Beta testing 2.10

[thebigh]

17 hours ago, stgatilov said:

And... how to reproduce it?
Does it happen on other missions?

I haven't noticed this problem in other missions. But for me it's easy to reproduce:

Play the mission on Linux, jump onto the little shack with the fire bucket, frob the window above you. It should at worst make the guard patrolling the courtyard stop and murmur.

Now play it on Windows and do the exact same thing- all the guards will freak out and run around with their swords drawn.

[nbohr1more]

Please open the console and invoke

con_noPrint 0

getviewpos 1

Then capture a screen-shot of the location.

Thank you!

 

 

[thebigh]

Is this what you need?

I captured this screenshot on Windows, after dropping the window and letting the guards go mental.

 

[nbohr1more]

Thank you!

I really need to get Windows going on one of my drives so I can evaluate this.

Does this happen regardless of graphics settings? ( Stencil vs Maps, SSAO, Bloom, MSAA, Render Scale ? )

( I doubt graphics are a factor but unfortunately the renderer touches the lightgem code and if anything spikes the render results then the AI will see you full bright. )

[thebigh]

19 minutes ago, nbohr1more said:

Does this happen regardless of graphics settings? ( Stencil vs Maps, SSAO, Bloom, MSAA, Render Scale ? )

I've tried it with a few different combinations of these and it always seems to happen. I also turned on notarget just to be 100% sure it's not related to vision, and the guards are just reacting to the sound of the window clattering on the ground. I'm reasonably certain it is a sound propagation effect.

[thebigh]

I think I might have tracked down the bug. I stupidly had a locationseparator with a negative sound_loss spawnarg- ie. it amplifies the sound.

I'm guessing the sound bounces around the map multiple times, passing through the megaphone repeatedly, getting amplified to crazy volumes. No wonder the guards were going berserk.

This patch fixes the problem on my Windows machine: https://forums.thedarkmod.com/index.php?/topic/21260-down-and-out-on-newford-road-by-thebigh-jan-26-2022-christmas-connections-contest-entry/&do=findComment&comment=470839

It's still not clear to me why this should be OS dependent though. Idiotic bug or not, it should behave the same on Linux as it does on Windows.

[nbohr1more]

OpenAL acts differently on Windows vs Linux. Thanks for the evaluation!

[stgatilov]

22 hours ago, Dragofer said:

Picking up some ragdolls, I noticed that they seem to oscillate slightly when they're in my grip. It's most noticeable with rats.

It is not easy to avoid: with a simple physics implementation, we either have oscillations or the item lagging behind.
I hope "slightly" is not a big problem... if it oscillates wildly, that's another thing

[stgatilov]

14 hours ago, thebigh said:

I think I might have tracked down the bug. I stupidly had a locationseparator with a negative sound_loss spawnarg- ie. it amplifies the sound.

I'm guessing the sound bounces around the map multiple times, passing through the megaphone repeatedly, getting amplified to crazy volumes. No wonder the guards were going berserk.

Is there any legitimate use for negative sound loss on a separator?

If no, then shouldn't we clamp the value to zero if it's negative?

[nbohr1more]

1 minute ago, stgatilov said:

Is there any legitimate use for negative sound loss on a separator?

If no, then shouldn't we clamp the value to zero if it's negative?

No legitimate usage that I can fathom, only mappers misunderstanding the concept of "reducing" volume using a positive number. IMHO we should probably just ignore the sign and reduce by the value amount.

Too bad sound manipulation isn't standardized in all defs as a range between 0.0 and 1.0 ... that would be too much sanity to ask idSoftware for...

[stgatilov]

52 minutes ago, nbohr1more said:

IMHO we should probably just ignore the sign and reduce by the value amount.

Clamping is better: that's out of range value, it should not have any meaningful effect

Quote

Too bad sound manipulation isn't standardized in all defs as a range between 0.0 and 1.0 ... that would be too much sanity to ask idSoftware for...

It is standardized to decibels, isn't it?
That's the most standard thing that you can get about volume.
 

[nbohr1more]

Interesting question.

To my view, config files are supposed to represent human readable offsets of the default audio files' baseline volume.

Doom 3 made volume offsets available that exceed "full volume" so that mappers could use quiet recordings and amplify them.

Allowing amplification means you are always at risk for clipping audio (distortion) where you move all frequencies up to the upper limit of volume range. Do we allow more volume amplification but risk either noisy audio ( noise floor is increased ) or clipping and just tell mappers "don't use poorly leveled recordings"... or do we lock down anything out of range?

A 0.0 to 1.0 range is just an abstraction that represents a percentage of a volume slider.

The complication here is that decibels are exponential and so is volume perception.

If we aren't internally converting to an exponential curve then mappers will encounter a strange disconnect where all of the useful volume adjustment is in the upper 20% of the ranges. As long as full integer values are properly mapped to the decibel curve in all configurations and mappers are made aware of the limits of the available range then all should be fine.

Unfortunately. (for example) we had one bug where people couldn't use the Ambient Volume slider because mappers choose huge amplification values: https://bugs.thedarkmod.com/view.php?id=4627

And if you review that wiki, the convention is always to use negative numbers since the assumption is that the 1.0 value is the full volume recording. The sound loss variable on location separators diverges from that design, so (of course) mappers are gonna see negative sound values in one place and apply them to other sound defs. So there is "standardization" for the actual volume representation but configuring modifiers for the volume values is inconsistent.

 

[thebigh]

I'm indifferent to whether it clamps the value or ignores the minus sign, just as long as it stops doing unexpected things.

[nbohr1more]

Yeah. I don't object to clamping the sound loss arg to 0.

I am just putting it out there in case anyone else ( mappers ) care about it.

Since we have a workaround, it probably doesn't count as a 2.10 blocker either.

[AluminumHaste]

Screenshot command is broken it seems. I noticed when running the screenshot command, that some shadows disappeared.

I also took the same screenshot with rivatuner's built in function and on the pillar in the middle you can see the shadow is missing.

Native screenshot;

Rivatuner screenshot

[Oktokolo]

On 2/20/2022 at 7:29 AM, stgatilov said:

Try to delete tdm_installer.ini?
The only possible explanation is some kind of permission issue.

Thanks, Worked. The installer first failed with ini not found. But after restarting it, it worked and found beta210_8.

Can't have been a permission issue - TDM is installed in an ordinary c:\games folder not protected by the system like all my games. As game devs are ultra-ignorant when it comes to adhering to any platform-specific standards even on the mainstream desktop game launcher OS, i don't install games into c:\program files...

[nbohr1more]

19 hours ago, AluminumHaste said:

Screenshot command is broken it seems. I noticed when running the screenshot command, that some shadows disappeared.

I also took the same screenshot with rivatuner's built in function and on the pillar in the middle you can see the shadow is missing.

Native screenshot;

Rivatuner screenshot

This is with the standard screenshot command that is mapped to F12 ?

Does this happen for both shadow modes ?

[AluminumHaste]

Well I bound F10 to screenshot. I forgot to check if this is just amd issue again or if it happens with stencil shadows also.

I'll check later

[AluminumHaste]

I also tried this on my other computer with the 980ti, and it only happens with shadowmaps. Stencil shadows work fine.

It's object shadows it appears. If you load up the bakery job, the shadows on the open window disappear when taking a screenshot with shadow maps, but they remain with stencil shadows.

Stencil Shadows:

Shadow Maps:

This is a clean Darkmod.cfg.

 

Darkmod.cfg

[AluminumHaste]

It appears that all the things missing shadows are func_statics.

The barrels, open windows, the frame around the door on the left, the awning on the right, the wooden beam above the hanging torch, all func_statics.

[Apiai]

Hello, 

I've tested beta 2.10b8 32bits version :

- TDM is loading more fast (just TDM i'm not talking about mission loading)
- New grab object animation is really nice (is it V2.10 new feature ?)
- I've translated news strings into french. Contrary to what is said in all.lang, french accented letters were well displayed only when I converted the file to ANSI and not to UTF-8 !?
- One little bug : If you enter HUD Settings menu, '(click for more details)' is still displayed when you go back to the main settings menu

- In Deadeye, one guard shadow is missing but I don't know if it's related to V2.10 : https://ibb.co/48GBffb

 

 

 

Edited March 1, 2022 by Apiai

[DeTeEff]

Mission is loading faster, but DMAP freezes on my VIP. It's a giant mission so I don't know how I can fix this. Where to start so to speak...I have tested with a smaller mission and that DMAPs fine... Perhaps back to 2.09 for me then...

 

It comes to

patches
entities etc
######### entity 0 #########
-- ( worldspawn: )
0.2 seconds daceBsp

 

then it freezes...

Edited March 1, 2022 by DeTeEff

[datiswous]

5 hours ago, Apiai said:

- In Noble affairs, one guard shadow is missing but I don't know if it's related to V2.10 : https://ibb.co/48GBffb

Are you sure this isn't Deadeye?

[Apiai]

Yes you're right 

[Dragofer]

27 minutes ago, DeTeEff said:

Mission is loading faster, but DMAP freezes on my VIP. It's a giant mission so I don't know how I can fix this. Where to start so to speak...I have tested with a smaller mission and that DMAPs fine... Perhaps back to 2.09 for me then...

 

It comes to

patches
entities etc
######### entity 0 #########
-- ( worldspawn: )
0.2 seconds daceBsp

 

then it freezes...

DMAP usually freezes for me too, but in most cases it gets done eventually.