Over the last few days there was a bit of a lively discussion on the sldev mailing list, caused by a Linden dude who, when inviting people to test a high end experimental feature, wrote some posts with some weird overtones about graphics cards. The issue is discussed here on the forum where Kitty Barnett's and Matthew Dowd's posts pretty much reflect my ideas of the issue.
However, the discussion went on a tangent as it aways does, about how Teh Laboratory handles additions which are meant to be optional and how they affect those who want to opt out of those. Atmospheric rendering came up, being pretty close to the topic discussed and how Linden Lab says that if you disable the blissful sky, the 1.19.1 series of the viewer should run better than previous versions, while many people say it runs worse (which is my opinion too). Kitty Barnett made comparisons which were indicating the latter. I made some quick tests here with similar results.
Now today comes an interesting comment from the Lab which is interesting. What it essentially says is, that except for the lowest end, in WL there are no settings which exactly match the older viewers. Of course you can turn off the sky, but a bunch of new features come in a packet (Preferences, Graphics, Custom, Basic Shaders), which will either hit your performance or which will leave you with noticeably less appealing visuals. I.e. with mid-range machines, you can trade off graphics or trade off performance, but there is no way in the middle as with old viewers. (If you are getting funny results with Eye Candy, even while you think you have similar settings, the Basic Shaders option may be the key).
This may be a pesky little technical detail, but I found myself being irritated by the Linden's claim that Windlight was optional and could easily be turned off to give you an experience as before and the fact that whatever I did, I either lost visuals or framerates.
This not-so-optional new feature is not unlike what happened when voice was introduced. You could of course turn off voice, but you were then left with questionable changes to the user interface and with grayed voice-related buttons scattered around the windows.
Humans learn from experience. The lessons learned from Voice, Windlight and Dazzle, coupled with derogatory remarks about hardware introduced 3 years ago make it hardly surprising that people are getting nervous at the mere mentioning of a new experimental high-end feature or that those "optional" improvements are often considered downgrades.
Thursday, May 29, 2008
Subscribe to:
Post Comments (Atom)
10 comments:
Thanks for posting this. I parsed the comment on the dev list and was rather slammed for it.
http://dusanwriter.com/?p=535
I realize Grace had the good sense to point out to me the fallacy of calling SLDev a customer relations touch point, however I'm pleased to see that my anxieties are at least articulated elsewhere.
Thanks.
Have you seen the pictures Kitty put up in the forum thread? Wow.
I totally wouldn't mind development so long as it remains truly optional. I'd opt in!
I am using a comp with a VIA C7-D @ 1.5 GHz with an ATI Radeon X1300. I tried and tuned the Old School and Bleeding Edge versions of the Second Life Viewer, and I actually got lower speeds -- about four frames a second or so in a not-very-crowded area. Then I tried the Eye Candy and RC versions and started getting eight to fifteen frames a second in a not-very crowded area, and it runs a lot better for me...
*side note*
Hey Nicholaz any sage speculation what is the "very important announcement" posted on the $L blog Tuesday, June 3rd ppls are abuzz about?
Cristalle: I've seen the pictures and they are gorgeous. I'm just not going to try the patch, because I don't have the hardware to support it (I wouldn't mind buying a $100 card, but the upgrade would make my computer quite noisy, because currently I have a fanless video board and would require a bigger power supply (which is currently also noise reduced)).
Randeimos: This is indeed possible. The lowest settings in the new viewer are "lower" than in earlier viewers, so they are bound to give you better performance.
Foe: I've read it but I honestly have no idea. But with the typical overexcitement on their blog, this could be anything from "Robin recovered from a broken fingernail" to "IBM will buy Linden Lab".
"IBM will buy Linden Lab"
I wish. Then we'd have an asset server database that works reliably.
Just a suggestion...
Have the "overview" post linked at the top right of your blog, along with the download links? It seems like a nice succinct explanation, it'll make it easier for new folks to find if it's up with the other links.
good point alexandra! thanks for the suggestion
I have tested the Bleeding Edge and Eye Candy viewers extensively on two very different systems: a dual-core desktop with an NVidia 7600, and a laptop with Radeon XPress 200M integrated graphics.
On the desktop system, the EC viewer gives lower frame rates than the BE viewer if atmospheric shaders are on. (Turning off atmospheric shaders gets the frame rate above the BE viewer.) But both do well, and the extra visual prettiness of WindLight is nice.
On the laptop, on the other hand, the EC viewer gives HIGHER frame rates. True, the only way to get GOOD ones is to turn off all the eye candy... but doing that gets the frame rate over 10fps, which is double the performance of the old viewer. With basic shaders and local lighting on (kind of like the old viewer but with slightly less pretty skies), it's still well ahead in performance. And that's true even in areas with no avatars around, so it's not coming from avatar impostors.
So, as is so often true, Your Mileage May Vary. One thing I really like about the WindLight viewers is the much wider range of tuning possibilities; you can opt for minimal graphics (no shaders on, or just Bump Map and Shiny) for speed, or everything on for beauty. You can even switch to improve your experience in various situations.
Shirley: It seems that mileage does indeed vary. But the best thing over all I guess is that people now (compared to last year) have ample choice to select what works best for them.
Thanks for the report.
Post a Comment