Any way to get default scale to be 1? Also, any way of turning off texture lock? 

Any way to get default scale to be 1?
Settings->preferences->brush settings->default texture scale

Also, any way of turning off texture lock?
You should have toggle buttons on your top bar for texture lock and texture vertex lock (???)

By the way are you starting the program up in quake 1 mode? 

Texture lock and vertex lock are the little lock icons on the toolbar or bindable, Edit->Shorcuts and you can search by typing.

Caulk .. You can middle mouse click the texture in 3d or select texture in texbrowser then "shift+a" to select all brushes with that texture. Not sure if that's what you're after there.

Some of the binds aren't ideal for constant use, "unselect" I bound to where I rest my hand.

Scale you can set in Edit->Preferences

Handy tip.. right clicking the toolbar icons (caulk etc) applies that texture to selected brush. Same thing can create func_group and make detail/structural. 

Kinn, starting in Q3 mode.

WW, the middle click thing shift a thing is interesting, not sure what I will use it for yet but its neat.

You guys have been genuinely helpful. Any way of getting netradiant to show brush edges? I like removing them but I work mostly with them on in TB. 

Camera->Render Mode->Textured+Wire
also Display->Textures->Render Mode->Nearest for nice pixel units 

overpronounced 3D workzone @negke: it is distracting imo too; how to make it being not and recognizable over any background?
- Ability to select back-faces of brushes in 3d view
in which mode?
- Clip tool allowing brushes to be split
m1 doubleclick either, besides respective command
- Smart selection of brushes in 2d viewports
SleepwalkR's strategy appears to handle more cases well, comparing to depth sorting; will try to think of it, not a simple change
- Abilty to select and move caulk faces even when caulk mode is on
i'm strictly against of selecting/editing hidden stuff; though ctrl+shift+a and QEtool resize modes do select them
- Object orbit in camera mode
one more orbit is on Tab
- Ability to drag and drop entities
where, from where, why?
- moving objects in 3d view locked on the y-plane and have z-axis as a modifier (and then locked on the y-plane)
what is y-plane? isn't it how it works now?
If I'm using the face tool in 2d view to move and I want to deselect it takes 3 escape key presses? I cant even just click a different brush to edit a face?
normally you need face components mode only for complex moves; QEtool m1/alt+m1 should do most things;
alt+m1 objects selector was a quick hack to get this ability w/o changing and confusing whole stuff; might worth to try to make component modes nonmodal, as overall if feels like an imporvement (not w/o a couple of drawbacks tho) 

https://www.youtube.com/watch?v=QeiDOeAwAkw 

 

https://github.com/Garux/netradiant-custom/releases/tag/20190705
With UV Tool and Autocaulk. 

Selection highlighting in camera window doesn't look quite right in the latest version: https://imgur.com/VuA6E47 

Which OS/card/driver is this?
We had the effect with ancient Intel @ windows, solution was to disable it (menu.view.show.3d workzone).
Does it look alright in previous version?
https://imgur.com/S9ratEt 

Yes, it's an old Intel HD Graphics 3000 on Windows 7. The previous version installed on this machine was from May 2018 so it didn't have the 3D workzone feature. All the download links on GitHub seem to point to the same version.
I've disabled the workzone display for now.
Odd that a seemingly simple feature like vector lines would break like this, regardless of (older) hardware. 

I see. The not simple part of the feature is specific drawing, when flush with surfaces plus depthtested.
So there are polygons with glPolygonOffset and GL_EDGE_FLAG_ARRAY; the latter isn't popular and likely not handled by old Intel driver. 

I tested the UV tool, and it's very cool - allowed me to fix some areas that were very very hard to do previously. 

https://github.com/Garux/netradiant-custom/releases/tag/20200320 

Currently testing it, had some build from 2018 before.
Has realy some nice additions to it, as far as i can tell.
Just as Kinn said, the UV tool is awesome and very useful.

To use it as a portable installation is also very useful, at least for me. 

One thing that came just to my mind:
Would it be possible to add a model browser to radiant?
In my imagination it would work similar to the texure browser, just that it loads previews of the models.
Next to the Entity Inspector, Console and Texture Browser it would make a fine addition, especialy as there would be space for some more tabs, methinks.

About that included version of q3map2, 2.5.17n, could it be that the lightmapping of models is broken?
Or has maybe something else about the behaviour in this regard changed?

Neither with spawnflags 4 nor the appropriate shader including q3map_forcemeta i seem to get my models lightmapped, only vertex lit. 

Model browser would be worthy thing, though i'v got no good notion of optimal implementation yet.
One idea was preview widget in open dialog to show selected model.
Other was close to your one: FS tree widget, showing folders, containing models. Selection change = model insertion or change selected misc_model. This one avoids adding extra viewport to render and navigate.
One more option is to have a map with all wished models to view and copypaste (e.g. netradiant-custom\q3pack\maps\grabmodels.map).
One more option is support drag and drop of model file to editor window, which particularly would make creation of preview .map faster.
Your idea looks most cool so far, though there are few questionable topics.
It's quick to preview small prop models with lowres textures, but there may be also detailed models/using a lot of hires textures/representing whole level. Possible solution would be to cache preview thumbnails (questionable points here are good angle to have representaive preview and lack of navigation around model).
Other thing is FS view: it definitely must show only folders, containing models, but how would it know, that you added a folder/model to some folder? Rescanning/recreating each time is potentially slow, showing whole FS may be expensive with a big one too.

Lightmapping of models: in general anything can be broken (furthermore there was massive refactoring of q3map2), but it works here. Can you share an example setup to reproduce the problem? 

Spawnflags 4 / forcemeta does actually work, incidentally i tested a array of models first which seem to not work properly.
It's more likely now that these models are "broken" in some weird way, due to my lack of modelling skills when i created them.
Could also be something with the darkplaces engine, the more exotic compile settings i use or any sort of things really.
So my post was a bit assumptious and careless in that regard, sorry.

Does q3map2 has requirements any model should idealy meet?
Like beeing closed, vertex colors, normals that sort of thing?

As for the modelviewer: One solution for the FS view might be a file similar to shaderlist.txt, eg "modelfolders.txt", in which a mapper adds the folders to be shown manually. 

There are no special requirements to models like that.

Limiting folders selection smells like an optimization indeed. 

I have created a small testmap with two rows of models, one row without spawnflag 4, another with.
Might still be interesting to mess around with.
It's a .pk3 file, includes the models and .map file.

Link:
https://drive.google.com/open?id=1gJj8gTtSKy1TAVABwGh2xbnkXqtWAFrt

As i mentioned some of my models are quite the mess, they were photogrammetry experiments that i havent come around to properly clean up yet.


In regards of optimization regarding the modelbrowser thing, what are the lowest specs you are aiming for?
I am using huge folders with pretty large textures for some 8-9 years now, with a few hundred mb each.
They load quit fast imho, that beeing on a normal hdd. 

I exported your .bsp to model with lightmaps, models are lightmapped: https://imgur.com/0hemZen
Even those, which you were not asking for :) This is known issue: > 1024 vertices push model surfaces to meta routine, which enables lightmapping.
Regarding to lighmapping quality (visible seams), this is general q3map2 problem. Enabling phong shading should help, also may be -trianglecheck for 'light bleeding' issue.
This should be improved once we get what's going on there, at least that 1024 limitation should be applied to separate surfaces of model, not a whole.

Loading folder with 137 512-1024 textures take a few seconds here, which seems to be realistic case for models. This is slightly too much imo, considering that models have to be loaded too.
I found some 50-100Mb models laying around in setup here. This is another topic, if to/whether it's possible to quickly evaluate if preview loading would worth the time spent/memory consumed. 

Got build with some working model browser:
https://cdn.discordapp.com/attachments/464535987993706518/703607488087457933/netradiant-custom-wip-20200425.7z
https://cdn.discordapp.com/attachments/464535987993706518/701421072611803146/unknown.png
That was a bucket of numerous small and big challenges there. 

Wow, very nice work!
Currently testing it, looks and works exactly the way i imagined.

The browsing through the model folders works fast and smooth, its nice to be able to rotate the models in the preview, and very cool to flip through those models with the misc_model entity selected.
Havent encountered any problems so far.

Anything specific you might want to have tested? 

It's good to know, that it works well with your practical cases.
Just tell, if something is not working, as expected.
I'v got a few questionable points:
-loading whole VFS here (root with max depth)(120k files) takes time, perhaps it makes sense to only load on 1st browser rendering
-there are some huge models laying around here, may be should load model file 1st to avoid loading, basing on the size (this is quite rough property tho)
-treeview expands quite widely and doesn't minimize, may be there should be draggable separator to adjust width and treeview itself scrollable horizontally