nbohr1more

I deleted the /darkmod/fms/deaddrunk folder and gathered the mission from the in-game downloader. I next restarted TDM and went to the downloader to get the translation pack. I installed dead drunk and set the language to Russian The briefing and objectives now render in Russian. The language pack for this mission includes Russian and English. As I can tell, all is in working order?

Committed at Rev 17318 ( @stgatilov )

Now and Then v3 is now in the mission database.

lol, I didn't realize we have rival platforms: I stuck with Intel because AMD's prices have gone up to crazy town even on used systems since the platform holds it's value longer with X3D and folks are scared of getting damaged 13th \ 14th gen processors due to the ring bus hardware failures. Also, I figured that there might be fewer driver issues on initial boot. I swapped out the board+cpu+ram and both Windows and Linux had no big complaints. Only pain point was enrolling the Nvidia driver into MOK for secure boot compatibility. I ran: sudo update-secureboot-policy --enroll-key but my MSI motherboard couldn't find the certificate ( it wouldn't let me browse to the right path /var/lib/shim-signed/mok/MOK.der ) so I had to copy it to one of my Windows NTFS drives where the MOK utility could easily find it and then Nvidia was happy running under secure boot. Basically your processor has very similar gaming performance to mine but wildly better productivity performance. I wonder if any newer titles with better multi-core support reverse the gaming situation between our chips? Does having more threads give UE5 Nanite powered titles a bigger boost than good IPC ?

Wooo! CPU Upgrade time. I got a gift-card from work so I figured I would get my PC ready for Win11 ( yuck ) ( So I can compile TDM on a supported MSVC version ). ( Also, new games are starting to require AVX2, my old AVX v1 i7-3770 can't do that ) I found a deal on a socket LGA 1700 platform. Now my setup is 3yrs old rather than 14 ( gulp ). CPU i3-12100F 16GB DDR4 Geforce RTX 2060 M.2 NVME Changing from i7-3770S to i3-12100F has increased performance in "Scroll of Remembrance" OLD Linux i7 = 72 FPS, NEW Linux i3 = 137 FPS OLD Win10 i7 = 50 FPS, NEW Win10 i3 = 90 FPS

1) Move your FMS folder somewhere 2) Create a new darkmod folder somewhere 3) Copy the tdm_installer into the new darkmod folder 4) Run the installer 5) Copy the FMS folder into the new darkmod folder ( overwrite ) Alternately, yes you can delete everything in the existing darkmod folder and run the installer.

Yes, that would work

Hmm... My deaddrunk folder has "deaddrunk_l10n.pk4", maybe something is going on with the mirrors? Normal style: Mission author runs their mission through the translation perl script and re-packs their mission in translated form along with a translation pack that only contains the strings folder. ( See Grayman missions for examples ) Darkfate \ Russian style: Mission author runs their mission through the translation perl script the makes a translation pack containing the altered map file, GUI files and Objectives. The original is left alone, the translation pack contains all the changes and overrides parts of the original mission on load. ( See Sotha missions for examples )

Fixed but the downloader doesn't correct it. You'll need to delete the language pack file from your darkmod/fms/builders_influence folder and ( if you prefer ) install the newly fixed one.

https://en.wikipedia.org/wiki/Mipmap Mipmaps are lower resolution versions of textures that are used by your graphics driver when objects are further away. Uncompressed textures typically don't include them because the default uncompressed formats don't support them ( not designed for 3D game engines ). DDS textures can include them in either uncompressed or DXT \ BC compressed formats and we typically do "generate mipmaps" when encoding DDS. How do uncompressed textures get their mipmaps ? We auto-generate them on mission load. ( That used to be painfully slow until stgatilov changed the mipmap generation code to use SIMD instructions and multi-core. ) What does the texture quality slider do? For uncompressed textures, it downscales the images before uploading to the GPU. For compressed images, it simply omits loading higher mipmap levels. It works perfectly as long as you properly encode your DDS with mipmaps.

I figured it out! The impacted textures are DDS encoded but do NOT have mipmaps. Edit: I fixed the core assets in SVN and uploaded a fixed version of the mission with custom textures with mipmaps to the mission database. I think at least one other mission uses the custom flagstone texture so I'll fix that mission ( someday we should add that one to core ).

I tried replicating in 2.12. The panels render properly with image_downSize set to 256 but the snowy stone brick ground texture is black. Something must have made it worse but it was already somewhat broken. Time to bisect I guess.

I think its the specular map. It seems that specular downsize is not working right...

Ah, the material def probably needs to be set to forcehighquality. I wonder of the diffusemap is DDS ?

I cannot reproduce this issue. Did you try deleting and re-downloading? I'll do likewise in case the mission database was updated without a new version flag to show the update. Edit: After deleting the scrooge folder from darkmod/fms/ and downloading v1.4, I still cannot reproduce this issue. I wonder if the textures are npot and thus causing issues for your video drivers? Edit 2: The reason I couldn't replicate is that I had my LOD settings set to High. The problem is due to the LOD defs or LOD models. Biker is working on it. Not caused by 2.13 though. Edit 3: Cannot reproduce on Lowest LOD. Working theory: When you set LOD to highest, the resized model gets auto-generated. Then when you restart at a lower LOD it already exists. If you start at a lower LOD, no auto-generated LOD model gets created? Edit 4: Deleted the darkmod/models/_roth_gen folder Started TDM with Lowest LOD Cannot reproduce. Textures and models look fine @MayheM is there a /models/_roth_gen folder in your darkmod folder?

I guess we'll have to look into editing the mission. @stgatilov ?

Cool! What if you uncap FPS but set Max FPS to 60?

Same behavior with capped and uncapped FPS?

Works in 2.12? 2.11 ?

Yes, diii4a would need to merge our latest code changes to their fork. Otherwise they might need to use scripts, defs, materials, etc from 2.12

See: Either manually download the old version or upgrade to TDM 2.13 beta 5 ( release candidate )

It's cured the issue for me too.

There is a new “last frame capture” used for cool menus that have a screenshot of in-game view as their background. Probably causing enough timing change to provoke this issue.

The old "Fixed FPS" ( com_fixedTic 0 ) mode is still as laggy and poorly performing as ever. With the new mode and maxFPS = 60 it's still waaaaaayyy smoother than the old mode. I can try setting it to max between r_displayRefresh and 120 to see whether a perfect double of screen refresh is still viable to remove the artifacts? I was trying to find a good way to query monitor refresh but it seems that the r_displayRefresh cvar is just an alternate to the GLFW mode data that feeds "parms.displayHz" . It seems that we treat that mode data a little differently in Win vs Linux? So I guess these are the paths to improvement here: 1) Try to use parms.displayHz instead of r_displayRefresh and hope the GFLW does a good job of picking the right value 2) Max between r_displayRefresh and 120 ? 3) Keep trying find a way to restrict maxFPS to 60 in the menus ( I got this one partially working but once you leave the game and return it's stuck at 60. I also had the inverse working, starting at 60 in the menu going to 300 then when returning to menu it was at 300 ) 4) Keep trying to find a way to force gameTic sync when entering and leaving menus ( I guess this is the "correct" thing to do... ). ( Thus far, all the glFinish, clear frame data and RunGameTic things I've tried have failed miserably. ) Sadly, r_displayRefresh is a sorta dangerous parameter to play with and thus it isn't archive-able. We could do it, of course but if a user has a legacy CRT and sets something crazy they could damage their hardware etc. Thus to enforce a higher refresh, users will need an autoexec.cfg or will need to constantly invoke the change in the console. Edit: Results thus far Max between parms.displayHz and any value ( 30, 60, 120 ) = artifacts appear Use parms.displayHz directly = artifacts appear Hard-code 120 = one blip artifact only on the first entry into the game Based on the above, and that hard-coding 60 also eliminates the artifacts I can only surmise that parms.displayHz is not picking up the correct value?

@stgatilov @snatcher I committed a change at Rev 10943 When using "uncapped mode" and vsync together, maxFPS will be limited to 60 or r_displayRefresh ( whichever is larger ). With this change, the menu change artifacts are gone.