Smoke Alpha channel messed up

[Dandel]

Another bug report against 6-30-2010 release on radeon graphics on linux.


The alpha on the smoke is incorrect... i'm getting huge blobs of white. (Not linked to shaders)... See my user gallery for a screen shot.

[joevenzon]

Good catch, that was a bug with the blending in the non-shader path. Fixed in R2834.

[nomoo]

rev 2836 or something near. I've got smoke with disabled shaders only without.

Also, i see strange behaviour of tcs it seems to be working all the time.

[NaN]

Looks like I broke a couple of things with the latest commits. I won't have time to fix them within the next couple of days, will roll back the changes I guess.

[NaN]

I've reverted to rev2835 and the smoke is still broken with shaders on. Can somebody confirm this? I haven't checked shaders for some time as I am developing on my laptop(no FBOs).

[portets]

confirmed here.

[joevenzon]

I was able to reproduce a distinct lack of smoke with shaders on (it's sometimes there but very very faint). Shaders off, the smoke looks fine. Is that what you guys are seeing too? I'll take a look at it more later today.

[joevenzon]

Fixed in data R663. I was accidentally applying alpha twice because I switched to pre-multiplied alpha.

[Lutz Mader]

Hi Sorry, NO.
I try to set shaders=off or schaders=false in the Display section of VDrift.config, but I get no smoke, it looks more like white walls.

Thanks for any help, idea to get 2010 working like 2009,
Lutz

p.s.
Mac OS X 10.6.4, MacBook C2D, GMA X3100.

VDrift 2009-06-15 works well.

[almos]

I don't think it is enough to fix this in svn, as most people download the latest release, which is still broken.

[NaN]

The last release was announced as a development snapshot. We should have added a big fat warning on startup I guess.

[almos]

You also should have added the big fat warning on the website. Neither the downloads box nor the release announcement mentions anything about development snapshots. They say "release", which means more-or-less tested code, which isn't the case, it seems. This is a perfect way of loosing users quickly. I'm sure lots of people who downloaded this "release", saw this very apparent bug (which is already known and fixed in svn for months now), and said "vdrift sucks", later also said "open source games suck" and "linux suck".

Now I checked the "about vdrift" page, and it really mentions in the end that "So far, every release has been a testing/development quality release. For this and other reasons, dates are used instead of version numbers. "

By the way my experience is that the longer a code is in a perpetual beta stage, the less likely it is to reach a stable release.

[NaN]

You are right. It should have been made more visible/explicit.

nor the release announcement mentions anything about development snapshots

http://vdrift.net/article.php?story=20100714200428239

... As a result this release is a snapshot during heavy development and so this release may be a little less stable than the previous in some areas...

[thelusiv]

almos, thanks for your concern about releases. We have been discussing that very issue lately. Perhaps you would like to read and comment on this thread:
http://vdrift.net/Forum/viewtopic.php?t=1455