New config vars
---------------

(These config vars are typically added to cam_ext.cfg as needed, with the exception of DromEd specific vars which
are usually added to DromEd.cfg.)


use_d3d_display
    use Direct3D 9 display system instead of the legacy DirectX 6 DirectDraw one.
    Enabling this will automatically also enable "force_32bit".

force_windowed
    always run in windowed mode (bit depth will be based on windows desktop bit depth)

force_32bit
    use 32bit rendering for both 2D (UI) and 3D (in fullscreen, in windowed the desktop bit depth is used)

single_display_mode [<1 - 3>]
    run game in a single resolution for both 2D (UI) and 3D, the optional parameter specifies the buffer config,
    default is 1.
    1: blit based buffer updates (same method that's used in windowed mode is used for fullscreen too)
    2: double buffering with buffer flipping (in fullscreen mode), the usual mode for games
    3: triple buffering with buffer flipping (in fullscreen mode)

ui_scale_mode <0 - 2>
    only has any effect in combination with "single_display_mode", determines how UI (/2D mode) will be scaled to
    the current display resolution, default mode is 0.
     0: scale UI to fit screen using basic method, result may or may not be filtered depending on graphics driver
     1: scale UI only an even multiple to maintain pixel perfectness
     2: scale UI to fit using 3D HW, result will be filtered
    When "use_3d_display" is enabled 3D HW is always used and the modes have following (almost same) meaning:
     0: scale UI to fit screen, result is filtered unless an it happens to be an even multiple which can maintain
        pixel perfectness (defining "d3d_disp_scaled_2d_overlay" will force it to always filter)
     1: scale UI only an even multiple to maintain pixel perfectness, no filtering is applied unless
        "d3d_disp_force_filter_scale2d" is enabled
     2: same as 0

d3d_disp_scaled_2d_overlay <val> [<val2>]
    only works in combination with "use_d3d_display", enables scaling of the in-game 2D overlay (HUD).
    * In the first form with only one value, in the range 2 to 64. The game will try to magnify the overlay the
      specified amount of times, if that isn't possible it will find nearest scale factor below that can be
      applied. It will always maintain pixel perfectness so it will choose a scale that results in en even
      multiple of pixel of the base resolution 640x480. For example any resolution that's at least twice the
      size of 640x480 can have an overlay scaling of 2. If you want the game to always choose the largest possible
      scale then set the value really high, i.e. 64, but if you never want it more than double on any resolution
      then set it to 2.
    * In the second form it takes two values specifying a virtual resolution for the 2D overlay. The overlay
      is scaled to fit the current display resolution which may result in slightly blurry results if the scale
      isn't an even pixel multiple, due to filtering. The specified virtual resolution must be at least
      640 480 and less than the actual display resolution. The aspect ratio of the virtual resolution will
      automatically adjusted to match the actual resolution.
    NOTE: Using scaling will actually improve performance (to some degree), in particular at high resolutions.

d3d_disp_force_filter_scale2d
    only works in combination with "use_d3d_display", forces scaled 2D (menus and overlay) to always use bilinear
    filtering. Normally no filtering is applied when the scale is detected to be an even pixel multiple of the
    original size, in order to get sharper pixel perfect results.

d3d_disp_2d_overlay_alpha <0.0 - 1.0>
    only works in combination with "use_d3d_display", sets the alpha/transparency of the 2D overlay, default is
    1.0 (opaque).
    NOTE: This affects everything that's drawn in 2D while in-game, including the command console.

d3d_disp_2d_surf_mode <0 - 3>
    only works in combination with "use_d3d_display", sets the method in which the 2D overlay (HUD) is handled,
    default is 3 (hardware based 2D rendering and 3D rendering in editor). Using the default HW method should
    significantly increase performance related to 2D/HUD drawing, especially at higher resolutions. If there are
    problems with garbled HUD texts and "d3d_disp_force_rgba_atex" doesn't help, then it may be solved by changing
    this to mode 0, 1 or 2.
     0: software rendering (normal method for SW rendering)
     1: software rendering (alternative method which probably ends up doing the same thing as 0)
     2: software rendering (alternative method using lockable dynamic vidmem texture)
     3: HW rendering (recommended/default)

d3d_disp_hw2d_cachesize <8 - 2047>
    max vidmem usage (in MB) to try to keep below for vidmem textures related to HW 2D overlay (HUD) rendering,
    default is 16. Normally nothing that needs to touched, but might be useful for cards with limited memory.

d3d_disp_hw2d_cachesize_editor <8 - 2047>
    same as "d3d_disp_hw2d_cachesize" except this applies to all HW rendering while in edit mode, default is 64.

d3d_disp_force_rgba_atex
    only works in combination with "use_d3d_display" and 2D overlay (HUD) in HW mode (see "d3d_disp_2d_surf_mode"),
    forces the use of RGBA texture format for alpha-only textures in HW 2D rendering, HUD texts in particular,
    the default uses the best suitable alpha-only format supported by the GPU (which should always work ok, this
    option is just provided in case some old cards are having problems)

d3d_disp_allow_a8l8_atex
    only works in combination with "use_d3d_display" and 2D overlay (HUD) in HW mode (see "d3d_disp_2d_surf_mode"),
    enables support for using A8L8 texture format for alpha-only textures, as a fallback when A8 isn't supported
    (by default the fallback format is 32-bit RGBA)

movie_sw_scale_quality <0 - 6>
    quality of software scaling filter (used during internal processing, not the same as the ui_scale_mode),
    default is 4 (bicubic). Lowest quality with best performance is 0 (point filter).

scale_movie_to_ui
    only has any effect in combination with "single_display_mode", will play movies within the UI area only
    when "scale_ui_even_pixels" is set, instead of scaling it to fit the current resolution

crop_movies [<left> <top> <width> <height>]
    only has any effect in combination with "single_display_mode" and not "scale_movie_to_ui", enable cropping
    of movies so they can fill more area on a widescreen display. The original Shock cutscenes all have a
    built-in letterbox so this option works good there, for Thief it would cut off the upper and bottom parts
    of the movies. Optionally the crop rect can be specified in 640x480 coords, regardless of movie resolution.
    The default crop rect, when not supplying a custom one, is optimized for SS2 and is 0 70 640 340.

movie_crop_exclude
    only has any effect with "crop_movies", a '+' separated list of movie filenames to exclude from cropping.
    For example: movie_crop_exclude credits.avi+success.avi+death.avi

movie_disable
    a '+' separated list of movie filenames to that should not be played.
    For example: movie_disable intro.avi+death.avi

no_unload_ffmpeg
    keep ffmpeg.dll loaded, otherwise it's loaded and unloaded each time a movie is played.
    This may or may not be helpful when there are problems with movies initially playing but then fail to play
    at a later time (which produces log messages about ffmpeg.dll failing to load)

ffmpeg_force_intern_indeo5
    force the use of ffmpeg's internal indeo5 codec even if ir50_32.dll is present

widescreen_lock_hfov
    use the same horizontal FOV as in 4:3 mode, this is generally disliked by widescreen users, but the option
    is included as that is how the code works natively

fov <val>
    set (horizontal) FOV in degrees, default is 90. the FOV is specified for the reference 4:3 aspect ratio,
    the effective FOV in widescreen will automatically be larger (unless "widescreen_lock_hfov" is specified)

vsync_mode <val>
    synchronize screen updates with display refresh, default mode is 3.
    The mode val is the sum of one or several of following values or 0 to disable all:
       1: enable vertical synch in 3D mode (in-game)
       2: enable vertical synch in 2D mode (menus)
       and optionally _one_ of these for non-standard flip interval in vsync:ed fullscreen modes
        4: flip every second refresh
        8: flip every third refresh
       12: flip every fourth refresh
    Graphics driver settings may override or not support some settings.
    For example: to enable synch in both 2D and 3D (=1+2) "vsync_mode 3", to disable all synch "vsync_mode 0"
    console command: set_vsync <val>

d3d_disp_limit_gpu_frames <0 - 3> [<0 - 3>]
    only has any effect in combination with "use_d3d_display", limits the number of frames that can get buffered
    in the GPU command buffer, default is 0 0 (limit disabled).
    The first value defines the buffer limit in 2D mode (menu screens), the second optional value defines the
    limit in 3D mode (in-game).
    The GPU command buffer can hold render commands for several frames. When rendering something simple with
    vsync enabled, like for example menu screens, the buffer easily gets filled. This results in noticeable input
    lag especially in menus, since there can be several frames between mouse movement and visual response. By
    limiting the number of frames that can get buffered the input lag can be reduced. It's not recommended to
    enable the options if no lag is experienced.
    May not have any effect on older graphics cards.
    IMPORTANT: Enabling limiting in-game should be used with care on multi-GPU setups, since it can result in
               performance loss caused by not giving all GPUs enough work.

d3d_disp_refreshrate <rate_hz>
    only has any effect in combination with "use_d3d_display" and running in fullscreen mode, sets a custom
    desired display refresh rate. The custom refresh rate will only be used if DirectX reports it as a supported
    rate for the current display mode. If the requested rate is not available, then the game will attempt to set
    it to the closest rate below. For example if the requested rate is 85 but DX only reports 72 and 60 as being
    supported, then it will use 72 instead.

framerate_cap <val>
    only has any effect in combination with "single_display_mode", enable software based framerate limiter to
    the specified framerate, default is 0 (disabled)

user_mode<1 - 8> <width> <height> [<bpp> [<aspect_x> <aspect_y>]]
    up to 8 user defined display modes. If only width and height is supplied the mode is assumed to be 32-bit
    with an automagic aspect ratio (assuming square pixels). Aspect ratio is defined in x:y form, so for 16:9
    it would be "16 9".
    For example: user_mode1 1280 720 32 16 9

gamma_ui <val>
    setting this enables UI display modes to have its own gamma setting, instead of using the "gamma" option
    for everything. This can be useful if you play in the dark and get blinded when opening the map or a
    a readable in Thief.

gamma_movie <val>
    gamma setting for movie playback, by default the game uses 1.0 for movies

force_32bit_textures
    load all textures as 32-bit, preserving original quality, as well as enabling 32-bit textures for various
    subsystems. Only works if "force_32bit" is also specified.

disable_alpha_test
    disable (newly added) alpha test support, when disabled, some objects like for example the plants in
    SS2 hydroponics, may exhibit rendering artifacts when other transparent objects are behind them.
    Alpha testing (aka alpha keying) is similar to chroma keyeing, which skips rendering of pixels that are meant
    to be transparent. This can create more pixely edges, but it perfectly solves alpha sorting issues.
    console command: toggle_alpha_test
    console command: set_alpha_test <0 , 1>

alpha_test_as_opaque
    treat objects with 1-bit alpha (chroma key) textures as opaque. Requires alpha test to also be enabled.
    The advantage is that these objects don't have to be depth sorted and it avoids sorting errors where
    other parts or objects get incorrectly drawn on top. The disadvantage is that in some cases overlapping
    decal objects may exhibit z-fighting.
    console command: toggle_alpha_test_opaque

default_alphatest_ref <1 - 255>
    alpha test alpha threshold value, default is 127.

d3d_disp_enable_atoc [<0 - 2>]
    only has any effect in combination with "use_d3d_display", enables alpha-to-coverage (ATOC) functionality
    for alpha tested rendering. This option requires multisampling to be enabled, and only works if the graphics
    card and driver supports alpha-to-coverage.
    Mode 2 is a special mode for nVidia cards that support supersampled transparency (SSAA). SSAA may be more
    expensive in return for bit better quality. If the card doesn't support SSAA but supports ATOC it will
    fall back to ATOC.
    ATOC/SSAA makes the edges of alpha tested surfaces look smoother and less aliased.
    This option may look better when combined with lower alpha threshold like "default_alphatest_ref 1", but
    "tex_edge_padding" should also be enabled or there may be ugly outlines.

disable_table_fog
    disable support for table fog, will fallback to vertex fog if supported or none

fog_mode_exp
    use exponential fog table mode instead of the default linear

tex_filter_mode <0 - 16>
    set texture filtering mode: 0 no filtering, 1 linear, 2,4,8,16 anisotropic (number is sample size)
    console command: set_tex_filter <0 - 16>

tex_filter_trilinear
    enable trilinear texture filtering (smooth transition between mipmap levels), only has any effect if
    mipmap_mode is set to non-zero
    console command: set_trilinear <0 , 1>

lm_filter_mode <0 - 16>
    set lightmap (the second texture stage) filtering mode, same modes as "tex_filter_mode" above
    if this config var isn't defined then lightmaps will have get same filtering as "tex_filter_mode"
    console command: set_lm_filter <0 - 16>

disable_lightmap_aniso
    (DEPRECATED) use "lm_filter_mode", this config var is entirely ignored if "lm_filter_mode" is also defined
    prevents the use of anisotropic filtering on lightmaps (the second texture stage)

mipmap_mode <0 - 2>
    enable mipmap generation for textures: 0 disabled, 1 fastest (quality depends on HW features), 2 software
    bilinear filtered. When "use_d3d_display" is enabled there may be little or no difference between 1 or 2.

lm_mipmap_mode <0 - 2>
    mipmap generation mode for lightmaps, this is ignored if "mipmap_mode" is 0. Disabling it may be helpful
    if there are performance issues or when experiencing visual artifacts (that look like light bleeding), then
    again it could also make the latter worse or have no effect at all.
    0 disabled, 1 fastest (quality depends on HW features), 2 use same quality as "mipmap_mode" (default)

lm_init_texmem <0 - 2>
    when initializing a texture for HW lightmap usage, clear surface memory to make sure there's no random data,
    default is 0 (disabled). Enabling this should reduce filtering artifacts (light bleeding) on lightmap edges.
    0 disabled, 1 clear when creating surface, 2 clear when creating and when flushing cached LMs from surface
    NOTE: Although option 2 should work without problems, it hasn't been tested quite as much. If there are any
          issues like lightmaps/surfaces suddenly turning black, then change this option to 1.

lm_filter_margin <pixels>
    only has any effect if "lm_init_texmem 2", pixel spacing to use between cached lightmaps in a lightmap texture,
    default is 0. Adding some spacing between lightmaps will help avoid/reduce light bleeding artifacts on lightmap
    edges, at the expense of being able to fit less lightmaps per HW lightmap texture.

mipmap_bias <-16.0 - 16.0>
    adjust mipmap LOD bias, only has any effect if "mipmap_mode" is non-zero. The bias adjusts the GPU selected
    mipmap level so with a value of 1.0 it would use one mipmap level smaller than it normally would.
    console command: set_mipmap_bias <-16.0 - 16.0>

tex_edge_padding <0 - 2>
    enable edge padding for alpha textures: 0 disabled, 1 1-bit alpha textures upconverted at load-time from
    palette based image formats, 2 all alpha textures.
    When enabled only textures that are larger than 4x4 and no larger than 256 are processed. The processing
    is done when textures are loaded (normally during mission load), so depending on the amount of textures and
    their sizes there may or may not be a bit longer load times.
    Edge padding increases the visual quality of mipmaps, as it fixed surrounding colors (often purple) bleeding
    into the edge areas. Padding should normally be done in the original image, but it was not possible with
    the old 8-bit textures the game used.

tex_max_edge_padding <1 - 999>
    set the max number pixels to expand edges, if edge padding is enabled through "tex_edge_padding" or material
    option. The default is 8. Increasing it will increase quality of the lower/smaller mipmap levels at the
    expense of more processing time.

texel_center_adjust <-1.0 - 1.0> <-1.0 - 1.0>
    texel/pixel center adjustment, default is 0.5 0.5. If the offset is incorrect textures from 2D components
    like UI and HUD icons will look blurrier than they should, and when multisampling is enabled you may see
    a pixel line at the edges of the screen. The default value should cover most somewhat modern setups.

multisampletype <2 - 16>
    only has any effect in combination with "use_d3d_display", sets the multisample (anti-aliasing) amount

precache_d3d_textures
    pre-cache all terrain textures to video memory before entering game mode. This will add a little delay (may
    or may not be noticable depending on amount of textures and system performance) when entering in-game after
    loading a mission or returning from menus to in-game. It may help reduce framerate stuttering caused by
    uploading of textures and mipmap creation, when viewing areas for the first time in a mission. Should be used
    with caution with missions whose texture usage might exceed available video memory. If any problems are noticed
    as a result of enabling this option, then disable it again.
    Specifying "precache_d3d_textures 2" will also pre-cache material textures that are used by any terrain.

max_hw_textures <1024 ->
    max number of vidmem texture slots, default is 4096 (used to be 1024 in older versions). If you have a card
    with not so much video memory and you get texture creation errors due to running out of local vidmem, then
    you could try lowering this. If lowering this does not help and "use_d3d_display" is enabled, you could look
    into "d3d_disp_managed_textures".

d3d_disp_managed_textures
    only has any effect in combination with "use_d3d_display", when enabled D3D managed texture memory is used,
    which may or may not affect performance a tiny bit (negatively or positively) or solve potential texture
    memory issues if such occur at the expense of increased system memory usage (as a memory copy of every
    texture is kept by D3D).

d3d_disp_forcevalid
    only has any effect in combination with "use_d3d_display", forces D3D initialization to succeed. Can be useful
    if Dark fails to init D3D due to ValidateDevice failing although it technically shouldn't, either caused by
    some issue in Dark that slipped through the cracks or a driver bug.

d3d_disp_enable_hdr [32]
    only has any effect in combination with "use_d3d_display", enables 16-bit (or optionally 32-bit) floating point
    render buffer (if supported), for increased quality in particular with postprocessing and software color/gamma
    processing. Enabling it will result in more video memory usage and some performance loss.

d3d_disp_sw_cc
    only has any effect in combination with "use_d3d_display", enables software gamma and color processing. The
    benefit of doing gamma in (shader) software rather than HW gamma, is that there is less quality loss
    because it can be applied on the higher precision floating point or 10-bit render data, before getting
    down converted to a displayable format. Whereas HW gamma is applied on the already downconverted displayable
    format. Another benefit is that gamma works in windowed mode, where HW gamma doesn't work.
    Enabling color processing comes with some performance loss.
    It's possible to override the CC shaders with custom ones. Either to customize the gamma or color processing,
    or to add additional processing like noise/dithering. When there's additional processing in the shaders, that
    should be done regardless of gamma/color processing being active or not, then define "d3d_disp_sw_cc 2". It
    will make sure the CC shaders are always used.
    For further details and example shaders see the "doc/sw_cc" directory.

d3d_disp_sw_cc_bright <val>
    brightness level when "d3d_disp_sw_cc" is enabled, default is 0. A value of 1 would make everything white.
    This option is only provided for completeness, it's more appropriate to adjust gamma.
    Does not affect menu screens.

d3d_disp_sw_cc_contr <0.0 - >
    contrast level when "d3d_disp_sw_cc" is enabled, default is 1. This option is only provided for completeness,
    it's more appropriate to adjust gamma.
    Does not affect menu screens.

d3d_disp_sw_cc_sat <0.0 - >
    saturation level when "d3d_disp_sw_cc" is enabled, default is 1. A value of 0 is completely desaturated (b&w).
    Does not affect menu screens.

d3d_disp_sw_cc_rgbfilter <0.0 - 1.0> <0.0 - 1.0> <0.0 - 1.0>
    RGB color filter when "d3d_disp_sw_cc" is enabled, default is 1 1 1 (white). For example to get a old photo
    type of look (sepia) you could do "d3d_disp_sw_cc_sat 0.25" and "d3d_disp_sw_cc_rgbfilter 1 0.7 0.3".
    Does not affect menu screens.

d3d_disp_sw_cc_hwgamma
    only has any effect in combination with "d3d_disp_sw_cc", forces gamma to be applied by display HW like it
    does when "d3d_disp_sw_cc" is disabled, but other color control like saturation and color filter are still
    applied. This _might_ help reduce any performance hit (induced by "d3d_disp_sw_cc") slightly, at the expense
    of gamma not working in windowed mode and potential minor quality loss. Another potential issue is that
    contrast/brightness calculations are normally applied after gamma, with this option enabled gamma is applied
    after brightness/contrast (if brightness/contrast isn't used, i.e. set to default, this is of no concern).

postprocess <0, 1>
    only has any effect in combination with "use_d3d_display", enable postprocessing effect (bloom)

bloomprescale <0.0 - >
    bloom intensity multiplier applied before blurring is applied, default is 1

bloomscale <0.0 - >
    bloom intensity multiplier, default is 5

bloom_range <0.0 - 100.0>
    bloom(/"glow") radius defined as percentage of the screen's diagonal, default is 2. The larger the value the
    more GPU processing is required.

bloom_saturation <0.0 - 1.0>
    bloom saturation, default is 0.7. Complete desaturation (black and white) is 0.

bloom_threshold <0.0 - 1.0>
    min value required (for R, G and B) in order for a pixel to be included in bloom processing, default is 0.6.
    A high value means only the brightest surfaces will glow.

enhanced_precip_trace
    enable more elaborate/expensive tracing for rain/snow impacts so it also hits more things than just terrain.

enhanced_corona_trace
    enable more elaborate/expensive tracing for corona visibility so it also hits more things than just terrain,
    fixes the issue where coronas are drawn on top of objects/characters.

enhanced_aifootstep_trace
    enable more elaborate/expensive tracing for AI footstep sounds so it also hits more things than just terrain.

enhanced_aidetect_trace
    enable more elaborate/expensive tracing for AI body and suspicious obj detection so it also hits more things
    than just terrain. Enabling this will for example prevent AIs from seeing bodies through closed doors (that
    block vision) or other objects that block AI vision.

model_lod_dist_factor <factor>
    modify object LOD (level-of-detail) distances to balance quality vs. perfomance. Default is 1.0. This only
    affects objects that have LOD support. Increasing the factor above 1.0 will make LOD changes happen farther
    away (higher quality), and decreasing it below 1.0 will make LOD changes happen closer to the viewer (higher
    performance).

model_lod_force_low
    disable the highest LOD (level-of-detail) of objects that have LOD support. This option could be useful if
    you have performance issues with missions that use a lot of high-poly objects (and that have LODs defined),
    where lowering "model_lod_dist_factor" doesn't help enough.

screenshot_format <type>
    select file format for screenshots, default is BMP.
    Available types: BMP, PCX, PNG

screenshot_timestamp
    use timestamp based numbering for screenshot names. Makes it easier to merge multiple screenshot sets
    and also to maintain sequence order even if some files have been deleted.

bob_factor <0.0 - 1.0>
    set a float scale factor for player head bob, 1 is normal and 0 completely disabled

shock_gun_bob_factor <0.0 - 1.0>
    (Shock only) set a float scale factor for player weapon bob, 1 is normal and 0 completely disabled

SlowFrameUI <ms>
    cap framerate when in UI mode (or game is paused) by setting this value to the minimum frame time.
    Setting it to 10 would cap the UI framerate to 100 FPS, and avoids the graphics card running unnecessary
    hot. If this is not set (or -1) then the original "SlowFrame" is used for both UI and in-game (which is
    disabled/uncapped by default).

phys_freq <30 - 120>
    set physics updates to a fixed frequency, this avoids variable movement speeds when playing without
    V-sync or can be used to lock physics at a lower frequency to play as the game originally was likely
    intended (60Hz). The default value is 0 which disables fixed frequency updates. It's always preferable
    to run with V-sync for the best experience.
    To get any useful effect of this at higher framerates the var "min_frame_time" should be set to
    something low like 1.

use_hi_res_timer
    use higher precision  timer functions, instead of the old 1ms standard resolution. Better precision provides
    a smoother experience, but some older OS versions (some XP versions and older) may not properly support
    hi-res timers due to implementing them with RDTSC or other reasons, which can lead to movement speeds going
    crazy and more.

snd_oal_device
    when using the OpenAL sound driver it normally uses the default AL device, if another device is preferred
    then specify the name with this config var, for example to use the device "DirectSound Default":

         snd_oal_device DirectSound Default

disable_oal_eax_reverb
    when using a EAX capable OpenAL device and reverb is enable (EAX option in options menu), the sound driver
    will use EAX reverb instead of AL's EFX reverb. Specifying this option will suppress the usage of EAX
    reverb and only use EFX reverb.
    NOTE: the menu option EAX does not specifically mean EAX, it only means reverb in general

snd_decompress_on_load <val>
    decompress sounds that use compressed formats when they are loaded, default mode is 1. Decompressing
    will use more memory and may cause a little hickup when the sound is first loaded, but may on the other
    hand provide smoother framerate during subsequent playbacks.
    The mode value can be one of following:
        0: Don't decompress any sounds on load (decompression is done during playback).
        1: Decompress OGG sounds on load.
        2: Decompress both OGG and ADPCM sounds on load.

snd_speaker_config <val>
    set a non-default speaker configuration (not all devices supports this, especially with OpenAL only the
    "OpenAL Soft" (soft_oal) device driver currently supports it.
    The supported configurations are:
         0: Default
         2: Stereo
         4: Quadraphonic
         5: Surround
         6: 5.1 Surround
         7: 6.1 Surround
         8: 7.1 Surround
         9: 5.1 Surround (Side)
        10: 7.1 Surround (Center L+R)

pixel_parts_as_disks
    when defined pixel particles are rendered as (small) disks instead
    console command: toggle_pixel_parts

square_parts_as_disks
    when defined square particles are rendered as disks instead
    console command: toggle_square_parts

fixed_star_size <val>
    set a resolution independent star size (for NewSky stars), the default behavior is to render stars as 3x3
    pixel large polys regardless of resolution, which will make them look smaller or larger depending on
    resolution. The fixed star size is defined as a scale factor based on the size the stars normally have at
    1280x960 resolution, so with a size of 1.0 the stars will pretty much look like they do at 1280x960 in any
    resolution. With a size of 0.5 they will look half the size of what they do at 1280x960, in any resolution.

disable_star_mipmaps
    disable mipmapping for (NewSky) star texture (only has any effect if mipmap_mode isn't 0)

d3d_disp_enable_distortionfx
    only has any effect in combination with "use_d3d_display", enables screen distortion effects like seen when
    looking through scouting orb or heat distortion from fire arrows.

enable_sw_heatdisks
    enable heat distortion effect, mainly from fire arrows, when game is running in DX6 mode (i.e. when
    "use_d3d_display" is NOT defined)
    WARNING: The performance is not good enough for real usage, this option is for reference and those curious
    about how it looked in Thief 1 with the software renderer.

indexed_mm_render <0, 1>
    enable/disable the use of indexed primitive rendering for character/deformable meshes, default is enabled in
    DX6 and disabled in DX9. Changing the default will have a performance degrading effect in DX6 and may have so
    in DX9 (except maybe on very slow CPUs, but that's a big maybe), and additionally some DX9 drivers have issues
    when it's enabled.

wr_render_zcomp
    enable z-compares for worldrep rendering (normally it only does z-write because it renders worldrep polys
    back to front). Enabling z-compare can slightly improve rendering efficiency because polygons may be
    batched more, due to rendering not being order dependent.

dark_zcomp_arm
    (Thief only) when defined the player arm/weapon is rendered with Z-compare enabled, to get rid of
    sword and blackjack being rendered on top of the hand. This does currently not affect arrow rendering
    due to their pos/rot diverging too much from the bow's.

scaled_inv_bitmaps
    enable (up-)scaling of bitmap based inventory icons, like Thief's body carry icon, at higher screen
    resolutions.

dark_map_decal_scheme <method> [<method_params>*]
    (Thief only) select and configure the way visited and current map locations are shaded/colored.
    Method 0: original, has no params (default)
    Method 1: dodge/burn, to customize this method supply either 2 params where the first is a monochrome factor
              for visited areas and the second for current area, or supply 6 parameters where the first 3 is an
              RGB triplet for visited areas and the last 3 for current area.
              The param values are floating point values, a positive value works like a Photoshop/GIMP dodge,
              a negative value like burn. "-0.3 -0.3 1.1  1.3 1.3 0.2" would look similar to the original,
              default when not supplying any params is "-0.5 0.8"
    console command: set_map_decal_scheme <method> [<method_params>*]

dark_map_lalign_notes
    (Thief only) left align map notes instead of centering them

dark_map_remember_last
    (Thief only) open last viewed map page if no automap location is found for current player location

loadout_buy_autoadvance
    (Thief only) auto-advance item in loadout screen after buying all available units of the current selected one,
    like it did in Thief1

vismeter_hide <0, 1>
    (Thief only) hide light gem. This config var is monitored continously so custom scripts could change it on the
    fly to show/hide the light gem during a mission. However if a script does that it's recommended that it checks
    the initial value and honors that, so that if a player has globally disabled the gem, it stays disabled.

hide_maxed_hpbar
    (Thief only) hide health bar when player has full health

ai_death_keep_phys <0, 1>
    (Thief only) enable/disable that AIs keep their physics shape after they die or are KOd, default for T2 is 0
    and 1 in T1 mode

dark1
    (Thief only) added to cam.cfg to tell game to run in Thief1 "mode". It also requires a couple of extra
    patch resources to make the menu system fully usable.

legacy_force_match_unrefs 0
    default value for "force matching unrefs" in particles, when loading SS2/T1 particle data created with previous
    game versions. The "force matching unrefs" parameter was added for T2 and defaults to disabled there, in SS2
    and T1 the engine behaved as if this option was enabled. To match original behavior it defaults to that when
    loading old data. It may however not be desired behavior because there are issues involved when enabling the
    option. It was required to so particle effects on weapons (i.e. fire arrows) were rendered, but it's now possible
    to enable that with "render_weapon_particles".

render_weapon_particles
    render particles attached to weapons (while held in hand) when they they have "force matching unrefs" disabled.
    In T2 this will make fire arrow particles visible (before they're fired). In T1 when combined with
    "legacy_force_match_unrefs 0" it will make the fire arrow particles work better.

analog_mwheel_bind
    mouse wheel now binds up and down events as separate button events by default, i.e. you bind mouse_wheel_up
    to Next Weapon and mouse_wheel_down to Prev Weapon. If the original behavior is desired, where the
    mouse_wheel control sends its analog value as a command parameter, then define this config var.
    Note that if a bind file still has a mouse_wheel only bind, that will still work regardless of this config
    var, it only affects the binding behavior in the options menu.

shock_single_bind_set
    (Shock only) when defined shock only uses "user.bnd" and disregards "user2.bnd"
	
shock_quicksave_slot <0 - 14>
    (Shock only) use a save slot for quicksaves, instead of the original behaviour of simply saving to the default
    autosave/current slot.
    Specify 0-14 to use a regular save slot, or -1 for the default behaviour of using the autosave/current slot.
    A value of 14 will use the lowermost regular save slot.

shock_autosave_slot <0 - 14>
    (Shock only) use a save slot for autosaves, instead of the original behaviour of simply saving to the default
    autosave/current slot.
    Specify 0-14 to use a regular save slot, or -1 for the default behaviour of using the autosave/current slot.
    A value of 14 will use the lowermost regular save slot.

shock_old_hud3d_order
    (Shock only) restore old drawing order for (spinning) 3D objects in HUD elements. This should never be needed
    and is only provided as a precaution in case some unexpected issue pops up.

force_underwater_reverb [<reverb preset>]
    always use the UnderWater reverb environment when the player is under water. This is useful if maps
    haven't explicitly set the reverb type for underwater areas so it ends up with some default. Optionally
    a specific (0-based) reverb preset index can be set, by default it's 22. The preset numbering is based
    on the "EAX Reverb Code" list in DromEd (which is the standard EAX preset list). 21 might be of interest
    for a more pronounced effect.

submerged_sound_occlusion <min blocking factor> [<distance penalty factor> [<reverb mix>]]
    when the player is under water sound distances are by default artificially increased to achieve a bit less
    direct hearing of sounds. This effect is negligible when EAX is enabled, making sounds very direct.
    Introducing a minimum blocking factor (occlusion / low pass filter), adjusting the distance penalty
    and reverb mix may give a better underwater experience. The reverb mix will be ignored if it's less than
    the normal reverb mix, which is 0.6 by default (or changed with "sfx_source_reverb_mix").
    The defaults are -1000000.0 1.0 -1.0 (which results in using the original method).

new_mantle
    enable new improved (hopefully) mantling code. The new mantling code should have a lower failure rate, so it
    doesn't start a mantle only to fail midways through as much as the standard mantle code, and/or requiring as
    many retries at different approach angles until successful. This means that it can sometimes be easier to get
    to places in a map that weren't intended to be reached, but then again a motivated/stubborn player might also
    manage to do so with standard mantle. New mantle however has the ability to mantle into crouch spaces, which
    standard mantle doesn't have.

force_correct_rope_lengths
    force all ropes/vines to calculate the rope lengths correctly so that their physical length matches their
    numerical length. The old way of calulcating the physical rope lengths is incorrect and results in ropes
    that are a certain percentage shorter than what they're supposed to be. The correct length calculation isn't
    enabled by default to ensure that placed ropes in existing missions retain their original lengths. It can
    however be helpful to enable this when running in Thief 1 mode, because handling of placed ropes in T1 was
    a bit different than in T2 so ropes can end up too short otherwise. Mission authors can also enable correct
    length calculcations directly on rope objects. Such rope objects will have correct lengths even when
    "force_correct_rope_lengths" isn't enabled.

rope_ground_margin <1.0 - 6.0>
    distance above ground where ropes/vines end when "force_correct_rope_lengths" is enabled or the rope objects
    have "Calc Correct Len" set in the Rope property. Default is 4.5.

extended_player_lighting [<0 - 2>]
    extended player lighting mode, if no additional numeric param is specified it assumes 1.
    Default calculations only sample lighting at one point in the player, with extended player lighting
    more sample points are added. Extended mode is disabled by default.
    Modes: 0 = disabled, 1 = normal weighted average, 2 = only uses the weighted average if result is brighter
    console command: set_extended_plr_lighting <0 - 2>

dynamic_light_obj_radius_scale <0, 1 - >
    radius scale factor for dynamic lights when calculating object lighting, default is 1.0. This does NOT affect
    the radius for how dynamic lights light up lightmaps(/terrain). A value of 0 restores original behavior where
    dynamic lights have an infinite radius with regard to objects.

sun_runtime_obj_shadows
    enable sun light to be obstructed by objects with the "Runtime Object Shadow" property, just like the property
    affects other lights.

frob_mode <0 - 2>
    frob line of sight mode, default is 0. This allows enabling of a more strict obstruction check for a potential
    frob target, to prevent frobbing through walls scenarios. That does however mean that sometimes objects which
    are only partially visible won't be frobbable, like an object on a shelf (made of terrain polys) above head
    height, unless you jump or climb onto something, or an object around a corner that's only partially visible.
    Use at your own risk, it could potentially break missions that placed "hidden" frobbable objects in such way
    that the player can never see more than a small fraction.
    Modes: 0 = standard (only objects marked as blocking will obstruct)
           1 = all terrain and objects marked as blocking will obstruct
           2 = (Shock only) like mode 1 but prevents HUD brackets/info to appear on obstructed objects, unlike
               mode 1 where the brackets/info will appear the same way as they do in mode 0, even if the object
               can't actually be interacted with. For Shock this mode is recommended over 1 because it's less
               confusing. (In Thief this mode is the same as 1.)

disable_headtracking
    disable head tracking (head looking at the player) of NPCs

ext_kb_country_check
    extended keyboard language detection. By default the game can under certain circumstances fail to correctly
    identify the keyboard language, it also does not identify UK or French Canadian layouts. This option solves
    those issues, but it does not add support for more layouts than already available in the game, which are:
    English-US, French, German, English-UK, French Canadian, Italian, Spanish.

skip_writeaccess_check
    skip write access test of game dir on startup

swim_chest_offset <0.0 - 1.15>
    customize the point of the player that needs to be submerged in water to classify the player as swimming.
    Default is 0 which uses the standard swim point which is at the "body" physics shape origin. When setting
    a custom (non-zero) value, the value is defined as height units below the "head" physics shape origin.
    Something around 0.6 would be recommended for a custom offset, which puts the point at chest height.

new_mantle_fwd <0.5 - 1.0>
    only has any effect in combination with "new_mantle". Minimum distance the player needs to be able to move
    forward during a mantle motion in order to accept the mantle as valid, default is 0.5. Increasing the value
    will reduce the ability to mantle narrow ledge situations (if you feel that the default is too "effective").

jump_speed_boost <1.0 - >
    speed boost factor when jumping, should not be altered as it will break missions that have locations carefully
    crafted based on the game's default jump distance. Default is 1.05 (or 1.4 when running in Thief 1 mode).

no_cretpose_in_sim
    disable fix that allows the creature "Current Pose" property to be applied to creatures that are created (i.e.
    spawned) during sim. Previously "Current Pose" only applied to creatures that were pre-placed in the mission.
    This config var is only provided as a safety precaution, just in case the fix causes any problems.

small_portal_repel
    this reduces the safety distance margin between the player camera and portals during rendering. Maintaining
    the safety margin is what can cause the small "jitter" that you can sometimes experience when crossing portals,
    such as walking across a door opening (portals are however everywhere and not just doors). By enabling this
    option, a smaller margin is used and the visible jitter is reduced or often eliminated.
    Although no problems are expected, in the unforeseen event that you experience issues like the screen
    flickering black when crossing portals, then disable this option again.

phys_deactivate_outofworld_ai
    restore original behavior that deactivates physics on AIs if they are partially (or completely) out of world
    after loading a savegame or mission. The new default mode of operation is to keep physics active if the AI is
    deemed to only be slightly out of world and hope it will recover. The original behavior would sometimes cause
    AIs to lose physics after loading a savegame or a level transition, so you couldn't damage them directly or
    collide with them. In case there are any unforeseen problems with the new behavior, this config var will
    restore the original way of doing it.

disable_32bit_fampal
    disable use of family palette (full.pcx) when load-converting palette-based (indexed) textures to 32-bit
    (see "force_32bit_textures"), instead use each texture file's own palette. Default behavior is to use the
    family palette.

legacy_32bit_txtpal
    enable use of the global game palette when load-converting palette-based (indexed) object textures, from
    the "txt\" directory, to 32-bit (see "force_32bit_textures"). Default behavior is to use each texture
    file's own palette.

incremental_save_dir <dir>
    (DromEd only) enable incremental saving when overwriting existing MIS/COW/GAM file. Default is disabled.
    When enabled, DromEd will make a backup copy of the file about to be overwritten, to the directory specified
    by this config var. The directory must exist, DromEd will not create it for you. It may also be an absolute
    path like for example "C:\Temp\DromEd_bak". The backup files will have a timestamp in their names of when
    the backup copy was made. Not to be confused with time when the file was last modified.
    Be aware that DromEd will not limit the number of backups, so you may occasionally want to delete older
    backups if you want to keep disk usage at bay.

brush_centers <0 - 2>
    (DromEd only) draw brush centers, 1 draws only the current brush, 2 draws it for all brushes
    console command: brush_centers <0 - 2>
                     or "brush_centers -1" to cycle between the three modes

validate_spot_cones
    (DromEd only) when defined it will validate spotlight cone angles during lighting and output invalid
    settings to mono.

validate_models [2]
    (DromEd only) when defined it will validate MD models when they're loaded to make sure that polygon
    data is correct and print warnings to mono otherwise. If the var value is 2 then invalid models will
    fail to load, otherwise load will succeed and a crash is likely to follow.

enable_32bit_lightmaps [<1 , 2>]
    (DromEd only) when defined DromEd will default to 32-bit lightmaps instead of the normal 16-bit lightmaps
    for new map. Setting a value of 2 will default to 32-bit 2X modulate mode. 2X mode allows for overbrightening
    at the expense of reducing precision to 1/128 (from 1/256), which is still 4 times better than 16-bit maps.
    Overbrightening gives lights a little more "oomph!", a cheap man's HDR-like effect on lightmaps. With regular
    16 and 32-bit lightmaps that maximum brightness a surface can achieve is that the texture is seen fullbright.
    With overbrightening the texture can be up two twice as bright as the fullbright texture.
    When loading an existing map the editor sets the current setting to match that, so re-lighting/re-portalizing
    will generate the same type of lightmaps as the mission already had.
    Note that generating regular 32-bit lightmaps for Thief 1/G missions, without improved shadow softness settings
    or added colored lighting, isn't worth while. The original T1 lightmaps already have 8 bits per channel
    precision just like 32-bit, the only difference is that they only have 1 channel (monochrome).
    The active mode can be changed with the "set_lighting_depth" command. Preferably the "Tools" menu should have
    commands to switch mode (see reference MENUS.CFG), but for manual use the command takes following values:
       16: original 16-bit lightmaps
       32: 32-bit lightmaps
       64: 32-bit lightmaps in 2X mode

lm32_params <atten> [<sat>]
    (DromEd only) customize the initial 32-bit lightmap calculation parameters. These parameters are saved in
    mission files, and are restored when loading a mission file. Be aware of that loading a mission and then
    New:ing DromEd without a restart will not reset the parameters, instead the parameters of the last loaded
    mission will be remain active. The 'atten' parameter is an exponent that allows controlling the attenuation
    curve, pow(org_atten_val, <atten>). A value of 1.0 results in unmodified attenuation. The 'sat' parameter is a
    color saturation factor, where 1.0 results in unmodified colors. The defaults are 1.0 and 1.0 except when when
    loading older missions where they are 1.07 and 1.4. The latter values are an attempt to get 32-bit lightmaps
    to look slightly closer to 16-bit lightmaps when loading original T2/SS2 maps and re-lighting them. This can
    in some places come at the cost of some smoothness/quality. They will never look identical to the 16-bit
    lightmaps though, due to mathematical differences with the much lower precision 16-bit maps.
    Note that when running in Thief 1 "mode", the defaults are always 1.0 and 1.0 because 32-bit lightmaps will
    look identical to the original 8-bit lightmaps (as both formats have a precision of 8 bits per channel).
    console command: lm32_params [<atten> [<sat>]]

editor_water_alpha <alpha>
    (DromEd only) alpha value to use when rendering water surfaces in editor solid view, when hardware 2D is
    enabled, default is 0.5. The editor doesn't render water with the actual transparency as seen in-game because
    for editing it's more useful to be able to see all geometry, and opaque liquids like lava would prevent that.
    console command: editor_water_alpha <alpha>

show_mouse_xyz
    (DromEd only) show XYZ info for the current mouse cursor position in a 2D viewport

show_camera_xyz
    (DromEd only) show current camera coordinates

vm_rot_speedfactor <float>
    (DromEd only) speed factor (or rather step size) at which the camera rotates in editor, when factor is a
    negative value the rotation will be independent of current viewport scale so it always rotates at the same
    speed regardless of viewport scale (default is 1)

mono_console [<cols> <rows> [<xpos> <ypos> [<buffer_rows> [<use_colors>]]]]
    (DromEd only) enable built-in mono console window. By default WARNING and ERROR messages will be colored
    yellow and red, to disable coloring set "use_colors" to 0.

mono_quickedit
    (DromEd only) enable QuickEdit in mono console window (allows text to be selected without opening the window
    menu), however be warned that while in selection mode the entire application is paused until the mode is ended,
    so with QuickEdit enabled, accidentally clicking in the console window could pause the entire editor without
    being noticed.

disable_cycle_mwheel
    (DromEd only) disable mouse wheel support for cycle controls (the controls that have < > buttons)

return_to_listwindow
    (DromEd only) make list dialog for Conversation edit and Mission/Quest/Gamesys parameters persistent until
    closed, so after closing the edit window after editing one of the list items, it returns back to the list.

fam_search_missing <fam list>
    (DromEd only) enables editor to search for missing family textures in other families, listed by this config
    var. This config var should only be defined temporarily to load a mission where family textures have been
    reorganized, and then save the mission with the updated textures and exit dromed again and remove this config
    var. Note that replaced textures will end up as single loaded (same as "load_a_texture" command) unless the
    new family already happend to be added originally.
    For example: fam_search_missing wood+mech_new+stone+walls

allow_wr_tex_assign
    (DromEd only) enable hack to assign textures in solid view directly to WR faces, i.e. emergency fixing of
    stripped missions. It works the same way as regular texture assignment in solid view, except you have to
    hold down ALT. Do NOT use this functionality for editing unstripped missions.

sdesc_decimals <2 - 8>
    number of decimals to use when formatting sdesc based float values, default is 2. Sdesc based values are
    the ones seen in many editor dialogs, like object properties, or values exported with export function in
    the object hierarchy dialog or commands such as "dump_tagblocks_vals".

mouse_smoothing <0 - 30>
    enable mouse smoothing for mouselook. Default is 0 (smoothing disabled).
    The value specifies the number of frames over which smoothing averaged, 0 or 1 results in smoothing
    disabled. A good starting point is 10 frames.

mouse_smoothing_weight <0.0 - 1.0>
    specify the weight factor used for weighted average calculations for mouse smoothing. Default is 0.5.
    A value of 0 results in no smoothing, 1.0 in unweighted average (i.e. input samples of all frames are
    equally weighted). The lower the weight factor is, the more the lag effect is reduced, allowing you to
    balance between lag and smoothness. The number of frames specified with "mouse_smoothing" of course
    also affects the lag amount.

use_raw_mouse_input
    use Windows raw input functionality to get mouse input, instead of the standard Windows input. The raw input
    is only used for mouse look and not for cursor movement. This can be useful to get full precision input from
    high resolution mice.

raw_mouse_sens_scale <xscale> [<yscale>]
    specify custom raw mouse input sensitivity base scale when "use_raw_mouse_input" is enabled. Default is 1.0.
    If only xscale is specified then yscale will be set to the same value. The game's regular mouse sensitivity
    settings, that can be configured in the options menu etc., still apply to raw mouse input, but if the mouse
    resolution is high you may want to set the raw input base sensitivity to something lower, so that the game's
    sensitivity range is more usable. You'll have to try different values, but a starting point is the formula
    400/mouse_dpi. For example if you have an 800 dpi mouse, you can try a sensitivity scale of 0.5 (= 400/800).


mouse_sensitivity_use_aspect
    this is actually a bind var ("user.bnd") and not a .cfg var. It allows to turn on/off the old behavior
    where the screen's aspect ratio affects the horizontal mouse sensitivity (making it differ from vertical
    mouse sensitivity).
    The default is 0 (disabled).

mouse_sensitivity_y_scale
    this is actually a bind var ("user.bnd") and not a .cfg var. It's a scale factor, applied on top of the
    general mouse sensitivity, for vertical mouse movement (mouse_sensitivity * mouse_sensitivity_y_scale).
    The default is 1.0.

crouch_unmount
    this is actually a bind var ("user.bnd") and not a .cfg var. It allows to turn on/off the new alternate
    action of the crouch and crouch hold keys that aborts climbing, so it's possible to drop off a ladder or
    rope without jumping.
    The default is 1 (enabled).
