#i don't post pics really i don't have a tag | Explore Tumblr posts and blogs

unproduciblesmackdown · 8 months ago

Text

shifted gears like alright after the [lucifer the mysteries please] centered approach that found mostly Other Stuff, centering [blade gunnblade please], and already have found something new but in the "possibly inaccurate & probably irrelevant even if otherwise" vein as a review read through the wayback machine says: "perhaps none moreso than the gun toting android Blade Gunblade, perfectly embodied by Asia Kate Dillon. They even get an awesome heart-wrenching backstory." which is so far the singular mention that blade gunnblade is an android, however, reviewers can Just Say Things, so. i believe the blade gunnblade full backstory clip that's available is from pre 2015's run at the pit, & rather a previous run as serials at the flea (and the prop & setpiece differences if it was simply an alex seife understudy performance from 2015 would be: inexplicable) but i have to imagine that blade's peak tragic backstory ft. murdered wife & child would be very similar, & i guess it would be funny if they were just out & about living their ordinary life with a family like that while also an android, but. however i'll also give them that: mentioned a bit more often is that at least at some point kapow-i Becomes part android / referred to as a cyborg, thanks to the same character as in 2015 & prior's tragic backstory. i don't think this fun fact is like super constantly relevant or obvious, so i Could imagine something along the lines of blade getting the same treatment & it also not being particularly more relevant. or else the proximity of such concepts like "kapow-i's a cyborg, the serials run clip seems to refer to like 'we need this specific rando who is now blade to be injected with Ultimate Fighter elixir (& have their family killed. because) & people's whose bodies couldn't handle it are now kind of like funny little robots (cybernetically automated but with some personality)" & like blade we need you to be a fighting machine was thus sort of conflated lol. but maybe they're a cyborg like i think it happens a lot anyways

#blade gunnblade #difficulties: going okay not sure there's as comprehensive a cast list anywhere as the mysteries' program being available #which also wasn't fully comprehensive!#however there is plenty of overlap b/w the mysteries cast & the the pit kapow-i gogo marathon cast #thus some more recent blade gunnblade finds stemming from Looking For Lucifer #and also i have already done some [afaik cast member who Wasn't in both] rifling around for blade material so #but there's still some more digging to do. some [fb &/or ig accounts now deleted or privated] to press f for. got any pics #some ''damn someone who took some relevant production pics was ig tagged but their profile doesn't have their actual name #and also only has like 8 pics & they're all selfies like who are you''#both productions having More production pics out there i knowwww please....i want to See #also shoutout to another archived review's mention of a green strobelight & cowboy bebop's rain.mp3 used in a scene #''for devastating effect'' or impact re: the 2015 run like ya that was the blade gunnblade Devastation we have crucial 8 sec clip of....#filed away as a Maybe. but i don't imagine it would actually really affect things very much at all either way so #finally we understand mafee taylor's bestie like i always knew you were a killbot cyborg sicko....all for a coy lil blade gunnblade ref (:#pausing partway through this post like this can all be briefer....it Could be but i'm writing it. read my posts boy or don't #hmm for example this review also lists a director as ''joel stern'' which i'm 95% sure refers actually to joel soren so #but then also many other names are cited more accurately. yet still the one error there. that's where we're at #kapow-i gogo #asia kate dillon #perfectly embodied by! So true

2 notes · View notes

chipped-chimera · 8 months ago

Text

Okay. Gamedev nerd here gonna throw in some inputs. Also neuroscience nerd because it's actually relevant here, believe it or not. (This is long so buckle up)

No game needs to run faster than 30 fps

So this one is a funky one. Whether frames are 'detectable' or not to the human brain, comes down to exposure. Human brains are amazingly efficient at sifting out what I'll call 'junk inputs' - floaters in your eyes (that everyone develops eventually), the phone you chronically keep in your back left pocket to the point you're slapping your body going 'where's my phone?' only to find it exactly where you left it - the same pocket as always. Your brain got so used to those inputs, it's filtered them out because they're deemed 'not important' to be constantly shoved in your experience of reality.

Up until maybe the past 10ish years, most media that we've been exposed to has run at about 30fps. Our brain is used to this and makes it indecernable. As tech has improved, allowing for higher fps rates, you can now be more exposed to 60fps - in fact 60fps has been the broadcast standard for quite some time. If you get used to the higher frame rate, your brain grows to EXPECT this frame rate in order to efficiently throw out the junk information (the pauses between frames) so anything below that threshold becomes noticeable.

I run a pretty beefy PC with a high quality monitor, and it's gotten to the point where I can now distinguish whether something is running less than 60fps to the point I start fucking around in graphics options because I feel like my game is lagging slightly. Sometimes I will consciously run at 30fps if it's more stable for various reasons, but I do notice. 60fps comes across much smoother to me now, whereas 30 has this barely perceptible jitter - that's because my brain is no longer used to filtering out those junk signals.

So with regard to this - it is true that games COULD only run at 30fps, but it would have to be literally EVERY GAME to stay imperceptible to the user. It'd also have to be every video, every film, every piece of media we ever consume in order to REMAIN imperceptible.

Studies have been done on this and while 60fps has been accepted as the 'upper limit', it's uncertain if that will hold. Current data suggests 75fps might be the last noticeable threshold before diminishing returns. Ultimately this comes down to the user and what kind of monitor they have available, and what that refresh rate is capable of. After all, we don't see reality in FPS. End of the day, more frames = closer experience to a frameless reality = smoother experience. Now since I'm talking about the tech component, that brings me to -

No texture should be bigger than 2k

While for the most part this would have been a yes ... the fact is tech is changing. Or honestly has changed. Like yesterday. 1080p, once known as your 'HD' has not been 'High Definition' since maybe the mid 2010s. Your average phone screen - arguably the thing that most of us are exposed to on the daily - if it has been built later than 2014 is likely to be running a 2k to 4k definition screen. If you ever wondered why text looks so damn crisp on your teeny tiny mobile phone, but not on your monitor - this is why.

I upgraded to a 2k monitor last year after my ancient, literally pre-hd Cintiq drawing tablet of 15 kilos made the death knell of glitchy buttons. She had dead pixels. She was 12 years old. She was my beloved, glass topped art workhorse I had gotten as a high school graduation present that I'd slap the side of and say 'don't make em like her anymore'. But as good as she was, the difference between her and even my basic 1080p monitor was noticeable. After doing some reading, and a lot of that reading was saying '4k is now the standard' which sounded INSANE but now they're putting out 6k/8k resolution screens, I decided it'd be more cost effective to just get a screenless drawing tablet and a better monitor. I could always upgrade the screen, therefore keeping my tablet in action longer.

Hooking up that 2k monitor for the first time, last year, when this was considered literally the 'lower end' of the current market? It was like I'd put on my glasses. But I was already wearing them. Everything was so much clearer and crisp to work on. It was VERY noticeable.

But what happens when you have a 2k monitor? Well now 2k is the baseline. A 2k texture on a model, displayed on your screen where it is close enough to be at full resolution is now noticeable if it's NOT 2k. You know when a character is talking on your screen and you're getting distracted by the pixellated quality of their armour texture? That's because the texture is no longer matching your monitor resolution. It has more pixels to fill, and if the information isn't there ... well that's how you get pixellated looking textures.

So what is a Gamedev to do? 2k resolution textures are NOT SMALL. Ballpark average for a single image might be around 2MB for 2k resolution with standard channels. If you need MORE channels embedded in the image, like an alpha channel for example - well that's more information for the image to hold, therefore more file size. Crank that up to hundreds of images - diffuse textures, normal maps are a bare minimum, add on other stuff like alpha masking, tintmaps - if you want a custom character creation system that's actually decent? Multiple armor options? Recolouring options? Textures? That's a lot of images already + a lot of file size and you haven't even put in the actual game yet.

So you've now got to get creative with it. This actually brings me to what made me want to write this post:

There’s zero excuse for Baldur’s Gate 3 to be 150 GB when Skyrim is 6 GB

I'm a recent modder, but I've had the privilege of messing with Baldur's Gate 3 and I intimately know exactly how their textures are formatted and used at this point - and they are the epitome of 'get creative with it'. So I mean it when I say you're lucky BG3 is JUST 150gb.

Anyone following me knows I have gone on about it before, but as someone who was trying to rebuild models in blender post launch, you'd know that BG3 uses a very unique texture system for character models to basically stretch these textures to the absolute limit while maintaining a high fidelity resolution.

This horrendous flow-chart-from-hell is me attempting to reconstruct this system (and this is maybe only half of it). While it's likely since been cracked by someone else since I last looked at it, the best I could do was get close.

Here's how BG3 textures work: they are not your typical 'baked into one diffuse texture' image thing we're used to for most games up until this point, where you've basically flattened the skin out into a 2D picture with all the colour information.

What Larian has done is make everything grayscale, and most of them work like layering masks. For those familiar with digital art, this would be your transparent layers layered on top of each other, with or without effects. For an even more basic analogy - think of stencils, on top of each other, until you have made a composite image. This means, instead of making a texture for every single skin colour available in the game it's now only three pictures -

These three pictures contain stupid amounts of information however, because while we see a composite image, actually you're looking at 3-4 images PER IMAGE. Because in reality, this is what it looks like -

This is my reconstruction folder of separated images to reconstruct in blender. Each contains a layer of information - how much freckles to put in the face, where the ambient occlusion goes, the vein amount yadda yadda. Remember what I said about 2k images being big? Larian managed to reduce that to three out of a potential seventeen. And while an absolute pain in the ass to reverse engineer in Blender shaders, honestly I was kind of giddy at the genius of this.

It's actually what makes the entire game really flexible and modder friendly, because once you understand how it works the engine is SUPER versatile when it comes to being able to recolour things. You can share these textures across multiple models and make them look completely different by altering the value of each colour, per layer, per texture.

So if we were to extrapolate this onto the old 'flat diffuse texture model', I'll be generous here and say it's three images (eyebrow/makeup masks + flat diffuse + normal map a.k.a. the thing that adds microdetails like pores and shit) at a very modest estimate of 2MB x the 350 skintones available in the character creator?

That's 700 Megabytes - because to get the colour for each one, you now need a separate image for each one. Sure that sounds 'small' but that's just faces, and only ONE face! For the character creator. Then you have your multiple face options with different textures. And multiple races. Then there's hair. Tattoos. Armor. NPCs. World textures etc. It adds up and it adds up quick. Instead this means they can use just 3 images, for one face in the character creator for all 350 skin tones. That comes down to a measly 6 Megabytes.

Even then they cut corners where they can. There's a reason there's so many upscaling mods available for BG3 - some textures aren't 2k. So to say that they're somehow at fault for not making their game smaller while maintaining the current visual fidelity is honestly doing Larian a major disservice. These Devs scrimped and scraped for EVERY inch of room they could possibly get out of their textures, and I've seen it. I'll say it again for emphasis -

We are lucky it's ONLY 150gb.

lower poly counts > high poly counts

Okay so as a Gamedev-student-turned-modder, I agree. Most gamedevs will agree too - because they're likely already using the smallest polycounts they can within the restrictions of current generation consoles (which are always less powerful than PC) and while maintaining the baseline fidelity of likely at minimum 2k textures. I spent 3 months learning how to make 'Real Time Hair' for games a.k.a. hair that can be rendered real time as you play the game, as opposed to engine heavy simulated hair you see in CG film. If you don't want your game to turn into a frame-sink slideshow you have to keep the polygons as low as possible. And it is actually really, really hard to get nice looking hair with volume while sticking to this. If you wondered why so many hairs in game tend to be super straight and flat - this is why. It's easier to obscure a lack of volume when you're basically constructing a cardboard cube. Once you start distorting layers, like with curls for example, you're more at risk of exposing that this is just a really thin mesh.

This is the hair I built. It looks great - but it's about double to triple the standard polygon count for most of Cyberpunk 2077's in-game hairdos. I only went with it because it was a personal mod and I knew my PC could handle it. (And even then, it's still more modest than some mods out there - I've checked).

Super low poly can work too! But it often affects style in turn. There's no point putting a 2k texture on what we associate with 'low poly' models, but that's a completely different game at this point like Zelda: Wind Waker. The unfortunate reality is, if you're going for realism in your game, the faster the frame rates, the bigger the textures = the closer to literal reality it feels, and the more current your tech needs to be.

There's a bit of a psychological phenomenon around this. I vividly remember sitting cross legged on the carpet of the library as a mere 10 year old listening to my computer science teacher go off topic to gush about the 'realistic graphics' of a game that'd just come out with the boys in the class.

For context, I am 31 this year. That game? Halo. The first one. No not the remaster, the literal, first, potato-ass looking Halo.

Also remasters are a fiddly metric to measure by - often what can be done to improve the game is limited by the engine, which is also probably a dated engine. But also using a tiny resolution image to compare larger resolution improvements in textures is not going to make a discernible difference. (And also arguably the technical skill of the Game Developer and given I call Bethesda 'Bugthesda' my bar ain't very high for them but that's my personal opinion.)

End of the day, the fact is this:

Whether a 150GB game looks any different from a 20GB game comes down to a lot of factors, and the majority of them come down to what you're running. For some people yeah! Below 1080p at 30fps will run absolutely fine and look great! And there are plenty of low-poly games with smaller graphics that do really well - a good game is a good game!

But as tech evolves, things change - and so do industry standards. The current 'standard' definition television? 4k. So what might not be noticeable to you, might be very noticeable to someone else.

And Gamedevs are aware of that. They are also keenly aware of the fact people complain about filesize, and you bet whatever size you are presented with (unless you know that company to be notorously incompetent coughs loudly in Bethesda and arguably whatever the fuck the sims 4 team is doing atm) that is the smallest size physically possible that they can present you with on launch day after way too many crunch hours.

Be nice to your Gamedevs guys. The majority of developers are out there just wanting to give you a fun experience and tell a story. And I think that should be appreciated.

here are my biggest gaming hot takes:

- no game actually needs to run faster than 30 FPS

- no texture needs to be bigger than 2k, and most don’t need to be bigger than 1k (save for megatextures). A good chunk should actually be smaller than 1k

- lower poly counts > high poly counts

Once you pass a certain threshold, it doesn’t even affect the style of the game that much, and you’re just using exponentially more power to get exponentially smaller results

Like, for example, the original Skyrim was 4 GB. The remaster is 22 GB. That’s five-and-a-half times more space for the exact same content! The graphical improvements are honestly negligible, especially when you consider the massive leap in storage and RAM used.

These sorts of things just hit diminishing returns so quickly— My eyes can’t tell the difference between 1k and 4k textures.

We have so many fun ways to get the maximum mileage out of every pixel and every polygon. It’s sad to see those techniques slowly trickle away from big-budget games!

Even as storage space becomes less and less of a concern, there’s something satisfying about keeping everything as small and tidy as possible

#kerytalk #my commentary #gamedev #like look I hate bigass filesize as much as you do - especially on my piss-grade Australian internet speed #downloading a big game is a whole fucking day thing #or more likely - running my pc overnight just to download it #there is a reason I blew all my savings and installed a 8TB SSD in my PC literally yesterday #it was half off and I know where this is going #shit's only gonna get bigger as storage in general gets bigger ad infinitum - may I never have to delete and redownload a game again #it's tech and tech privilege and I know it sucks #but that doesn't mean we should start railing on gamedevs - they work really really hard #and often any kind of shittyness you're experiencing is related to some cut-corner demanded by SHAREHOLDERS #most people working in gamedev just want to be proud of their work you know?#also I feel most people say a lot of this shit and don't actually look into how the processes ACTUALLY work #like yeah a small resolution comparison pic of a HD update of a game is going to look like nothing special #that's because you literally crunched it into a small picture #how am I supposed to see the texture resolution difference if it's been CRUNCHED DOWN #also yeah some people are wizards at upscaling and downscaling. Not everyone is though #main takeaway = any actual bloat and poor coding in a game #likely some shareholder influence bullshit - not what the devs actually want or have done #so your buggy ass game launches #Sims 4 frankenstien's monster code of expansions etc.#throw the shareholders out - not the poor developer on a shoestring salary sleeping in the office to get your game out on time #this has been a tag rant #long post

8K notes · View notes