[Fox]

It's just a tile 4429 above another with cstat 4. May you have confused it with tile 913?

[Fox]

Some of the animations displayed on the screen are too fast depending of the framerate, for example the moving arms after being hit by the Shrinker. In the GIF above it is faster than it should be in the vanilla game, despite it still being slowed down by the recording software.



This also occur with the Ripper or Freezethrower "shaking" while firing. I guess this is because of the code being too primitive and was written under the assumption that a PC would never be capable of more than 30 frames per second...

I suppose one solution would be to have it working with totalclock divided by 4 or something like that.

[Kyle873]

 Fox, on 08 March 2013 - 02:01 PM, said:

Some of the animations displayed on the screen are too fast depending of the framerate, for example the moving arms after being hit by the Shrinker. In the GIF above it is faster than it should be in the vanilla game, despite it still being slowed down by the recording software.



This also occur with the Ripper or Freezethrower "shaking" while firing. I guess this is because of the code being too primitive and was written under the assumption that a PC would never be capable of more than 30 frames per second...

I suppose one solution would be to have it working with totalclock divided by 4 or something like that.

Dividing totalclock by 4 is only a stopgap. What should probably be done is to base the animation on something which increases at the same speed regardless of the current FPS, such as the player_par in the player struct. (As an example)

[Fox]

I specifically mentioned totalclock because it is unsynchronized with the game.

[Gambini]

 Fox, on 02 March 2013 - 05:21 AM, said:

Is it possible to fix the bug which sprites offset doesn't flip with it in Polymodes?

bump

[Plagman]

 Plagman, on 02 March 2013 - 09:27 AM, said:

Can you post this map?

bump

[LeoD]

Just in case it is easy to fix/implement: the pitch and roll maphack commands don't work on voxel models (but angoff does), so they can't be rotated to fit on sloped floors.

[Gambini]

Here. I hope you dont mind it´s not the same map but I have no access to Fox´s HD.

Notice it happens only with viewalligned sprites.

[Gambini]

DOUBLE POST, THE ATTACHMENT WENT AWAY IN THE PREVIOUS. SORRY

Attached File(s)

•  offset´s offset.rar (225bytes)
  Number of downloads: 198

[Fox]

 NY00123, on 15 February 2012 - 08:53 AM, said:

Well, checking out it a bit, I've spotted a mention of dethtoll.mid in the source for the case of failure.
A test I've done with a modified USER.CON file has then revealed what I've just guessed:

If, upon loading a user map other than E1L8.MAP, the game fails to locate a MID or OGG file of the same name as the map's, not only it loads a "vanilla" MID; It always loads dethtoll.mid.
So, even if you have a modified USER.CON with the track for E1L8 set to something other than dethtoll.mid, it'll still playback dethtoll.mid in this very specific case (which is hardcoded).
The alternate track defined in a modified USER.CON should be played back properly as long as you don't load a user map other than E1L8.MAP.

After looking a bit at the code, I'd say that it may take a little more to fix than what one may think... maybe not much, though.

Bump.

I still stand that Eduke32 shouldn't be using E1L8 for the usermap, which is a fucked up hack. Right now you can't define a eighth leve for the first episode without returning a bunch of errors.

[Hendricks266]

^ This has been on my to-do list but as part of rewriting how the game stores music file information, including saving the current selection in savegames.

Spring break is in a week so you might see some stuff from me then.

[Jblade]

 Hendricks266, on 11 March 2013 - 02:04 PM, said:

^ This has been on my to-do list but as part of rewriting how the game stores music file information, including saving the current selection in savegames

Spring break is in a week so you might see some stuff from me then.

Cool! I assume this'll also mean that the track won't restart from the beginning when you load the game? There is a workaround I've used with the EVENT_LOADGAME that works great for playing a changed music track but it still restarts the music track when you load a game. Having a built-in solution always seemed like the best way.

[Fox]

I am having some trouble with making a mod compatible with HRP. It seems that the duke3d.def file from the custom content directory doesn't override the one in the main directory. So it requires the file from the main directory to be deleted first...

Another problem comes from the setuptile command which affect tiles defined after it. For example, I am using the defs from the HRP and a certain mod which contains new sprites for EXPLOSION2; because the HRP use setuptile to resize EXPLOSION2, the new sprites will also be resized.

One satisfactory solution would be undeftexture also clean up the setuptile for the specified tile. And I suppose it also makes sense.

[LeoD]

 Fox, on 15 March 2013 - 06:04 PM, said:

I am having some trouble with making a mod compatible with HRP. It seems that the duke3d.def file from the custom content directory doesn't override the one in the main directory. So it requires the file from the main directory to be deleted first...

You could use

setdefname my_mod_toplevel.def

to override any duke3d.def in your toplevel CON file. (Well, if you have one. You might get into the same trouble with more than one EDUKE.CON available however)

[Fox]

Hmmm, interesting. I will check it out, thanks!

===//===//===

I still think Eduke32 should not correct the aspect ratio for 8 x 5 resolution. Otherwise I don't see the point of having an option for 320 x 200 resolution.

[Hendricks266]

 LeoD, on 09 March 2013 - 11:09 AM, said:

Just in case it is easy to fix/implement: the pitch and roll maphack commands don't work on voxel models (but angoff does), so they can't be rotated to fit on sloped floors.

I tried looking into this but was repelled by the unreadability of the code in polymost.c and mdsprite.c.

[Mblackwell]

Make pausing or entering a menu (or otherwise interrupting play) only pause sounds rather than stop them.

Thanks.

[Fox]

Eehh, I think that would require some full rewriting of the sound system.

[Mblackwell]

No, it wouldn't.

[Hendricks266]

Have you ever looked at ASS (Apogee Sound System), now incarnated as jfaudiolib?

The thing is quite frustrating. It's a holdover from at least the days of ROTT, possibly earlier. Vorbis and FLAC support is bolted on in a way that you can tell the system was designed for lower-level formats. Don't forget that CON "volume control" is just a hack using the sound distance.

There is definitely no way to pause sounds using its API, at least at present. Perhaps there could be a way to freeze the mixing buffer, but the fact that the menu also makes sounds give the problem a new dimension.

As a novice programmer, I've already had nightmares trying to separate Vorbis and FLAC background music into a separate, second buffer, because with just one the music reverberates along with sounds in sectors with echo MUSICANDSFX sprites.

[Mblackwell]

 Hendricks266, on 18 March 2013 - 10:48 PM, said:

Have you ever looked at ASS (Apogee Sound System), now incarnated as jfaudiolib?

The thing is quite frustrating. It's a holdover from at least the days of ROTT, possibly earlier. Vorbis and FLAC support is bolted on in a way that you can tell the system was designed for lower-level formats. Don't forget that CON "volume control" is just a hack using the sound distance.

There is definitely no way to pause sounds using its API, at least at present. Perhaps there could be a way to freeze the mixing buffer, but the fact that the menu also makes sounds give the problem a new dimension.

As a novice programmer, I've already had nightmares trying to separate Vorbis and FLAC background music into a separate, second buffer, because with just one the music reverberates along with sounds in sectors with echo MUSICANDSFX sprites.

TerminX says it can be done; I believe him.

[Hendricks266]

It can be done: implement multiple parallel buffers with pause controls into the sound system and expose the functionality to the game using new API.

No one said it would be easy.

[TerminX]

Hendricks266, see FX_PauseVoice()/MV_PauseVoice().

[Helixhorned]

 LeoD, on 09 March 2013 - 11:09 AM, said:

Just in case it is easy to fix/implement: the pitch and roll maphack commands don't work on voxel models (but angoff does), so they can't be rotated to fit on sloped floors.

Since voxels are mostly a classic feature and there's no way for rotation around all 3 axes there, I think it would be wrong to implement these for the Polymodes only. Similarly, voxels don't support x/y flipping ([F] key in Mapster32, bits 4 and 8) or floor orientation, so even if those transformations could be coded in into the GL modes, making them look differently than classic seems only an increase in trouble.

 Fox, on 15 March 2013 - 06:43 PM, said:

I still think Eduke32 should not correct the aspect ratio for 8 x 5 resolution. Otherwise I don't see the point of having an option for 320 x 200 resolution.

With r_usenewaspect 0 and the -rotatesprite-no-widescreen command line option, everything should look like Duke3D 1.5. Otherwise, we assume that screen pixels are square (there's a correction for non-square ones actually coded, but almost never tested) and want squares in the world to be square on the screen.

[Spirrwell]

Sorry if this has already been addressed, but...

GRP file case sensitivity. I just realized that under Ubuntu 12.10 that the GRP file is case sensitive even though I can see it under the Game tab. I had to make all the the letters capitalized and mapster32 just crashes on startup if you don't have the case right.

[LeoD]

 Spirrwell, on 22 March 2013 - 12:49 PM, said:

GRP file case sensitivity. I just realized that under Ubuntu 12.10 that the GRP file is case sensitive even though I can see it under the Game tab. I had to make all the the letters capitalized and mapster32 just crashes on startup if you don't have the case right.

Interesting. This page would imply just the opposite case. No case issues on Debian64 Wheezy btw. (self-compiled executables).

[Spirrwell]

 LeoD, on 22 March 2013 - 01:28 PM, said:

Interesting. This page would imply just the opposite case. No case issues on Debian64 Wheezy btw. (self-compiled executables).

I just re-tested it, it works either way. As in either all lowercase or all uppercase.

I think it would be nice for it to not be case sensitive, Although it's so minor... And I really don't know if I want to dive into the EDuke32 source code again. The damn thing is massive!

[Hendricks266]

 Helixhorned, on 22 March 2013 - 02:50 AM, said:

Since voxels are mostly a classic feature and there's no way for rotation around all 3 axes there, I think it would be wrong to implement these for the Polymodes only. Similarly, voxels don't support x/y flipping ([F] key in Mapster32, bits 4 and 8) or floor orientation, so even if those transformations could be coded in into the GL modes, making them look differently than classic seems only an increase in trouble.

I'm not a believer in the caboose model of doing things (everyone has to move as fast as the slowest or last one). I already added support for the alpha member to voxels in Polymost. (Alpha works for 8-bit tiles in both Polymodes.) For some time there was talk of a rotatemodel command in IRC that would give HUD access to pitch and roll. (Why I gave up on that is a lengthy explanation.) Right now I am planning to give alpha access to rotatesprite (rotatespritea) and a new text function, even though alpha currently has no effect in classic. (I assume it is theoretically possible but much more difficult due to the sampling math involved.)

I have not comprehended the classic voxel code but from what I have seen, the limitation of x/y flipping comes more from the use of cstat&(16|32) [which are the "flatten" bits, oops] as a hack is just a lack of implementation to iterate in the other direction from the center point. In theory, I can see pitch/roll/etc being implemented in classic (again, limited by how much math is involved, what the support functions are capable of, and performance) by changing the position of the individual volume element squares.

[Fox]

I am having a little issue, I figured out that EVENT_GAME fails to return the player id with THISACTOR if the player in question is dead. I can counter this by getting the yvel of the current sprite, but it's a little annoying to have to use getplayervar instead of ifvare.

[Mblackwell]

Are you saying getplayer[THISACTOR].i returns a bad id?