, but this is depending from their agreement terms with Geomerics.
     
    RB_lashman likes this.
  45. IanStanbridge

    IanStanbridge

    Joined:
    Aug 26, 2013
    Posts:
    333
    Hi Sonicether it sounds to me like cascades are probably the way to go. I wouldn't agonize over the trade off between quality and performance too much and instead simply provide options to change it's quality so it can work at various performance levels. I think all voxel cone tracing techniques have trade offs and no one is perfect for all solutions. Since it sounds like you have experimented with a few simply include all of them and allow the user to select which options they want to enable for their situation.

    I imagine the ideal setup for performance would be the one that was used on the game The Tomorrow Children for ps4. Bear in mind though that you can buy far more powerful graphics cards for pc that are relatively inexpensive so don't be afraid to use higher resolution grids and filters that would be possible on a modern pc.

    A few of the comments on here from people concerned about performance and 4 year old gpus or complaining of minor light bleeding are being unrealistic about what is possible with a voxel cone tracer. A voxel cone tracer by its nature demands a high end gpu unless it is running at a very low resolution.

    If low end performance is something that is critical on a project they would be better off using Unitys standard lighting rather than a voxel cone tracer such as this anyway. The same with light bleeding it just isn't feasible to have realtime GI without light bleeding on modern hardware. If it was why would lightmaps still be a thing ?

    Anyway hope your personal situation improves, personaly I would only open source the project if you felt you had to move away from it which would be a shame. If I was you I would simply implement the cascade system and make sure it works with the upcoming Unity 5.6 . As it sounds like you are close to that I don't really see why you would need other developers to help you with that so keep your asset your own as you built it.

    Ian
     
    ThomasVeil, No0B25, ftejada and 3 others like this.
  46. macdude2

    macdude2

    Joined:
    Sep 22, 2010
    Posts:
    685
    I'm not sure open sourcing it is really the way to go. I'm skeptical that there are that very many developers with enough experience and free time to push the project in the direction that you want it to evolve in... But maybe that's not the point? It would certainly be a valuable tutorial for many budding graphics programmers if you decided to do so. I'd also hate to feel that you weren't being fairly compensated for your tremendous and groundbreaking work. But again, that might not be the point for you.
     
    RB_lashman likes this.
  47. Ascensi

    Ascensi

    Joined:
    Sep 7, 2013
    Posts:
    572
    @sonicether getting this error "Expanding invalid MinMaxAABB" on Unity version 5.5.2f1 there is no additional error message.
     
    RB_lashman likes this.
  48. Frpmta

    Frpmta

    Joined:
    Nov 30, 2013
    Posts:
    461
    Listen to this man.

    Don't focus in old hardware. You are making something modern in here.

    Just look at how Star Citizen began development targeting the 680GTX as base and people complained. To this day, the game is still under development and the 680GTX is now considered an old and weak graphics card.
     
    chiapet1021 and RB_lashman like this.
  49. Flurgle

    Flurgle

    Joined:
    May 16, 2016
    Posts:
    368
    @sonicether would the scriptable render loop allow major positive changes to SEGI?
     
  50. ThomasVeil

    ThomasVeil

    Joined:
    Aug 7, 2014
    Posts:
    4
    I second this idea. Unity must be salviating at the idea of having this as a first in their engine - it's like perfectly made for indie-games too, as it makes everything simpler. Probably worth tons of money for them - and they have the perfect staff to help you.
     
    hopeful, Howard-Day, punk and 2 others like this.