My observations


#1

First I want to thank all those programmers who have put such hard work into High Fidelity. I remember when it started and I first poked in to see what it was all about. I liked the idea but was not impressed with the implementation. The whole cartoonish nature of it just didn’t seem forward thinking to me.

After looking at all the available offerings I thought, “What would I want if I could write my own VR social platform?” This is what I came up with:

  1. It would have to be open source
  2. It would have to P2P
  3. It would need to use the block chain to ensure identification
  4. It would need to have the world engine on the client side
  5. It would need to be as close to AAA quality as possible.

I came back to find that High Fidelity has fulfilled most of my list! I say that because we are barely approaching AA quality at this time. I don’t want to be flamed here. My familiarity is with PHP and Javascript so I understand that writing a 3D engine is no small feat. I know there are allot of other pressing issues such as file formats, content creation, ect…

All ego aside why are we not using an Apache friendly open source 3D engine like OGRE? Sansar is getting much accolade over having a “better” 3D engine. I won’t get into why I can’t stand Sansar a this time. Why aren’t we soliciting the expertise of contributors like Clement Foucault who is the brainchild behind Blenders EEVEE? EEVEE is the real time viewport which will be included in Blender 2.8. Google it and you will be amazed. Accomplishing AAA quality would be a turning point for HF in my opinion.

I can only think of one reason not to complete this and this would have to be an issue with bandwidth. I say let’s bump up the poly count and have some sort of animation play as things load. You could play an animation of traveling through a wormhole or entering an event horizon or how about Dr. Who’s phone booth flying through space? The idea is that we would play something to keep the user entertained while a domain is being loaded. We could allow content creators to create new “traveling” experiences for their own domains.

Other than this I am very happy with the direction of High Fidelity and have decided to make this my platform of choice. For now, I will be focusing most of my attention toward MakeHuman to make it as HD compatible as possible. To be successful HD needs tools for non technical people to create their own avatars.

That’s it for now… thanks for listening.

Haptic Monkey


#2

I did a somewhat similar post some days ago. I was answered that to have high quality rendered scene you have to set good lighting environment and good zones, which is obviously true. I was also answered that HF puts much emphasis on avatars compared to the environment.
The thing is that under default conditions the quality of the rendering in HF still appears cartoonish, and needs deep and diligent work to have it look AAA quality acceptable. Looking at most of current domains it seems that most of the creators are still producing cartoon better than realistic domains.
Snapshots taken from HF tend as well to remain quite dull colored and 2Dsh.

This is not what happened with other platforms. I’m not technically enough expert to say if this is due to the qt5 engine and if ogre could do better. But repeating myself, something should be done to improve the A- perception people has while entering for the first time in HF.

I like cartoonish and low-poly as well, so what about have the option to distinguish AAA domains from the low poly domains so that HF can have a better quality status for people looking for photorealistic vr, or low-poly?

If you look at the high fidelity entry point for exploring


only the last place seems to be AAA photorealistic, while the others seems essentially low-poly.


#3

claudio.pacchiega, thank you for that. I will need to look at the National Park domain and try to discern where the disparity comes from. If the capability is there then it’s more than likely that the older domains have not been updated to use the new features made available… I agree that some people like the cartoonish and that certainly is a viable style. It just comes down to a matter of taste. Businesses on the one hand likely wouldn’t appreciate a cartoon experience, that is unless you are Google… ?;O) I am thinking that the benefit of an AAA 3D engine resides on the client side.


#4

Photorealistic and low-Poly, 2 different things.
You can do a cube with a photo photorealistic texture that will stay low poly.
But a tree is definitely costing lot of polygons to look good (independently of the texture)

You can them make a low-polygon building but with photo photorealistic textures.
But the cost will be in download time for the user.
This is certainly why people take it easy on the textures.

There are ways to get the thing more optimal too. Like keep non-embedded texture for landscape element to allow reused the texture on many element to be loaded on time.

For what I have seen, the catoonish we feel in High Fidelity right now is because most of the domain are build to look like that. (intentionally or not)

Note that it’s not very easy to get the realistic aspect. HF is for what I have understood it’s own standard, if you don’t build for it and hope to have things imported from other source to work fine without any rework or adjustment, it might not be perfect.


#5

If blender eevee and hifo worked hand in glove together,the Blender army would descend in here and we would be swamped in amazing creative builders
Maybe we cant take that


#6

This is not exactly true, even at it’s best Hifi graphics barely approach single A in my opinion. It has many gaps that are very concerning as we get closer to full release:

*No post processing (blur / depth)
*Shadows yuckky
*No bloom/HDR
*Main world light yuccky
*Fog/haze needs more work
*No LOD referencing
*No occlusion or occluder areas
*Zone lighting limited options


#7

What do you mean?
Depth of Field?


#8

Disclaimer: I really love HighFidelity and so this is just my brainstorming thoughts, so my posts are not meant to be critics :slight_smile:

I was just considering the “marketing impact” of an experience from a casual visitor coming to High Fidelity. The name itself should be a “letter of introduction” to this virtual world. So entering in a virtual world so named would give some expectation to visitors. Should it be named “CartoonLandia” everybody would expect comic- like characters or anime. When I use the term photorealism or high poly I’m not referring to the technical aspect, but on the perception on being in a High Fidelity virtual world, so that people expect to experience this aspect.

Consider that the default avatar is the wooden mannequin which has a good pbr wood, so it is exceptional from that point, but definitely comic like.
The sandbox is as well a bit cartoonist so that the first impressions on the creators is to go that way.


#9

claudio.pacchiega, I so agree with you. As a first time user the sandbox confused the heck out of me. I had to talk with someone in the information booth at the hub to understand I had to delete all that stuff in my sandbox to move forward. Then I still have all these scripts auto-spawning presentations. I know I have to learn the system but think about the experience of a non-technical person. Some kind of “clear” function to allow the sandbox to be reset would be useful. It would be nice to be presented with a few options for pre-built experiences. Let the user decide if he wants to live in a cartoon land or on the beach.

I don’t think it’s critical at all for us to point these things out. What value are we if not to provide High Fidelity with feed-back both good and bad. A good motto is “Question, don’t defend.” Certainly, they have the option of ignoring us completely. I am a little fuzzy on how the open source community works here. Is there really any democratic system in place? I’ve seen invitations to HiFi events but I have not experienced them yet…

I also have to be careful not to be impatient. The development process takes time and resources. I would think however that these kind of things would be hammered out before we leave Beta, else I doubt HiFi will get the traction it deserves.


#10

On that aspect of the question, I do aggree with you. It’s a bit too much welcoming people in Toontown. (And it was worse 8 months ago… Remembering ‘Cupcake’ domain)
It’s discarting many people at the entrance because of this homogenity in the style. Clearly, it doesn’t induce the dream for many people.


#11

Yes Depth of field, it just seems like cutting edge rendering is not a big consideration for them at this time.


#12

deep breath
Okay, I really didn’t want to jump in, but this has been on my mind for quite awhile… so… here are my own observations (and proposed solutions):

  1. The Marketplace really needs a redesign
    Okay, let’s start this off with saying I’m not against the marketplace itself, but rather the design of it. I’m no Leonardo Da Vinci of CSS and HTML, but why are we viewing items one at a time? If I go onto Amazon, even on my phone in a listing style, I see at least 2 to almost an entire third item. Meanwhile, I can, at best, only see one item and maybe part of a second one if I’m actually looking. Somehow Clara,io is a thing so important that the banner will never go away and will forever use up screen real-estate as to cover the Notepad app below it.

    Some solutions:

    • Why can’t we just hide the banner?
      Seriously, why can’t we just have an option to not show the banner? I don’t care about Clara.io except for one time when it was funny to put a muscle man on top of a dog to ride into battle.

    • Show more than just one item
      So this one I’ve heard as a rumor that it had to do with allowing the thumbnails to be big so people who make items have their work shown at its highest quality. As much merit as it has, it also has a massive downside: No one can see anything else including the stuff other creators made!
      Not to mention, when I click on the item to view more details, the image size is STILL the same! Look, explore how every other marketplace works. Small thumbnails to give users a general idea on the item, and bigger ones for when the user is viewing the details of it. Second Life does it. Amazon does it. Walmart does it. This isn’t exactly a new concept, so I’m confused as to why this is such a problem. It’s doing a disservice to creators when they make something new only to be dethroned because they aren’t the latest item anymore.
      It isn’t like you can’t already do this. Check this out!


      This is the marketplace as viewed on my Amazon Kindle Fire HD10 (2017 edition). I can see SO many items, the Clara.io banner is gone, and it looks like… a marketplace! So why is it I have to view the marketplace outside of Interface to make it look like what it’s supposed to be?

    • Let free items be free
      Alright, let’s kick this debate again: if an item is on the marketplace place and is listed to be free, give the user the OPTION to not use their wallet. I really don’t understand why this is such a hard concept to grasp, but it has been one that has lingered honestly for too long. High Fidelity is already suffering from the walled garden that is the system requirements (VR or otherwise) and having a large learning curve, the moment the wallet comes into play, it’s typically the last straw for some people. This was never a problem in the past, so why is it still a thing?

      Okay, so lets say by some miracle that the wallet is made optional for items. The other side of the argument is then how do we encourage users to use a wallet, be it for free or paid items? Well, just tell the users the benefits of using it! If a user clicks No, just give me the item, show them a nag screen listing the benefits of using the wallet. I know some people will moan about it, but I feel that’s a good happy medium between what users want and the developers want. Eventually, if people feel they can trust the platform enough, they’ll set up the wallet and be more active on the marketplace itself, which I feel is better than simply forcing down people’s throats.

  2. Return the help button
    Okay, this one will turn some heads, but hear me out: a big problem I can see already is that even if a user goes through the tutorial, if they have additional questions or need a refresher, they need to overcome the midboss called the menu system. Assuming the user is in VR, the menu isn’t exactly the first place one person would go for help. Even then, I do like the idea of the help button being gone.

    Happy medium powers, activate!

    Keep the help button as the first listed item on the tablet or hotbar, but in the Help menu, add the option to hide the help button. Experienced users can hide a button they’ll most likely never use and new users who feel they can take off the training wheels can have the option to do so. Until that point, the button is easily accessible and easy to find.

    While on the topic of help, there is the Help domain. One thing I find absolutely mind boggling is why doesn’t the help button have a button to go to Help? Instead, we have two contact me buttons that bring up Outlook, which I never have set up and is even more useless while in VR. I’m not saying to remove the contact me button, but keep one and make the other just a shortcut that takes the user to Help. You have greeters who can help new people out, so use them!

  3. Ease of use, this ain’t (aka, information overload)
    How many people have tried Eve Online? Elite Dangerous? Kerbal Space Program? Other than all being space games, they also have a lot of depth (though the first two more so than the last). They also have a lot of options, buttons, shortcuts, do-dads, buttons, things to do, buttons, and buttons. Did I mention buttons?

    The UI on each of them is pretty overwhelming at first and can scare off a good chunk of new people, Elite Dangerous especially. However, none of them are bad games: they’re just hard and complicated, with a great deal of depth that rewards you for learning the system. Being a space pioneer? Pretty hard. Piloting a space ship? Pretty hard. Building one in KSP? Fun… but also can be pretty hard.

    What I’m getting at is that when it comes to High Fidelity, it wants to be the place where everyone goes yet has the same complexity as the three examples in the UI department. A user opens up their menu (or looks at their tablet) and immediately have a million questions before they even find out what button to press to move forward. In fact, the concept of the bubble can be a bit like first timers in Elite Dangerous when looking for the landing gears.

    Now I know this is also a tough thing to work out with, since Second Life did try this in the past with the ‘Simple UI’ which failed badly because it oversimplified things and left out access to important things (cough cough voice wheeze). High Fidelity, however, could benefit from a ‘Simple UI’ concept, but how that would be done is sadly beyond me. I thought about maybe restricting it to 6 or 8 buttons that floated in front of the user like Mute Audio, Safety Bubble (Bubble), Go To Location (GoTo), People Near Me (People), Snapshots (Snap), My Avatar (Avatar), and MAYBE the Marketplace/Wallet. You could even have a side button just to bring up the tablet so a user could disable Simple Mode once they are fine with everything. Things like Create are interesting, but a new user coming in would only be able to make things in Maker and that’s honestly it, so having it not initially visible would be okay in this simple mode. Of course, in normal mode, it’s the old UI as always, so Create would always be around.

    The ability to transition from ‘Simple Mode’ to ‘Normal Mode’ should be simple as well as to not annoy regular users. I’m not going to touch controls since that’s a whole other argument in itself, but the same could apply there as well.

  4. Hosting Content is still a Nightmare
    Thankfully, I’ve gotten around this with SSH and shell script wizardry, but that’s because I have my own server already set up and running smoother than butter. When people come in who are creators, or even are from VRC, getting their content in world is just a migraine waiting for both the person explaining and the person following the instructions. I can remember past discussions about allowing the ability to have a public ATP system (aka, an ATP server that can be accessed by other domains), which I would agree could help out a lot, but unless a user already has a server up, it won’t work.

    Making a 3D model appear in High Fidelity is like compiling a program at this point, in that you make the model, wait for it to upload to your favorite web service, bring it in world, cry because you forgot to include the textures in the FBX, re-export, wait for another upload (now even longer), bring it in world, then cry because you didn’t get the material settings right again. It doesn’t make sense why I have to have a separate program and service just to upload content into High Fidelity for use in High Fidelity that isn’t a world component. Again, I’m lucky I got my setup as simple as possible so that uploading my avatar’s new pair of pants is as easy as drag and drop in Explorer, but this should be looked into somehow.

    While not the most graceful solution, I did look into using the Digital Ocean servers for file hosting and… well, shocker: it’s easy. I could totally see just including on future Digital Ocean cloud servers a simple file server to host their own public content, which would mean a person would still need a server, but at this point, that’s the best I can think of for now. I know there was an offer to use the marketplace in draft mode as another means until otherwise stated, but not everyone will know that and it still has the same issues as previously mentioned.

    Another solution could be to have a temporary hosting system local to the system itself and have it be client side only. If the asset is an avatar, have everyone else see something else. This way, a user could test an item without even needing to own a server, and focus on only uploading content when they are ready to do so.

A major reason I’m not touching graphics is mostly because that is mostly up to the artist’s capabilities and style. When I make props, they aren’t the most elaborate and fancy item in the world, because I am not a 3D artist. When I had my avatar created, it was created to the capacity and the style of the artist I hired. Things like AA, glow, HDR, bloom, and other post processing things can only polish things so much, but if what to polish is already in a style that isn’t desirable, then it can’t make it look any more desirable. Having a strong foundation where users can actually move around and do things I feel is a bit more important, which some improvements to graphics being done on the side to compliment the improvements of the user experience.

If I had to pick from the common list of graphical improvements, I’d pick LOD control and Occlusion over bloom/HDR any day of the week, especially with the push for an Android build and because some avatars, namely the 3D scanned ones, are over 200k poly and a large majority of it isn’t even needed. I’m not here to brag (after all, I didn’t make the avatar), but I’m resting at 30k and looking just as well, and even then, I’d love to be able to properly decimate certain aspects and features so that users who are either experiencing slow down or because I’m 1km away don’t render unneeded stuff. At this time, High Fidelity feels I should lose my eyes first rather than the dew claws on the back of my legs or the claws even on my hands. Some say the eyes are the window to the soul, and it does suck that the system thinks the window should be obliterated first.

As for Occlusion… well that’s because I could see that being a way to massively improve rendering for some users in certain developed domains.

.

Now don’t get me wrong; I’m not here to be the Negative Nancy and I don’t think anyone in this conversation does, either. High Fidelity has been the primary reason I took Javascript lessons and started using the mystical divinity known as VIsual Studio Code and ESLint. From 2015 when you had to compile your own code just to log in to now, there have been many major improvements:

  • KTX/Draco support (and introduction of the easy bake oven)
  • Multi threaded downloading (this one was a major improvement)
  • Body Tracking via HTC’s pucks
  • Improvements to the 3D audio
  • Ambisonic introduction (though I wish some better explanations were given on how we can really use this)
  • Shader introduction
  • Worklist introuction
  • Voice recognition (short lived via Limitless, but was still an interesting thing)
  • Clara.io integration (though as mentioned, I don’t really care much for it. Despite that, I will still give it credit)
  • Improvements to the audio server handling (I think this was just after the 100 user test if I remember)
  • Introduction of the secondary camera system
  • Blockchain integration
  • Near-field audio support
  • Actual metallic PBR support

To say nothing hasn’t happened over the course of High Fidelity’s operation would be ignore some of the listed above achievements. Granted, we also have to acknowledge the elephant in the room: High Fidelity isn’t the only one running the VR world marathon. VR Chat, Sansar, and others are equally in the same area and that does make it easy to notice the differences.

Cheers!


#13

Well… For that, that would need that we can focus on objects. (Which is note really possible now in HMD as we don’t have vision tracking)
it’s not even possible in Desktop too. And in Desktop, that would be mainly for photography…


#14

SL has had it since 2010 I think it could work fine without gaze tracking… http://wiki.secondlife.com/wiki/Depth_of_field


#15

its horrible tho, do u have it turned on?


#16

In SL, It works with the focus on the object that the camera is pointing. (As in real)
In HF, there is no way to do that right now.
Without a point to focus on, that become pointless. (wow, I wrote that)
Also in our day to day, as a human being, we are to much conscient of the DoF since we always looking on what we are focus on. (We see it when we watch a flat representation that allow us to see the phenomenon.)
In SL, this is definitely a nuisance (except for photography and video)… but well some people seems to like it… (So much that there is an abusive usage of that.)
That said, that would be a plus value in HDM if we could have the eyes tracking… we won’t be too much conscient of it in HDM , no more than in the real world… but the view would probably look closer of the real sight.
For these reasons… I would put that low in the list.


#17

Ultimately we need some way to control how the objects appear at a distance… this is the key to achieving more realistic graphics. The fog/haze is not good enough, in combination with poorly rendered skyboxes, just can’t get away from the feeling that your plopped in the middle of a cartoonland.


#18

I don’t mind it in SL, adds to the overall look for me. It may not be for everyone, but what is concerning is that no-one gets any post processing effects in Hifi of any kind.


#19

quotes myself cos my ego is that big

april 2014
i stopped holding my breath for a sudden simple interface people can understand
maybe ill go down the can we make it harder to use and keep user numbers down route


#20

I totally understand this point, a lot comes down to the artist, but…

I think it’s very very important to be able to create beautiful looking screenshots and video clips that have that ‘wow’ factor. Since it’s a newly developed platform why are are being asked to have graphics that look like pre-2010 SL.

As an artist, I want to use the best tools… Hifi is like trying to paint the Mona Lisa with a toothbrush and a bowl of soup. I get enjoyment from building something and then thinking ‘ohh that looks cool’ , just doesn’t happen very often here…