stgatilov

Well, it is clear that error reporting is good enough I'll fix this case, but in general killing the installer when it is doing something can leave your installation in trouble in many ways, so I'm afraid it is expected that user might have to do something after that...

beta209-02 is available. Changelog is provided in the third post of this thread (and will always be there). Run tdm_installer to get new version, and don't forget to check "Get custom version"

I hope you have heard such thing as "trunk-based development"? That's how games are usually done

I would say it looks very much like C++, but it is quite far from it. It is closer to javascript, PHP, or all the rest of high-level language. Yes, the idea of classes and inheritance is important, but it is now present in most languages. There is major difference between static and dynamic typing. And "auto" keyword in C++ does not make it dynamically typed, so there is huge real difference with how javascript works. I would say C++ is the "language of lifetime". You need a lifetime to master it enough to feel free using it. And the rest of your lifetime to catch up with new things piled upon the old things I guess only very scarce knowledge of C++ is helpful to deal with DoomScript. Basically, the C basic syntax, C preprocessor, C++ class/inheritance/virtual functions, and probably that's all. Then just practice and get generic programming skills, then look at doomscript examples and try to do something similar. And for God's sake don't ever read anything about "threads" in C++! You will be lost for scripting forever

There would be GitHub mirror at some moment. Right now the admin resource is occupied on the better FM database. Then it would be 2.09 release and syncing tdm_installer data to mirrors. Not fast enough, unfortunately.

You can take the modified file and replace the line: draw_Color = vec4(color, 1); with line: draw_Color = vec4(0, 0, 1, 1); Then either restart TDM or execute reloadGLSLPrograms command in console. If the file takes effect, then you should see strong blue color (normally, the whole screen should be full blue). Yes, it is OK.

Yes, I would appreciate if someone improves spacing and layout. The FLTK designer application is located here https://svn.thedarkmod.com/svn/darkmod_src/trunk/ThirdParty/artefacts/fltk/bin/win64_s_vc15_rel_mt/fluid.exe. And the file for GUI is here: https://svn.thedarkmod.com/svn/darkmod_src/trunk/tdm_installer/GuiFluidAutoGen.fl

It should update itself when you try to go to the second/third screen (unless you explicitly disable it). It would post a message when it does so, so you won't miss it.

The resolution is ignored. The whole method does not use any textures.

From what I see, "mirror" keyword is almost equivalent to "subview" keyword, and only affects draw order. It does not enable the actual mirroring effect. You must set mirrorRenderMap keyword so that the mirrored geometry be rendered. It is the problem of bc_teatray material. UPDATE: So both keywords are required to have proper mirror.

I'm sorry but I don't understand what is the problem. Normally, the GUI should catch the "CinematicEnd" event, and use it to stop the video and switch to the next state. That's how stock GUI behaves. Make sure this switch code is correct.

I think you can apply mirror texture to model, but you must ensure that the corresponding surface is planar. Because engine computes mirror plane from arbitrary triangle of the surface. So I believe you can create a round mirror, but you cannot create a curved mirror. The engine handles mirrors by simple linear transformation, it does not have some magic ray tracing features to handle any deformations

Created issue 5462. Is it true that you had such issues with doors which a) are horizontal, and b) have no handles ?

@Goldwell, in case you decide to update this FM in some future, please fix the mirrors (4707). I guess starting from 2.09 dmap will warn you about them.

Performance is a good enough reason. Imagine the engine would render all five faces. That would double the number of frontend runs, and will significantly decrease performance. And if there is no diagnostic, then nobody knows the reason. As for correctness, I have no idea. You can always duplicate material and and apply different material on the other side. To be honest, I believe all sort of mirror effects are hacks, better don't stretch it too much

I tried to test AC2, which has mirrors, and its mirrors are buggy. They have been buggy for quite some time, also in 2.09 they are buggy in slightly different way due to r_useClipPlaneCulling. So I digged deeper, and realized that the mirror that has the problem (near -115.41 415 1157.25) is created with a brush having many mirror-textured sides. The engine only renders mirrored effect for one mirror surface (per material stage), so it just chooses a random surface out of 5-6 available and it happens to be the wrong one. The question is: should I put a warning into dmap to complain about brushes who have several sides of same mirror texture? I guess it applies to all textures with mirrorRenderMap and remoteRenderMap keywords. As for @MirceaKitsune problem, I suggest opening the .map file in text editor and checking that "mirror" only occurs once per every different mirror in your FM, and never on several sides of same brush.

It is probably related to sky. Anyway, could you share your map? At least privately with developers?

Most likely that's the old version dmapped with 2.07: Uninstall the FM, then go to in-game downlowder and get the update, then try again.

Please write exact FM and location (getviewpos). Did you check it on 2.08?

Posted new version of tdm_installer. Fixed name of config file: Darkmod.cfg instead of darkmod.cfg

I'm really sorry for that. Yes. These are the new defaults: seta r_postprocess_desaturation "0" seta r_postprocess_colorCorrectBias "0.0" seta r_postprocess_colorCorrection "5" seta r_postprocess_colorCurveBias "0.0" seta r_postprocess_brightness "1" seta r_postprocess_gamma "1.2" seta r_ambientGamma "1" seta r_ambientMinLevel "0" These values disable all the color correction that was "added" in 2.08. If you also set gamma to 1.0, then you should get exactly the default look of 2.07. By the way, you can copy/paste your old values and use them it you don't like the new defaults.

Could you provide some examples of "good" and "bad" doors?

Every discussion of "automatic settings detection" immediately bumps into problems: We don't own diverse set of machines where we can test such detection. The experience mostly depends on FM. There are too many FMs, constantly getting bigger and tougher, and we don't control it. Thus we cannot make good detection. And it's better to have none than to have buggy one. On the other hand, we can probably implement some code which would detect VRAM size or out-of-memory error, and modify the downscale cvar, post warning, and restart loading from scratch.

Could you please try a few more things with r_tonemap 1 and see if it helps: Set all r_postprocess_XXX cvars to defaults (they should be after config reset). Set r_postprocess_gamma to 1 (its default value is 1.2). Set r_postprocess_sharpen to 0 (its default value is 1). By the way, how does the erratic picture changes when you change r_postprocess_gamma and r_postprocess_brightness (you can do it in console or by moving sliders in menu) ? Also, please try putting the attached file into glprogs directory, i.e. if your executable is at /home/tdm/thedarkmod.x64, then it should be at /home/tdm/glprogs/tonemap.fs. Restart TDM and check how if the erratic picture changes or is gone. This hacked shader file simply passes the input color to output. Be sure to restart TDM for it to have effect. Also, be sure to delete it when experiment is over. tonemap.fs

I guess not... This date is hardcoded in the "zipsync hashzip" command. I hope it is present in all calendars, regardless or DST and local changes, which is not guaranteed if I try to put real date to there. By the way, you always see build time when you start the installer. Here they are: https://yadi.sk/d/0KKFiwpXNjvVrg?w=1