On low precision in darker colors being a function of gamma...
There were issues in Quake/Quake 2 with bright areas not being quite right, so some measures were taken to try and achieve some form of overbright lighting [1]. Especially in Q1 where the light baking did not perform bounces/radiosity, there were a lot of black spots with adjacent highlights for dramatic effect, and you could see some banding there with the OpenGL renderers, which is more apparent with lower texture settings and pumped gamma.
But also, as a side effect of this overbright ability, lighting got flattened even more in GL [1]:
>The original quake (software) engine used overbright lighting, which means the lightmap brightness can go up to 200%. So, light.exe creates BSP files with lightmaps that go up to 200%.
>In glquake […] there is no overbright lighting, so every part of the lightmap that goes above 100% is flattened to equal 100%. This is why glquake looks just fine in darker rooms, but in bright areas the lighting looks very flat.
>[Some engines] restore the overbright feature of the software renderer, so that lighting hot-spots will actually be as bright as intended.
A similar effect was pursued further in Quake 3 with r_overbrightbits/r_mapoverbrightbits, which fakes a higher dynamic range through futzing with the hardware gamma ramp and bit-shifting/normalizing color channels [2], achieving highlights that pop a lot more... at the expense of introducing even more banding in dark spots.
He mentioned it in passing in a .plan in 2000 that touches on higher precision buffers [3]:
>Q3 hacks the gamma tables to sacrifice a bit of precision to get a 0.0 to 2.0 range, but I wanted more than that for even primitive rendering techniques.
Eventually, in his .plan dated january 29th, 2003 [4] he appears convinced,
>The future is in floating point framebuffers. One of the most noticeable thing this will get you without fundamental algorithm changes is the ability to use a correct display gamma ramp without destroying the dark color precision.
One other aspect of 3dfx looking a bit washed out beyond the gamma problem is the trickery performed in the RAMDAC before it blasted those cool pixels to the VGA port.
If you remember, 3dfx screenshots were just as dithered as, say, an ATI Rage, but on screen it looked smoother (although a bit blurry) while the others looked visibly dithered still (see some 3dfx screenshots https://www.vogons.org/viewtopic.php?f=46&t=36548 )
Part of it is the gamma shenanigans mentioned, and also the blur filtering pass taking place in the DAC which "upgrades" the image (eventually claimed to be so-called 22 bit precision for Voodoo3), introducing some quite visible smudging and hints of unexpected colors as a result.
Earlier this year I studied a few of the Amiga portraits and yes, they reduced VGA to fit in 5 bitplanes (32 colors), but they actually don't even use all of the colors. I found them to be around 20-25, keeping _more than a few_ aside for, say, the GUI.
Except in close-up chat mode, from memory all you get is text and a background color for the text, and not much cursor or anything else. So I'm still wondering why they didn't allow themselves a couple more color indices, which can go a long way even without dithering and anti-aliasing (which there isn't too much of in the VGA portraits). Maybe it ended up compressing a bit better.
I imagine you are vaguely familiar with subtractive synthesis; oscillators, filters, envelopes, modulation. It wouldn't hurt, although the building blocks are laid out differently in an FM synth.
Anyway, I found these hands-on demos on the DX1 to be some of the most "aha" moments I've had trying to understand how FM patches are made.
TLDR, an FM "operator" is an oscillator. A carrier is the oscillator you hear, its modulator is an oscillator that wiggles the carrier and brings out different harmonics. You can stack more modulators and you can have multiple carriers going in parallel, the various combinations of operators available in a synth being the different algorithms.
For some more complete gameplay, I would suggest a mission pack that was actually produced and finished by a modder for the original Quake 4 engine [1]: False Dawn, released 2014 (https://www.moddb.com/mods/false-dawn)
This is a well built "episode" of a few levels built by a one-man team. He even recorded some voice lines. The mod borrows some shader effects from Sikkmod, but actually uses them tastefully.
This is a finished, polished product and IMHO more worthy of your time.
(I was not involved in this, but discovered it a couple years ago and was very pleasantly surprised)
[1] OpenQ4 is not binary compatible with the original game libs
"finished" is a bit misleading, I'll try and explain.
Long time Quaker. I was quite involved in Q4 Multiplayer (in fact, I adapted and rebuilt a Quakeworld map for Q4MP a year ago "for fun", and got to see what the idtech4 modders have been up to).
Every now and then, I replay the SP campaign and every time, I am reminded of why I forget so much of it. The expansion, despite some interesting twists, was shaping up to be very much the same, and you can see why it was canned.
Some of its completed parts were rolled into Q4MP when id Software had taken over maintenance from Raven
I played the Ritual expansion as-is when it was leaked a few months ago. Impressions:
- First couple levels had gone through enough art passes to be about as good-looking, if not better in places, than Raven's. As an aside, from the levels and the new creatures/weapons, you can definitely tell Ritual had the chops to pull it off.
There were some awkward sections to navigate (cramped corridors where the marines would just block you), probably little playtesting had occurred - makes you wonder when they consider a level works well enough that you can art it up, because building levels in this engine for hardware of that era is somewhat tricky.
- Level quality then progressively degrades to what you expect from alpha. Not necessarily wholesale, but maybe a quite fleshed-out section is poorly connected to the next set piece, encounters are weird all over, etc.
- You acquire powerful weapons and upgrades way too quickly, including the new ones - I assume this was made to be able to get a feel for the whole thing quicker, before playtesting and balancing passes.
You even get the new guns, one of which (I can't remember) was essentially a different look of a gun you already have, with different ammo (no wonder I forgot what it was).
This is something I also remember from Q1/Q2 expansions: you could totally do without some of the new guns. You're just happy you get a "spare" of something else for bullet-sponge enemies.
- In this game, you remain a regular marine. You don't get the base HP, stamina and running speed enhancements from being stroggified (they couldn't have pulled this twice, understandably). So to veteran Quakers, this would have felt even more Call-of-Duty-Quake than retail Q4 already did.
- Retail Q4 had some quite tedious vehicle sections. There is still some of that in the expansion (good and bad), but Ritual tried some interesting twists, and some of these sections actually could have been an improvement over Q4's.
Unfortunately, this is particularly sensitive to pacing IMHO and the alpha suffers from pacing problems everywhere.
- Placeholder content everywhere, including in the more complete sections: barren cutscenes to try and get some time allocations here and there, temporary voiceover/animations or none at all (you wouldn't burn too much money on mocap at this point).
One of the vehicle sections, you're a gunner on a ship. Every now and then you have to get out of the turret and perform some task in the back. The task, in alpha state, is essentially click a GUI and go back to gunning down infinite enemies, and you're not sure how your ship is still going somehow.
- New monsters and bosses - except they're super buggy. Interestingly enough, some of the jank carries to enemies already existing in the Retail game: either they made it worse in code, or missed some supporting level design elements, or started working off an incomplete pre-release Q4, and this is prior to them merging in some of the fixes Raven would have made.
- You start getting some new enemies early, but since you also have access to too much firepower, you barely notice they're supposed to be tough. You can still get near one-shot by some basic troopers thanks to the wonders of Alpha Jank TM, though.
There's one miniboss encounter where the door opens right on him, and you can get insta-killed at range because you turned left instead of right trying to get cover in the maze of boxes facing his closet. If you turn the other way, you can just trivially cheese him behind a pillar (except when the collision is bad, and you get insta-killed anyway).
- Another thing from Retail this suffers from is, your cohorts seem invincible and their rifles are better. So, no need to bust your ass: you can just let them work (when they don't block you).
Except since this is alpha, sometimes they die in weird ways (maybe the game designers were trying to make them more important?), but the game doesn't care (yet), and you can continue with events that would depend on them being triggered :)
- Final level was so janky in all of the ways I just had to go through it with /noclip and /god to see the end.
Some of these problems can be addressed in code, and going by the moddb page, it seems this is what has happened in this fixup release of that vibe-coded Q4 source port, quoting:
> Where the alpha's own code or content stopped something working as designed, openQ4 fixes it.
Back in the day, people weren't afraid to use the SDK to modify game behaviour you couldn't alter by just changing interpreted scripts. It was the most accessible, you already knew how to program.
This somehow got lost, so recent "mods" largely remain at a basic "I changed this gun, altered a monster".
It is now possible to vibecode yourself out of this, and thus we get a more playable alpha. But it is by no means _finished_.
You could probably generate some better-than-placeholder voiceovers and music with AI now.
The big problem before you can call this anywhere _finished_ was, and remains everything else pertaining to content:
- You need to take editorial, game-design decisions and run balancing and pacing passes on the entire product.
- You will need to get some good animators to invent what is barely there in the alpha expansion.
- You will need to complete some of the new models, etc.
Textures and modeling were a generational upgrade in tech4, leading to the foundations of the current art pipeline (high-poly -> low-poly+normal maps, diffuse, specular).
You barely saw any of this at the amateur level, and when you did it often wasn't pretty.
Nowadays, hobbyists have caught up, and i.e. in the D3 modding scene, they endlessly remaster the existing models.
- You will need to run design and art-passes over most of the levels.
Levels were tricky due to the tools and increased difficulty of texture making from reasons above. They started using more static 3D models, so you needed to level-up there too.
Back then, it was comparably easier to make new levels, and you'd try to reuse stock assets as much as possible.
You probably didn't understand how the lighting model worked and suffered performance and/or visual issues, assuming the 60fps cap and hardware upgrades would sort you out. That part has not changed.
Nowadays, the level editing tools have changed the least. No Trenchbroom for you, and vomiting meshes out of Blender only takes you so far. It has remained the most tedious part IMHO.
Unsurprisingly, the tech4 diehards have output very, very few new levels, and at comparably low levels of quality.
This is unfortunately a very, very visible aspect of this alpha.
This right here should be the top comment. Everybody seems hyped about this "long lost" expansion that was full of promise, but what we see in the alpha was far... very far from it.
And it's sad that most people fall for the superficial take of "12 new levels" of a nearly-finished expansion, but as you correctly pointed out (in great detail, thank you!), there is nothing once you look beneath the carpet.
As a reminder: This expansion was supposed to bridge the Strogg Quake 2/4 universe with the eldritch Quake 1/3 but what we see in the alpha is a far cry from that. Overall, this looks like an expansive fan expansion with ambitious ideas but struggling with the id tech 4 technology.
I remember the time when id tech 4 was made public and we were hoping for a similar revival of the games built upon it but id tech 4 was an engine that was built on old hardware foundations and tooling that was geared exclusively towards experts to make it work.
> This expansion was supposed to bridge the Strogg Quake 2/4 universe with the eldritch Quake 1/3
Wow, I had no recollection of this. I certainly had no idea whatsoever when playing it!
From the alpha, Awakening just seemed like a second take on the same "lone marine saves the day", but happening within hours of the first.
Story-wise, it severely undermines the impact of what you do in the main campaign: the Strogg can just spin up another Makron like that? Pff... way to deus ex-machina your parent product.
Maybe that's part of the reason(s) the project was canceled!
Yeah, I could imagine that at this project it simply did not show enough promise to carry a game that sold below expectations. In contrast, the D3 expansion had some gimmicks like "hell time", "hunter bossfights", "Gravity Gun" and Awakening had... a freeze-thrower?
If you are curious about the whole premise of the expansion, the following video summarizes it nicely. Or perhaps this is all just speculation, it's difficult to know for absolutely certain: https://www.youtube.com/watch?v=E6R99ShYpZU
My fork of DarkRadiant focuses on experimental editing, including editing in the 3D view like in TrenchBroom, as well as mouse wheel extruding (Cube-like), procedural tools, wall/floor tooling, etc. It also fixes all issues with Q4 and Prey. You should check it out:
I've been maintaining and adding new features to DarkRadiant as well, which aims to be more stable than HF and less experimental. NetRadiant-custom (https://github.com/Garux/netradiant-custom) also supports id tech 4 to some extent, and WifeRadiant (https://github.com/erysdren/WifeRadiant) aims to improve that support soon.
In the making of Bravado Remix (https://imgur.com/gallery/bravado-remix-quakeworld-map-quake...) I did end up using both DarkRadiant and NR-Custom, the latter I have recommended here in the past.
I have no firm intention of doing another such project in the near future again, even though I am now familiar with the texture sets :D
But in general, these tools are the least evolved. Outside of the Doom, Q1 scene and maybe a couple of related engines, "modern" level designers are just used to throwing models around and counting on engine/hardware to manage culling for them. In the stock Q4 engine, this doesn't scale well, especially not at high fps for competitive multiplayer. I shouldn't need a CPU with a ton of threads and a 4080 to run a map which, in ultra settings, still looks worse than what we were doing 20 years ago (actually, the 4080 barely matters. A 1070 would still be twiddling its thumbs, the bottlenecks are with the CPU). I shouldn't have to kill all light sources and impair shape and depth perception to get good performance, which was the thing in that game.
Anyway, DR dropped support for a number of Q4isms, namely declarations in sub-folders (why the hell would you remove this) and material guides, a form of template with macros added by Raven to simplify common texture definitions. As a result, working on Q4 files in DR has by default some missing items (entities being defined in sub-folders of /def) and missing textures (a bit of this from materials in sub-folders, and lots of .guides). So I had to write a little script to just expand all the materials etc. just so I could open a map and be able to place stuff.
It quickly became apparent that DR was annoying in a number of other ways so I ended up settling on NR-Custom as its complement for refinement and texturing, despite its (at the time) broken handling of grouping/ungrouping in func_groups/statics (losing the origin, awkward texture locking etc)... so I also had to lose the use of layers, which although super useful, are a purely DR construct and require its own file format.
Good things for DR:
- Editing lights and 3D preview are nice.
- Material preview and editor
- Hot-reloading of images, materials, models
Good things for NR-Custom:
- VERY usable texturing tools, somewhat Trenchbroomist; also some of that in the shaping and cutting up of brushes. Great time-saver.
I'm taking note of that feedback for DR. We currently have an active thread to prepare a 3.10 release on the The Dark Mod forums, so I invite you to join us if you have any bugs to report.
Oh, and I also invite you again to try HellForge. I've fixed most of the issues around Q4 and Prey mapping. They're not stable yet to go into upstream DR, but it's good. Oh, and you can quickly import and re-scale any Q1, Q2 or Q3 map.
Skin tight is fine. Clothes, depending on cuts and styles, will drape over the shape underneath in different ways that don't necessarily represent what the shape is.
Thus, your understanding of the shape can be misled, and your attempts at adapting a pose to something slightly different, or coming up with your own (let alone dressing them up differently) will likely be stilted and wrong in a number of ways - because you're trying to imagine what's underneath without understanding that there is a muscle belly there that doesn't bend that way, or a bony protuberance being an anchor point for other landmarks, and what have you.
Imagine "learning anatomy" from studying a cohort only wearing the ultra baggy styles that are around currently.
You don't have to know all the names for all the parts... but you will probably get better results by learning from the ground up then inferring how the outfits drape over that.
1 data point of Yep.
I went to my local Monoprix to grab a few things during lunch break. For some reason they seemed to have suffered a fridge incident, good for me I didn't need anything from those aisles. I did wonder if it was just a recent problem or if a more sneaky malfunction had silently semi-spoiled some of the products I brought back days before, until it just blew over.
Let's see. Yeah, the lighting is good
- Location isn't quite right. The side gardens on the Champs de Mars are much wider than this. They're not in this style either :)
- Melty coins, and keyboard.
What else?
- Bunch of visible shape inconsistencies: building window sizes and alignment, fence railing, rivets on the bench, coins, accordion features, perspective violations etc...
- Metal clips on the bottom and top edges of the carrying case are on different sides. Said clips look inconsistent.
- One of the case straps originates inside (from under the velvet, even), the further one originates outside. The carrying handle is oddly mis-centered (as with most things).
- I'm not sure that accordion, compressed, fits in that case. She can just about stand in it (diagonally, on one foot).
- It has weird folds that disappear partway through and turn into adjacent folds going the other way. Between this, the keyboards and the dots, I'm sure there are other things wrong with accordion anatomy.
- Awkward left hand: not sure what's going on with the metacarpal joints, and index finger appears to begin forking into two tips pressing into two (misshapen) buttons. Her right hand is getting there too.
- Weird hair midline. I haven't seen a "sub-midline" like this in the wild :)
- The burgundy coat clips through the bench under her right leg. It's like the front slat goes through a hole in the side of the coat.
- We don't usually have that side curling bar on each side of the sitting area. It looks like it's clipping through the flats next to the leather handbag
- Ill-placed buckle on that bag.
- The green metal fence on the right begins behind her for one vertical bar, next one disappears into the plants, no further vertical bars, no top horizontal railing. Actually, it appears to only have fencing on one side, that's not something I'd see often. On the side where it does have a top, its top rail has very irregular width.
- Sand looks like... breadcrumbs? Very repetitive coarse shapes, like it was poorly done with a clone stamp.
- Something weird about the leaves.
- The foremost lamp isn't sure whether it has flat sides or not.
- The other lamp emerges out of nowhere from behind one of the blob-people in the alley.
- A bunch of unfortunate tangents and occlusions which make you wonder whether the diffusion hallucinated detail out of bigger shapes, or if it's been trained on photos where the tangents were done on purpose.
ie. center bottom bar of the bench being exactly aligned with the gray edge tiles, side curly railing-not-present-on-our-real-benches fully formed on her right, almost completely foreshortened out of existence where the bench's perspective forbids it...
Most people are fooled by much cruder fakes though. This is a lot better but keeps plenty of tells for observant people (and most people aren't)
When I was a kid, there was a game in the funny pages that asked you to find six things wrong in a picture. Someone should run a contest like that for AI-generated images.
It might be an interesting benchmark for vision models, too.
There were issues in Quake/Quake 2 with bright areas not being quite right, so some measures were taken to try and achieve some form of overbright lighting [1]. Especially in Q1 where the light baking did not perform bounces/radiosity, there were a lot of black spots with adjacent highlights for dramatic effect, and you could see some banding there with the OpenGL renderers, which is more apparent with lower texture settings and pumped gamma.
But also, as a side effect of this overbright ability, lighting got flattened even more in GL [1]:
>The original quake (software) engine used overbright lighting, which means the lightmap brightness can go up to 200%. So, light.exe creates BSP files with lightmaps that go up to 200%.
>In glquake […] there is no overbright lighting, so every part of the lightmap that goes above 100% is flattened to equal 100%. This is why glquake looks just fine in darker rooms, but in bright areas the lighting looks very flat.
>[Some engines] restore the overbright feature of the software renderer, so that lighting hot-spots will actually be as bright as intended.
A similar effect was pursued further in Quake 3 with r_overbrightbits/r_mapoverbrightbits, which fakes a higher dynamic range through futzing with the hardware gamma ramp and bit-shifting/normalizing color channels [2], achieving highlights that pop a lot more... at the expense of introducing even more banding in dark spots.
He mentioned it in passing in a .plan in 2000 that touches on higher precision buffers [3]:
>Q3 hacks the gamma tables to sacrifice a bit of precision to get a 0.0 to 2.0 range, but I wanted more than that for even primitive rendering techniques.
Eventually, in his .plan dated january 29th, 2003 [4] he appears convinced,
>The future is in floating point framebuffers. One of the most noticeable thing this will get you without fundamental algorithm changes is the ability to use a correct display gamma ramp without destroying the dark color precision.
[1] https://www.quaddicted.com/engines/software_vs_glquake
[2] You can probably read the C/OpenGL code better :)
[3] https://github.com/ESWAT/john-carmack-plan-archive/blob/mast...
[4] https://github.com/ESWAT/john-carmack-plan-archive/blob/mast...
reply