Utilized Mark V v1.99 DirectX.

Even bigger problems with fps occur at Rubicon Rumbke Pack. Same mouse delays , fps down to 20-30 in some areas. 

External scrolling sky textures are usually comprised of two images, not one:

sky1_trans.tga (transparent overlay)
sky1_solid.tga (opaque underlay)

There are another two possible suffixes in some engines, but I don't remember the name. 

MarkV Direct3D is not the same as DirectQ, and you shouldn't expect the same performance.

To be specific: the reason DirectQ gets good performance is not Direct3D. It's batching, vertex buffers, multitexturing, shaders, etc, and this is all stuff that can be done in OpenGL too, if the engine author is willing to do so.

In the case of MarkV, Baker is not willing to do this. So in order to run in Direct3D, it needs to store out vertex data per frame, then try to mangle unoptimized OpenGL calls into something that works in Direct3D. You shouldn't be surprised if it doesn't always work well.

If you want an engine that has good performance in big maps by using the right kind of code, MarkV is not that engine. Use QuakeSpasm instead. 

Speaking of which, any chance QuakeSpasm will ever support Nehahra? 

Quakespasm is not in active development ATM. So there's that little issue. 

True, but FWIW folks shouldn't read that as "dead project". Updates were still landing in the codebase through the end of 2018. So nothing going at the moment, but Quakespasm development seems to come in spurts.

Dunno if those developers have ever commented on the likelihood of Nehahra support though. 

Poor Baker of inefficient coding, he might never return, man. :P We need him! 

Someone said somewhere that Nehahra is too much of an oddball to support properly. Too bad for me, haven't played it yet and Mark V doesn't run well on my system. 

In mark_v.exe (not _winquake), if a water brush is not at least overlapped by at least one unit by a non-water brush on one face, it will not render as transparent unless you activate r_novis 1. With developer 1, it claims that the level is not vised. In mark_v_winquake.exe, it does see the vis information and properly renders the water. 

I've played through Nehahra recently with Mark V and had no personal issues at all, just specify -nehahra in the command line. The fog and pretty much all graphical features work, the cutscenes do not mess up the HUD afterwards unlike the original executable, the custom music will not play however, maybe it doesn't support xm files but the cool outro music on epilogue texts between chapters does play. 

There's some funky things Nehahra does that technically work, but that it really shouldn't do. In other words, the kind of thing that normally comes back to bite you in the ass at some later date.

Two of the big ones will be that it uses a modified protocol but doesn't change the protocol version number, and that it relies on sending stuffcmd every frame for certain things rather than the built-in messaging or reading options from worldspawn. 

If you use winquake with vid_stretch 1 or 2, water in dopa has a bunch of lines through it.

https://imgur.com/a/8xyVixJ

To correct this you can set r_wateralpha to 1.

Figured I'd mention this in case it hasn't already been brought up. 

is normal behaviour on those maps which support it (the original game doesn't as standard) 

Ahh I see. Shouldn't winquake default this cvar to 1 automatically though since it doesn't support transparency? 

The lines you are seeing are the transparency. It's exactly the same transparency method used in old games such as Sonic 1 for the Sega Genesis.

However, I agree that it doesn't look good in MarkV. A stippled pattern would look better, as implemented in old versions of Makaqu. 

Shouldn't winquake default this cvar to 1 automatically though since it doesn't support transparency?
Stock WinQuake doesn't support translucency, but several engine mods have added it, of which MarkV is one.