HDR in Secondlife?

Actually you can make your own shaders and use them in Second Life. Check out these sweet lens flares.

Well, i'll have to disagree with it. The hardware compatibility is still an issue:

http://www.steampowered.com/status/survey.html

note, 35% or so of these who responded to the survey has computers only able to support directX 7-8 level of graphics. This is quite a significant amount and in case of SL even greater since LL picked GLSL 2 for their new shaders, for which the driver support is flakey at best. And as far as patching things goes... you only need to remember recent Battlefield 2 drama to see how well the companies resolve through patches the issues they have no intention to solve in the first place. And you only need to ask around developer boards to see how fast (not) card manufacturers can be when it comes to implementation of standards and to patching errors in their drivers :<

This steam survey is kinda skewed, for example I know for a fact that 16:10 is the standard widescreen format for desktop wide screens (as opposed to widescreen tv's). And they are only listing 16:9 there.

Then if you go down to the resolutions given, you don't see any 16:9's instead you see 16:10's. So there is some diusperancy here, either through user error or the survey error.

The simple answer is the users with old/crap hardware are limited to surfing 2d web sites and filling in online surveys while the rest of us are marveling at the latest ultra real shaders...

Of course the survey results are skewed.

It's true that the world is still littered with ISA buses, old-style PCI videocards and even 10Mbit NIC cards. This shouldn't stop progress.

Of the 35% you cite, approximately 10% of the group (the ones defaulting to DirectX 7) are running cards not designed for 3D (the NVidia MX series, a bunch of on-board chips and some other oldies-but-goodies). The other 25% fall into a DirectX 8 path which does support shaders, real bump-mapping and more.

While Valve has gone out of their way to be inclusive with their engine, and I certainly wouldn't begrudge Linden Lab for doing the same, there is no inherent benefit in doing so. People may hem and haw at the thought of upgrading their hardware, but there are sub-$50 cards on the various DirectX 9 lists, and such costs to run "modern" 3D applications should be accepted.

As others have already pointed out, this is a skewed survey which ended in April 2006 (well before the current high-end ATI vs. NVidia price war which lowered costs dramatically) and doesn't even cover a fraction of the sales of Source-engine products.

Changes are right around the corner in another way, with Windows Vista and DirectX 10 breaking compatibility. Microsoft is biting the bullet to support future growth. Although different circumstances, Linden Lab should make the same decision if they must (as in quality sliders being too difficult) and at least start playing catch-up with rendering technology from years ago.

I think the catch here may be, your idea of progress (high-end graphics few percent of computer users worldwide are able to run comfortably) is likely to vary wildly from LL idea of SL progress (SL hitting WoW-like user numbers and being accessible with minimal effort required, and becoming "3d 'net"

The mention of WoW isn't accidental here -- i strongly believe WoW is actually a case where extreme popularity is to large degree caused by *low* hardware requirements -- the fact their engine can run comfortably on nearly toaster level hardware significantly lowers the entry barriers, making it easy to reach the userbase counted in millions. On the other end of spectrum you have the "we code for the future and screw everyone not on the bleeding edge" way of thinking, shown by makers of game engines like that for EQ2 and suchlike... which leaving aside all the differences in gameplay *does* reflect in amount of their subscribers.


Again, it'll depend on what you actually consider to be beneficial. 35% of playerbase *is* considerable amount... and while yes, it may be possible to upgrade one's computer for $50 or so, at the end of day it's still someone else's money you tell them how to spend... while it's up to person in question to decide what kind of spending can benefit them the most. And while it can be tempting to say "well then, screw them because we can't be arsed to code engine flexible enough so it's able to cover multiple render paths and fall back gracefully when necessary" ... what ultimately gets screwed is the number of players perusing your application. If this can be seen as beneficial end-effect or not, depends again on what goal you have for it. But generally, lesser number of players rarely 'means more' and isn't perceived as success. No matter how much eye candy your game can draw.


No, Microsoft is biting the bullet in a gamble to force people to 'upgrade' from OS that already meets their needs. Don't kid yourself it has anything to do with "support of future growth", they simply painted themselves into corner where they cannot make a product significantly and noticeably better from what they sell already, so the only option they see is to make sure new 3rd party software only works with their latest product.

The thing is, just as all XBox developers know exactly what hardware their games will be running on, all game developers have total control over what goes into their games and what the players are allowed to do. Hence they have total control over how well the game performs on older, limited hardware simply by limiting the content within the game.

LindenLab do not have this luxury. Regardless of how good their data compression or code efficiency, WE are the ones who actually create the content and it is OUR fault if a server lags or we're only getting 2fps. There will always be people that simply do not understand that when they create a virtual office desk and put a virtual photo of their pet dog on it in a fancy picture frame, the photo does NOT need to be at native camera resolution of 6 megapixels...

your ideas intrigue me and i wish to subscribe to your newsletter.

...for serious. tell me more that's cool!

It all looks like photoshop to me. When you look at the rest of Torley's pics in her archive you'll see a bunch of normal looking pics. Though if you want SL to look noticebly better with a slight performance hit try cranking up your FSAA (mines set to 8xS now), Anisotropic Filtering (mines set to 8x), and enable Transparancy AA SuperSampling if you got it (available to 7 series Geforce cards, available for series 6 if you're using the 91.45 or 91.47 drivers like I am). Alot of Geforce cards like to be pushed so you may not see any performance hit though some actually see gains from pushing their Geforce cards. Dunno about ATI though.

Yes, WoW's requirements are low, but maybe that's just because their client is better programmed. They require DirectX 9.0c, which supports all of the features we are talking about here. Your previous post pointed to the Source Engine poll that included support for DirectX 7 and 8 level technology. While Second Life uses OpenGL, are you saying now that it would be okay with you to extend Second Life to support the technologies that WoW supports by requiring DirectX 9?



Second Life is ultimately only about Eye Candy and the gameplay is nil. Without the tools to build proper gameplay experiences, it is just a glorified chat and 3D construction set, therefore the more "bleeding edge" support that is added the prettier it will look desite its vaccuous underbelly.

I would love HDR, but it should come after many other technologies. I'd prefer the server environment be upgraded before the client, as in a modern physics system (and its better performance that would affect all of Second Life) and a respectable programming interface (be it Mono or something else, just something that could perform properly and was implemented like modern language). I'd also love changes in the handling of prims, such as object classes and other tools that would make implementing gaming system possible without reigning in the open creativity allowed now.



I don't know if this coming from an anti-Microsoft mindset or not, but it is not the reason for the change coming with DirectX 10. Rather than boring most everyone, I'd suggest you research the important reasons they made this decision after a decade of advancement and compatibility.

In the end, I'd love to see what the talented builders in Second Life could pull off with true bump mapping, normal maps, shaders, terrain painting and more. I hope that hardware choices aren't the reason we haven't gotten these technologies, though from some of the posts that appears to be the case.

While WoW requires directX 9 to run, the video card to run it is listed as 'GeForce 2 or better'. This means their engine can utilize very simple render paths on systems with outdated graphics, rather than _require_ state-of-the-art video just to run. Which was the case with updated SL client.

I believe the requirement for dX9 runtime is merely because dX9 was used at compilation, and this is what game client code is linked with, for simplicity sake.


I think for plenty of people it's rather a social 'meeting place' and --quite literally-- a sandbox to exercise their creativity. If it was indeed all about eye candy, it'd be long dead because the graphics it offers are completely out of date. So whatever it is that makes people hang around, graphics it ain't.

There's interesting catch here, by the way. If it's ability to 'make their own things' that makes people stay and play in SL, then significant facelift of its graphics abilities can be counter-productive -- higher level graphics of modern games can take considerably more time to prepare because with increase of abilities, so do the expectations and demand of your audience. Which means, at some point the bar for developing content could be set too high for your 'casual creator'. And while i'd personally love getting access to even such basic things as glow/gloss/normal maps, at the same time i can imagine how it could result in either further increase of stress on the SL client/servers (3-4x as many texture data to handle) ... or in some people giving up in frustration due to 'no longer being able to cut it'. Or perhaps, in both.


I have absolutely no reason to have anti-MS mindset, so it's certainly not the case. But i looked into what both dX10 and Vista does, and there's no good *technical* justification there for this kind of move. So i have to go with simplest explanation, which is creation of incentive for people to switch, in situation where there's even less reasons for it than there was to switch to Xp. Ockham's razor and all that.

Last night I installed Catalyst 6.3 (the last version ATI supported smartshaders) and tried to do some HDRish shader rendering in SL. It looked pretty nice, uploaded one pic to snapzilla as well. But alas when you take a photo SL strips the nice custom shaders out of it.

Due to lack of preformance on those drivers I had to go back to 6.8, but that's one way to get HDR or some custom shading if your so inclined and have a relatively modern ATI card.

The next place to tackle (which I feel is the proper place) is the shader folder in SL program directory, these have glsl shaders, I've messed around with them a bit, and tried to replace one of the shaders with a free HDR shader I found on Ogre site. I did not get any errors when this was done, the custom shader loaded properly. But once again, the SL glsl implimentation is lacking and does not fully support evertying. Regardless I got some funky rendering through that and another cartoon type glsl shader I found elsewhere on google.

You are absolutely correct, I only checked into it because from what I remember during my WoW research (I wrote a lengthy paper on larger MMO environments include Star Wars Galaxies and a few others) was like Second Life (and most game applications) you had performance options allowing the environment to support a breadth of features yet scle back when necessary. Earlier in this thread Michi mentioned (and other commented) that they removed some features as not being ready for prime time. Yet I read daily about people who crash continuously and "live with it" while I have run the Second Life client for days at a time (once six days continuously) with nary a crash. Mt only continuing comment is that if they couldn't solve the downscale problem then, was that the reason for the removal, or was it truly that they weren't ready ad needed to deplot 1.9 immediately. Since that version was in place before my joining, and the only render-related changes since were local lighting and flexible prims, was the progress completely scrapped, back-burnered, or what?



It depends on implementation. Every suggestion I made publically (and those I'm still trying to solidify) has been with the intention of extending the environment and not replacing any proven tools. I undertsand why we have prims instead of splines and meshes: they are easier building blocks to learn from, and allow instant visual gratification when building. Some features such as reflective surfaces and true bump-mapping could be added with no modification to the building metaphor. Others would require changes, or alternative tools. Much as Adobe's slow addition of vector tools to Photoshop, there's no reason to believe that those comfortable with their options wouldn't slowly learn new tools and back off tortured prims to achieve less performance-crushing options available with meshes. I would hope that the current tools could be extended but remain as easy to use, and new tools could be introduced to advance the state of the art as it were, in Second Life terms.



I guess we'll have to agree to disagree. The root reason, that to deliver the new performance meant that had to remove compatibility, makes perfect sense. They will continue to upport old DirectX through emulation, but the binaries were rewritten to leverage current technology (PCIe) and to take advantage of the bus and memory changes they had two choices: maintain a less efficient multi-path layer or optimize and target current technology, drawing a line in the sand to start a new future path. I am happy they made this choice for once, since they debated it back during the PCI to AGP transition and stuck with less-optimal solution.

I'd figure it's been shelved until fully openGL 2-compliant cards become more widespread, or until the SL client drawing code gets rewritten so it can optionally utilize these new shaders where supported... neither feels like something to happen in the near future, though :/


I don't quite understand this. It only makes sense if what you say is a) dX10 relies on PCIe functionality in order to run and b) this functionality is not possible to provide on systems like Xp ... neither being true, from what i know. (especially when you consider performance and functionality of agp and pcie matches all way up to 8x speed, so 'higher performance' isn't really argument to use, unless you also restrict Vista to 16x pcie systems which quite obviously isn't happening -- people run current Vista builds on agp based systems fine, dx-based aero bells and whistles et al. So any such restriction doesn't happen for technological reasons)

I 've been using static and scripted hardware shaders in sl for two months they are great ,its allowed me to prosper somewhat. I am very thankful for shader 2.

Just out of curiousity, how are you guys adding shaders to SL?

mmmmm.......secret?

Umm I guess I will need to buy a 16 core computer soon to turn that thing on.

Like a couple of posts above it's described how it's done. I guess now you should pay me for reading up threads and informing you of that.