#this is purely from memory bc I can't find an example post

[firestorm-fox]

Black Death: I made the ultimate alien capable of countering the ultimate lifeform! Eclipse Fans: You messed up a perfectly good kid is what you did! Look at him, he's got anxiety!

[wolfboyvirus]

ok originally i was gonna reblog this with some tags and thats it but i kept going and i feared i would hit the tag limit, so, reblogging with my thoughts in the actual post this time! hopefully i don't ramble away too much.

basically i agree with most of this post but i just wanna add more details because im a nerd.

i think, the mix of data they use collectively as their memories, would consist namely of environmental scans instead of videos (similar to Connor's reconstruction and Markus's deconstruction), and digital audio samples (think like, waveforms). both of these can be turned into something for human processing as needed; the pure data from scans can be compiled into a visual, and the audio samples can be converted into an actual recording you can hear- which is actually how audio works normally irl! i only learned that bc of this post fhfjhkd

additionally, there would be a lot of other less important things that could be generated, saved, or linked back to different data. for example, the audio transcripts mentioned in the original post, as well as visual transcripts, both generated based on the initial scans and audio samples. lidar scans, also already mentioned in the original post. "links" redirecting back to personnel files that the android already has saved, instead of generating new ones for the same person every single second in each memory. catalogues of their system status, objectives, etc. and of course the date/time. prolly a ton of other shit im not thinking of too.

also, of course, i think androids would prioritize the actually important parts of their memories in this way, using some kind of algorithm to sort out what's "significant" and what isn't, and the rest would be compressed as small as possible, possibly sent into cloud storage instead of local storage. for example, with Markus. the memory of the junkyard would be very readily available, because that was a life changing and traumatic moment for him, whereas he would need to go through a bit more retrieving and processing to remember a random day years ago with carl, where neither of them really did anything of note. easy access when needed, but it doesn't take up the android's space when they don't need it, and in the full memory's place you'd likely have a much smaller bit of data that would essentially be a summary. so the android isn't left with a complete blank space. though, that does bring up an issue of deviants being tracked through that information transfer into, presumably, cyberlife servers. nnnnot actually sure what to do about that part, really? if someone does lmk

a particular concept im not really seeing is the idea of an android like... capturing moments fully in a picture or video clip, and then basically remaking the full memory with information from that, like ai generated videos in real life. because that would make it just as fallible as human memory, since it has the opportunity to remake the memory in a way that isn't actually completely accurate, no matter how insignificant the minuscule details really are. which, can be okay, its technically completely functional this way. but i have a feeling elijah cyberlife would have higher standards than that, yknow? the way i see it, its a lot more like.. a mix of motion tracking animation (ironic), lidar, and... some form of color mapping. a lot of laser scanning involved basically. im actually not sure if that would be better or worse as far as storage goes though; I can't seem to find anything about it online, as most people irl are focused less on the size of the files and more that scans can offer accurate 3d renders while videos can't. scanners seem heavily used by law enforcement in particular to scan crime scenes. hell, i think hank was seen using a handheld one during the carlos ortiz chapter, he was using it to scan the stab wounds. im just gonna assume its worth it??

went slightly off topic sorry, anyways

one thing im still kinda trying to figure out is the glitchiness of the memories we see in-game. if retrieving memories is basically just remaking a 3d animation using various saved components, then, the glitchiness just doesnt make much sense?? it could be the compression of a specific component involved, likely the colors, because iirc all the android memories are extremely desaturated as well.

i think thats everything i had to say actually, i didn't have many new ideas to contribute now that im reading back but. shrug. idk this is fun to think about!! cool post op :]

I don't think androids store memories as videos or that they can even be extracted as ones. Almost, but not exactly.

Firstly, because their memories include other data such as their tactile information, their emotional state, probably 3d markers of their surrounding...a lot of different information. So, their memories are not in a video-format, but some kind of a mix of many things, that may not be as easily separated from each other. I don't think a software necessary to read those types of files are publicly available.

Even if they have some absolute massive storage, filming good-quality videos and storing them is just not an optimal way to use their resources. It's extremely wasteful. I think, instead, their memories consist of snapshots that are taken every once in a while (depending on how much is going on), that consist of compressed version of all their relevant inputs like mentioned above. Like, a snapshot of a LiDAR in a specific moment + heavily compressed photo with additional data about some details that'll later help to upscale it and interpolate from one snapshot into the next one, some audio samples of the voices and transcript of the conversation so that it'd take less storage to save. My main point is, their memories are probably stored in a format that not only doesn't actually contain original video material, but is a product of some extreme compression, and in this case reviewing memories is not like watching HD video footage, but rather an ai restoration of those snapshots. Perhaps it may be eventually converted into some sort of a video readable to human eye, but it would be more of an ai-generated video from specific snapshots with standardised prompts with some parts of the image/audio missing than a perfectly exact video recording.

When Connor extracts video we see that they are a bit glitchy. It may be attributed to some details getting lost during transmission from one android to another, but then we've also got flashbacks with android's own memories, that are just as "glitchy". Which kinda backs up a theory of it being a restoration of some sort of a compressed version rather than original video recording.

Then we've also got that scene where Josh records Markus where it is shown that when he starts to film, his eyes indicate the change that he is not just watching but recording now. Which means that is an option, but not the default. I find it a really nice detail. Like, androids can record videos, but then the people around them can see exactly when they do that, and "be at ease" when they don't. It may be purely a design choice, like that of the loading bar to signalise that something is in progress and not just frozen, or mandatory shutter sound effect on smartphones cameras in Japan.

So, yeah. Androids purpose is to correctly interpret their inputs and store relevant information about it in their long term memory, and not necessarily to record every present moment in a video-archive that will likely never be seen by a human and reviewed as a pure video footage again. If it happened to be needed to be seen — it'll be restored as a "video" file, but this video won't be an actual video recording unless android was specifically set to record mode.