We can start with a bare image, with the default engine processing. The one change here is that interior lighting relies more on direct sources than it does in vanilla. I also can't just disable all of the high detail textures. But the overall look when it comes to the image shaping is still vanilla. This is what a regular FO4 interior looks like.
I don't see it much, so it's like... shockingly hideous to me lol
But I guess it isn't terrible. The lantern does throw some direct shadows out, with a slightly rotating source flicker to boost it. But things are looking pretty flat, especially in the highlights. All of the stark shadow transitions make low-polygon objects look like cardboard. This is a SUPER 'video-gamey' looking game. Something I don't appreciate often enough with how far I've deviated from the base look. Lets put some ENB on it.
Now, we have some ambient occlusion for objects - things are no longer uniformly baked in lantern ambience and there is better separation between objects as well as a better sense of polygonal detail across surfaces. Transitions from dark to light are easier, with very distinct background/foreground separation. Light has contour. Tone shaping and LUT grading give a darker but more natural image profile that is far easier to look at, which is further aided by upgraded adaptation and bloom algorithms, as well as some slight DOF that all comes together to give a better sense of light diffusing across space. We even get some tiny local shadows around some smaller contours. Not nearly as bright and in your face as vanilla, but far less flat. This particular look is not what everyone likes - many go higher contrast with this stuff but personally I like a smoother image for fewer complications and less eye fatigue. I also happen to think it just looks better toned down rather than up.
FO4 is a game easily overdone with stacked post-processing. Many do not have the self-control for it IMO. Go peep some screenshots over on the Nexus, check out some of the reshade presets shared on there. I don't wanna throw shade(lol) but you'll see a lot of examples of how easily it goes overboard. And you yourself do not see it as well as other people, because your impression is niched-out by staring at it as it gradually changes - you adjust and it looks normal. I've taken a lot of time to understand what every aspect of what I mess with is actually doing, and it has made me increasingly more conservative with everything. I have a built up intuition for where I'm at with the image, without needing to hone-in and really *see* it to know what's happening because I understand how what I'm changing works. You have to accept reasonable goals for the end result, let things fall into their own place and it'll look consistently better in the end.
After this, we put the icing on the cake with an improvised take on ray-traced global illumination. I use ReShade to bring in RadiantGI and RTGI, which cast rays in order to draw new shadows and light onto scenes in real time, quite a lot like the real thing, though nowhere near the same quality levels. RTGI is currently set to be pushing out 6 rays per pixel, which then branch out on 18 raymarching steps per ray (which is part of what takes it from ray-tracing to path-tracing). So it really is calculating a lot of lighting and the effect can be pretty dramatic.
It almost looks like too much, but that's because I'm running RTGI in 'still' mode, which cranks quality but doesn't have the same amount of filtering as the real-time mode yet, causing it to over-accent close to some edges.
But you can see that transitions between lighter and darker surfaces have changed yet again... only this time they follow the light angle set by the lantern that is illuminating this little corner. You have more shadows drawn on the parts receiving less light from the lantern all over the screen - even on the gun. You have bounce light reaching over the top and sides of the table, climbing over the terminal, and even between the two wooden file cabinets. It's doing as much of that as it is determining what actually is not receiving light and darkening it accordingly. Some things blend more, while others stand out. You can also see how everything has taken on more of the color qualities of the surrounding light. Everything caught by the light now glows slightly against the contrast thrown by new shadows, with all of it warming up the color temp in accordance with both the yellow-orange lantern and the pinker sunset ambience flatly filling the room.
These plugins basically 'learn' that information by shooting rays out from the camera across what the engine tells it about the depth layer. The depth layer can be represented in the form of a 2D image not unlike a normal bumpmap for a texture (the reference image that gives it the appearance of having 3D surface contour under light,) only it is really just an abstract matrix generated organically on a per-frame basis, after mesh collision and visibility within the screenspace has been calculated, basically revealing the on-screen Z-axis to any 2D raster shaders that work dynamically with camera-based depth. It's that crucial 3D to 2D conversion needed for so many basic processing tasks handled by game engines, letting those 2D shaders 'understand' the shapes and contours that the camera sees in a given frame. It's essentially baked physics for the render pipeline to use, like tracing paper to help the shaders draw better and smarter. It even contains information about the physical light sources in the worldspace, and mesh-flagged material emittance properties. Nearly every 3D game engine has and uses one to facilitate shader calculations that rely on scene geometry but only need what the camera actually sees. It's where most of what you actually end up seeing on the screen begins being drawn. The GI shaders use that to inform the behavior of the rays, which they then use to distribute light and shadow with literally unprecedented accuracy, pixel-by-pixel and frame-by-frame.
I never thought something like this could be possible in ReShade. It's crazy to me that this runs smoothly. It definitely has some ghosting but frame rates are good! All said and done, I'm down ~5 frames.
What's most fascinating to me about it is how well it manages to hide how drastically it alters the images. I have the benefit of being able to toggle it instantly at any time, so I can see what it does and make a lot of comparisons compared to RTX GI. It changes so much, and yet you don't register what it's doing as part of an effect. It's almost disappointing when you stop and toggle. The changes just 'disappear' into the images the more you look at them. Then, you explore, get in a few firefights, and moment to moment everything just clicks and jives. I think this is the sort of thing that hits on lower level brain processing... like the part of your brain that makes depth and grouping 'real' for your vision in the real world. It's hard to get around that effect and see what it's doing that makes it hook into your perception in that way. You only get one or the other... big picture impact, or one detail that you can see as different, but can't see how it could make a big overall difference. Being able to really see what it's doing, and freely see how the parameters work/react to changes has told me a ton about what I've seen in RTX games. You could say it's been... illuminating.
Once you play with it on for a while, it's kind of hard to see the game without it. Everything just starts to look so much flatter.