HBAO or not to HBAO..

So we received our new batch of Oculus Rift DK2 kits. INCREDIBLE. The future is here. BUT. Development is slow due to the new SDK and 75hz limitation “cap”.. looong road a head.

Trying to develop true “future” generation content at 75hz, is near impossible. Most “next gen” consoles and “next gen” console titles can barely manage 60Hz at HD, let a lone 75hz at over 2K rendered in stereo …and we are not a fully fledged development studio with $100,000’s nor $1,000,000’s to dip into.

We are an indie studio in the truest sense of the word.

We’ve had lots of comments from people who have shown a huge amount of interest in what we are doing. I wanted to say a massive thank you for being as patient as you have been with us. We are doing our utmost to try and get some products developed. Because we are such a small development studio with such tight budgets, we rely heavily on other freelancers from all over the world to help us integrate cutting edge effects and shaders into Unity. This is very difficult to do. Time and money are a constant battle. We are winning! We’re getting closer and closer each day but it takes time.

Users have posted asking to see some screen shots of progress, so I wanted to share some new results. Slightly technical. We’ve been working hard on integrating hair, better skin and some other effects like HBAO and HDAO. AO = Ambient Occlusion.

UE4 already has AO built in, so we are trying to make sure that Unity can match UE4’s visuals, as this is the best real-time visual bar there is at the moment. Because Unity is much more open ended and can supported hybrid Forward and Deferred rendering, whilst UE4 cannot, this makes Unity more attractive. UE4 has some Screen Space SSS in the works but for the time being we are sticking with Unity. So the plan is to beef up Unity as much as we can.

UE4 HBAO

ue4-ao-03 ue4-ao-02 ue4-ao-01

Unity HBAO

Chris-HBAO-HDAO-02 Chris-HBAO-02 Chris-HDAO-02

With HBAO and HDAO

Chris-HBAO+HDAO-01

Without HBAO and HDAO

Chris-HBAO-HDAO-NON-01

HBAO and HDAO only

Chris-HBAO+HDAO-FLAT-06 Chris-HBAO+HDAO-07 Chris-HBAO+HDAO-08

Comparison with and without

Chris-HBAO-HDAO-NON-03 Chris-HBAO-HDAO-03 Chris-HBAO-HDAO-FLAT-03

Ambient Occlusion is a subtle effect but it’s integral to ground objects into the world. Rather than baking in AO we now have dynamic AO, which will coincide with our real-time skin deformation research. We are trying to convert everything over to be Physically Based.

The hardest part so far has been the jump from DX9 to DX11 and trying to go fully deferred which just isn’t an option for real-time skin and hair. So we have to go hybrid.

The next hurdle we face is the self imposed limitation by Oculus of 75Hz.

Some experts on Reddit pointed this out:

elexor

75hz is not a limitation. I hope cv1 has 120hz display. If i can tell the difference between 60 and 120hz on a pc monitor, imagine the difference it will make using a vr headset. fps > graphics you can’t just cut corners for vr, unlike console games which run at a horrible 30fps. I would take worse graphics and better motion to photo latency any day.

I have a dk1 and it gives me a headache because of the 60hz and the motion blur. low persistence tech and high framerates are essential to vr’s success.

For vr to be perfect we would need a 1000fps 1000hz https://www.avsforum.com/t/1484182/why-we-need-1000fps-1000hz-this-century-valve-software-michael-abrash-comments

.
and another:

sinoth

Agreed. Calling higher framerate a “limitation” is short sighted and defeatist.

So the plan is to beef up Unity as much as we can.

This is the problem… I cringe to think of what the shader stack they are using looks like. If you can’t manage 75Hz with a single model in a scene then you are doing something very wrong.

 

.

Agreed. 75Hz and eventually 120Hz is essential for VR. We don’t dispute that.

The next hurdle we face is the self imposed cap by Oculus of 75Hz.

I used the word limitation before because we ARE limited by current GPU technology. Trying to recreate realistic humans at 75Hz in an application like Unity is very hard to do in even in off-line rendering. Think Benjamin Button, Avatar, TinTin. Even then they are stylized examples, with teams and teams of creative people and access to rather large budgets!! Rendering realistic humans on a budget at 75Hz is near impossible. Show us some examples of the problem solved?? and we will bow in defeat!!

If you drop even one frame below 75Hz (with Vsync on) we will get vertical screen tearing and judder. Which is the equivalent of putting hot pokers in your eyes!! With the DK1 if you dropped below 60Hz, it was no big deal, you could just about cope with it. With the DK2 forget it. 75Hz or GTFO!

Our next challenges are numerous. SSR is the biggest. Screen Space Reflections are fairly trivial to implement, the trick is getting good frame rates. This is our next challenge, as well as order independent transparency.

More updates to come. The “bottom” line is, if you want realistic demos in VR. You’re going to need some powerful hardware to run them!!

Chris-HBAO+HDAO-10 Chris-HBAO+HDAO-11 ssao+hbao-test-01

Seeing HBAO, mixed with HDAO and SSAO in VR is a real treat, screen shots do not do the effects justice. More optimizing to come.

We hope to launch some products this month.

V

 

Posted on September 15, 2014 in Research, Virtual Reality

Responses (60)

  1. Need
    September 15, 2014 at 4:45 pm · Reply

    Thanks and great works !

    We waiting to test this new demo.

  2. Laudrac
    September 15, 2014 at 6:21 pm · Reply

    Dear Veiviev Team,

    Thanks you for sharing on what you working actually and it really Looks good and fully understand your strugle. what do you think of allowing People here to jump in to croudfunding like Kickstarter but just on your page. Creating PayPal allowing People to pre-purchase packages. so we get a better product and you the funding you Need to create perfect experience. you said 20-30 per package so if you have 10 package i am in for all then and i think others as well. Cheers

    • veiviev
      September 15, 2014 at 8:26 pm · Reply

      Thanks Laudrac, we have thought about it. It’s possible but PayPal doesn’t allow donations to LTD companies they also certainly don’t allow payments for anything that is adult orientated. But we will look into it further.

      • Ati
        September 17, 2014 at 2:40 am · Reply

        Consider Bitcoin?

        • veiviev
          September 17, 2014 at 10:11 am · Reply

          Yes we have it set and ready to go.

  3. Johnny
    September 15, 2014 at 6:29 pm · Reply

    Thanks for the update. Very impressive work! Most of us with a DK2 never got to see your older DK1 work in VR. Anything at all that you can release that has DK2 support would be sweet 🙂

    Great Job!

    I really think you are going to make a nice profit in the future. Looking forward to your next update!

    • veiviev
      September 15, 2014 at 8:28 pm · Reply

      Thanks Johnny, something will be published soon. What do you think is a fair price for a 10 pose demo?

      V

      • Slartibartfast
        September 16, 2014 at 1:13 pm · Reply

        That’s a simple question with a bunch of complicated answers. If I were you, I would seriously consider the following: Let people pay whatever they want for the demo. If they want it for free, they get it for free, but if they want to pay for it, there’s no upper limit to what they can give.

        Why do it this way? Because the people owning VR units at this point are all enthusiasts or fellow developers. By telling them that paying for the demo is vital to the future of these products, I suspect a lot of them would be willing to pay a lot more than if you set a fixed price. They are more likely to want to help push the development of cool new VR software along than the regular users that will be buying the Oculus Rift CV next year. This would also increase your exposure, as there would be nothing to stop those who are merely curious from trying it out, and then maybe coming back to pay after they see your demo for themselves. With the limited amount of VR users right now, encouraging a few to give more, rather than forcing everyone to give a little, might result in better revenue. Once you’ve got a wider selection of products out, you can keep this first one on the same business model, essentially maintaining its status as a demo without causing any PR issues with neither new customers nor old ones.

        There’s also the risk of a small, public backlash if you sell demo software. I believe that making payment voluntary, you’ll ironically make people more likely to want to pay, and pay more.

        Of course, any future products you put out should probably follow a more traditional business models as this is a trick that is likely to have diminishing returns.

        • veiviev
          September 16, 2014 at 2:07 pm · Reply

          Some great ideas Slarti. The only problem with that is, it’s an unknown. No one has really done that yet. We struggle as it is financially. I’ve personally put in multiple £100,000’s into all the research and technology behind the technology to capture and replicate people in 3D. The custom software development has also been a huge expense. Server costs are also expensive. The more that people download and use the bandwidth, the higher the fees. The risk is (to cater to the now generation) if we get 10,000 DL on the first demo. Similar to Lucid Dreams and maybe 100 people pay $10, it’s a failed system. Each one of those human scans are worth at least £1000 in development cost terms. So selling them for $1 each in the hope people will pay, that is a huge risk.

          • nomi
            September 16, 2014 at 2:43 pm ·

            If you want or have to make money with the first tiny demo, then yes, I agree. Its a risk.

            If you want every Rift DK1/2 user to try it out (and spread the word) you should hang the fruits low.

            The use and repetition value of looking at a (granted beautiful looking) 3D model is relatively low for the average person and I’m not sure I would spend more than a symbolic amount for that. Its not a game nor a real application, its a demo and should imho be handled as that.

          • veiviev
            September 16, 2014 at 2:59 pm ·

            Interesting. Well the fist batch of demos will be just that viewers but we are working on hand interaction, eye follow and skin deformation to allow for actual contact into the VR world. Touching and manipulating the models skin. Perhaps this might justify higher prices. Unknown territory. Kind of an adult type “Madame Tussauds” experience, hopefully a little more realistic and a little more tactile!

  4. belowzero
    September 16, 2014 at 7:24 am · Reply

    Can’t wait to see the finished product! Regarding crowd funding as some have already suggested, I think it is a good idea. Have you guys checked out https://offbeatr.com/ ?

    • veiviev
      September 16, 2014 at 10:47 am · Reply

      Thanks, yes we are aware of OffBeatr, the problem is they have a cap of $30,000, they don’t allow any more (and we need to raise a lot more!) they will also take a %. We might try to self crowd fund on the main site.

  5. Name (required)
    September 16, 2014 at 9:13 am · Reply

    This month as in September or this month as within thirty days?

    • veiviev
      September 16, 2014 at 10:47 am · Reply

      September is the plan!

  6. DiVR
    September 16, 2014 at 12:25 pm · Reply

    Nice read. It’s a wonderful job you’re doing. Being a perfectionist myself, I know how hard it is having to compromise quality to make something practical, especially due to technology limitations, but of course it’s better to have something “very good” than not having anything at all. There’s a point after which your efforts increase exponentially but your product value stays almost steady. Be careful not to go much beyond that point, as it’s bad for business.
    About crowdfunding, it really could be what you need right now. I believe for a 10-model demo (no animations or interactions), something between 10-20 dollars would be fair at this stage (1-2 dollars per model, for that matter).
    Have you ever considered sharing one model on a single, non sexual pose with the community (like a free sample on http://www.triplegangers.com) to have more people collaborating on shaders and implementation? You could get some interesting feedback.
    Anyway, thanks for the update!

    • veiviev
      September 16, 2014 at 2:01 pm · Reply

      Thanks DiVR. Some good ideas on price point and sharing content. We will certainly think some more about that.

  7. Laudrac
    September 16, 2014 at 5:39 pm · Reply

    Fully agree for the 10-20$ for the 10 models demo, you could also create a 1 model demo for the first People wo don’t own DK1 and DK2 like me to convince that the 10-20$ it is absolutelly worth to spend. Secondly create pre sale packages like 40$ demo plus next model that is to be released 60$ for 2 model package and so on. So this should give you a nice amount in cash that supports the Overall process.

    • veiviev
      September 16, 2014 at 10:22 pm · Reply

      Sounds good. Will think about it somemore.

    • Slartibartfast
      September 17, 2014 at 4:47 am · Reply

      I would strongly advice against those prices. As a consumer-level software product made for people mostly interested in computer games, comparisons will be drawn to game prices. If people can draw direct parallels between the price of a couple of static models and a full-blown AAA computer game, your product won’t come out of that comparison intact.

      The idea is to find a price level that is so low that people “can’t afford” not to buy your products. This is, in my opinion, one of the pitfalls modern porn sites have fallen into. Most of them cost around $30 a month from what I can recall, which is a large enough sum that it starts to stick out in anyone’s budget. Compare that to services like Netflix for example (disregarding their enormous user base that allows them to keep prices down), and users end up letting the subscription keep going because they hardly notice the monthly cost.

      I don’t know what is being planned for this technology, product-wise, but I think relatively frequent releases of new models/poses at a reasonable price per “pack” (around $10 to make the cost “invisible”) is better than a few packs at a higher price. But at no point should actual profit be expected until the release of the Oculus Rift CV1. It’s almost impossible to be profitable while selling to the current Rift market because of its small size, something a bunch of Rift-oriented Kickstarter campaigns have noticed to their detriment.

      The solution is not to increase prices, but to plan ahead and be prepared for when the market finally arrives, treating the currently limited amount of users as a test bed that can at least keep the ball rolling, economically.

      • veiviev
        September 17, 2014 at 10:15 am · Reply

        This is all true but it’s a bitter pill to swallow. What we are providing is also something very new, something that hasn’t been done before. It’s a way to change the (faltering) adult industry. I doubt anyone pays for that any more, due to all the free to view content. We’re offering literally a new dimension to the market. We are going to try and stick to the $10-$20 range. I want to stay away from subscription based models. I would imagine the CV1 wont be out until 2016/17, then it will take time to be adopted. It’s a long road a head. Plus side, bigger and better GPU’s!!

        • Slartibartfast
          September 17, 2014 at 11:29 am · Reply

          The real plus side to this for your company is that, by the time the CV1 arrives, you will be years ahead of potential competition in terms of technology and content. All of this is a risk (but you knew that before you even started), yet it will give you a market position that most entrepreneurs would kill for, figuratively speaking. I say that as a former technology entrepreneur myself. The hardship and bitter pills of the moment can lead to great things over time, as long as you play your hand properly and skillfully. From what I’ve seen of your products, you should have no issues with that last part.

          • veiviev
            September 17, 2014 at 11:47 am ·

            Thanks for the words of encouragement Slarti!

  8. Floyd
    September 16, 2014 at 5:45 pm · Reply

    You responded to Laudrac’s suggestion of crowdfunding/preorders that Paypal does not allow adult content or donations to LTD companies. Well there are other options. For example bitcoin, as it’s decentralized they have no usage restrictions. It also has very low fees, is non-reversible and anonymous. In other words it sounds perfect for these kind of uses. It is already accepted by for example oculusrealporn.com

    • veiviev
      September 16, 2014 at 10:22 pm · Reply

      Thanks yes we have Bitcoin in place for the store, we could use that. We can also use PayPal until it gets reported and shutdown.

  9. Drake
    September 16, 2014 at 6:09 pm · Reply

    Look at async timewarping. This allows you to maintain 75fps, but without updating the scene that fast. you basically timewarp the last motions from the last scene rendered. And you render in a separate thread to be later timewarped. Basically, timewarp will sometimes allow the same scene to be rendered twice, but in a slightly different head location. If you can get this to work, then you will also be ready for the CV1 (90fps we hear)…

    • veiviev
      September 16, 2014 at 10:22 pm · Reply

      Sounds too good to be true 🙂

    • Jeremy
      September 16, 2014 at 11:06 pm · Reply

      I’ve read that time warping has been the cause of jitter on DK2 demos like Tuscany when I searched why I was seeing it myself. The recommendations in forums were to turn it off.

      • veiviev
        September 17, 2014 at 12:36 am · Reply

        I’m not sure I agree here. We have time warping integrated, you can turn it off with a push of a button. The difference is quite apparent with it. It does have a use, the same with UE4. You can notice the difference with it off. But everyone’s machine and experience seems to differ. On my test machine it’s useful but for others, maybe not.

  10. Jeremy
    September 16, 2014 at 11:10 pm · Reply

    I might be among some of the people here who, while grateful for your meticulous attention to detail and perfection, are wondering what’s hindering the release of even a DK1-quality demo for the DK2 that allows you to at least use the basic head tracking for looking around?

    • veiviev
      September 17, 2014 at 12:40 am · Reply

      Perfection. The DK2 was a spanner in the works. The SDK and OVR integration has been a nightmare. I’m not sure if you’ve seen the user experiences posted but there have been hundreds and hundreds of problems and issues. Not just for users but developers. For VEIVIEV’s stuff all our previous scripts and shaders have had to be re-written for the DK2. New OVR. DX11, deferred, all this is required for a better experience and to just work on the DK2 to overcome judder and lag. A lot of optimizing. We can release a simple demo but I’m certain people will complain “it’s not good enough” “they don’t look real” “I can’t fap to this” or “that’s not worth $10” etc We’re doing our best. We are also a 2 man development team, building this, whilst trying to get money in from other work to put food on the table.

  11. Dustin
    September 17, 2014 at 1:26 am · Reply

    You guys are doing great work here. Keep fighting the good fight!

  12. Loki
    September 17, 2014 at 2:37 pm · Reply

    ever thought about croudinvesting?
    I would invest € 1,500 in their company and there are certainly many others would invest more …

  13. Fretchen83
    September 18, 2014 at 3:20 pm · Reply

    Looks great so far. But only models with small boobs 🙁
    How many girls will be in the next version?

    • veiviev
      September 18, 2014 at 4:49 pm · Reply

      We plan many different shapes and sizes to cater for all. On launch it will be a limited number to start with.

  14. Oculus Rift, Facebook « Khannea Suntzu
    September 18, 2014 at 6:02 pm · Reply

    […] takes complaints from users about nudity seriously. So when I posted an article about the applications of Oculus Rift in a few dozen groups on Facebook, I made the header image […]

    • veiviev
      September 18, 2014 at 6:59 pm · Reply

      Yes Facebook (much like YT) don’t like tits, nipples or ass. Beheading and deaths… that’s OK.

      • VRSilver
        September 20, 2014 at 5:57 pm · Reply

        Yeah everyone complains about there being too much sex and violence in todays media. I think we could take a lesson from bonobos (our closest primate relatives, right up there with chimps) – less violence, more sex. Everyone is happier when everyone is having more sex.

        Also, the earlier comment was spot on about $10 being “invisible”. I’d spend $20 for the 10 model pack, probably, but I’d think about it. $10 wouldn’t even be a question at all. If $20 is a pricepoint for projections of profitability for 10 model scans i’d recommend breaking into two packs of 5 models for $10 each.

  15. michael
    September 20, 2014 at 1:01 pm · Reply

    DK1/2 owners love to back new companies, only have to look at kick starter VR related products to see this.
    A little like, “build it and they will come”, “give a way of giving you money and they will!”

    /great job so far, can’t believe you don’t have a huge office + dev team by now considering how huge your product is going to be!

    PS
    Im sure there are plenty of people willing to donate their time to scan those beautiful models for you :D)

    • veiviev
      September 22, 2014 at 10:03 am · Reply

      Thanks. Whilst that is true, a lot of software kickstarter projects, based on VR, have failed to reach their target. People love to support hardware but not so much for software. We are looking at crowd funding options.

  16. André
    September 22, 2014 at 4:33 am · Reply

    Please, make Ambient Occlusion optional (and I would even make it off by default). Or yet, a slider.
    It used to be called dirt maps for a reason.

    You are rendering people, which bleeds and reflect a lot of light. Ask your nails (if they aren’t dirty). But maybe I’m focusing on minor drawbacks. I trust you guys are thriving for the best RT balance of realism and feasibility. I just got scared if old habits taken as best practices by many seasoned developers would drag my porn into the uncanny valley. XD

    So I’ll leave my 2 cents anyway:

    At least for most architectural visualization artists (the most static of static subjects) on these times of path tracers and no light leaks, AO is now used mostly only when it’s a style choice rather than as a shadowing tool. Realistic renders prefer not to use AO, since nowadays contact shadows of baked light are much more precise.

    We all do love scattered diffuse lighting and shadowing. The problem of gaming render culture is that just recently it left the 0-255 universe. AO is often used still on that context ignoring linear workflows.

    Unity 4 is still very shy on the matter of HDR. If I’m not mistaked, sunlight in it is by default assumed with a value of 10, against an environment of 1, and artificial lights just a little bellow that. I suppose closer intensities of light sources helps Unity 4 and Unreal very biased and lightweight CPU renderers workflow.

    Even some offline render artists that use HDRI as light sources do not properly boost the sun light. Rarely HDRI (to tell the truth, I’m not sure if there is any) have enough exposure stops to properly represent the natural sunlight. It all means that ambient lights (and shadows) are often more pronounced than they should be. This way is also easier to composite the final image into a LDR exposure. But it does drag the image a bit back into the uncanny valley.

    Keep it up! Can’t wait for DK2 support.

    • veiviev
      September 22, 2014 at 10:01 am · Reply

      Just to be clear, we aren’t implementing dirt maps, nor are we baking them in. Real-time HBAO is very different from old school techniques. Our HBAO and SSAO reacts to light. So it dissipates when light shines on it, similar to what we see in real-life. We also use IBL combined with HDR.

      • André
        November 9, 2014 at 5:14 am · Reply

        Semantics.

        Although I recon (and am aware of) improvements to the different RT AO techniques, the shared suffix indicate similarities. I’m aware of the AO-removal-from-light-incidence trend/improvement/belief-it’s-finally-“realistic”.

        But I tell you that AO is indeed uncanny outside of the technical appreciation. I would even go as far as call it more kitschy than the old times of dirt maps because back then there weren’t better options. SSAO just passes as realistic by the STYLE considered acceptable by “seasoned game developers and players”. In my opinion IBL+the captured texture you are already capturing (as it already has all the AO you could really have) are better alone for realism. AO would only do any good when crossing others models and the ground which is not really the main focus of the static models.

        • veiviev
          November 10, 2014 at 10:40 am · Reply

          AO is extremely important for digital realism. Our models will rotate in world space, we will also have finger to skin interaction. AO will simulate skin contact. Trust me, without AO, in any real-time simulation things look very flat. It’s why all the AAA titles use it. Whether SSAO or HBAO or combined.

  17. Need
    September 22, 2014 at 10:39 pm · Reply

    Hello, i need news.
    Demo fot this week?

    Thanks and good works!

    • veiviev
      September 25, 2014 at 11:47 am · Reply

      No comment as of yet.

  18. laudrac
    September 23, 2014 at 3:16 pm · Reply

    Hi Veiviev Team,

    IS still end of September your Target Date to Delivery the dk2 Demo?

    Kind Regards
    Laudrac

    • veiviev
      September 25, 2014 at 11:47 am · Reply

      It’s the plan but time is running out. We are very busy.

  19. Chase
    September 24, 2014 at 8:42 am · Reply

    You can count me in for those £10 packages, definitely be willing to pay for the work you’re doing.

    Also, Oculus will be releasing the mobile SDK for the Gear VR headset end of October. A lot of people will be buying the Note 4 and with it that Gear VR. Since both are made by Oculus i’m hoping you can convert the software and make it compatible with Gear VR. They already mentioned that the SDK will be supporting the unreal engine and that had me jumping back here cause I remember the work you did with that.

    Anyway, excellent work. Keep it up!

    • veiviev
      September 25, 2014 at 11:47 am · Reply

      Thanks for your support. By the sounds of it GearVR is going to be a very carefully monitored closed off system I doubt there will be much adult content allowed, until it’s hacked for more freedom in some way.

  20. belowzero
    September 25, 2014 at 5:44 pm · Reply

    If you are going to go with the pack scheme, I’d let people download it for free with say a couple of model poses baked in, and then sell the packs on top of that. That would give you a lot of exposure from the free downloads, and at the same time earn the monies by selling packs. As a person who’s tried your dk1 demo I’ll buy any and all packs in a heartbeat, and that’s kind of the point. I think it would be a lot harder to sell to someone who hasn’t tried the software, than to someone who has.

    Oh and don’t call it a demo. Early access is all the rage 😉

    • veiviev
      September 26, 2014 at 10:05 am · Reply

      Interesting. “Oh and don’t call it a demo” – What should we call them?

  21. Dino
    September 25, 2014 at 9:38 pm · Reply

    Check this out…. amazing video.

    It would be great to have your 3d models in nature.I know its hard work, but maybe in the future ;).

    • veiviev
      September 26, 2014 at 10:06 am · Reply

      Thanks for the video but we don’t support Euclideon’s efforts. Models in environments might come in the future, once technology has improved and we have the budget for it.

  22. silence
    September 26, 2014 at 10:10 am · Reply

    pls post just a few pictures of the demo which you will offer.

  23. OulusGuy
    September 27, 2014 at 9:38 pm · Reply

    i know you keep being asked all the time but can we assume you won’t release the DK2 demo this month anymore seeing as it is already the 27th?

Leave a Reply to Johnny Cancel reply

Your email address will not be published. Required fields are marked *

Back to Top