Person you responded to said av1, I didn’t mean to imply you did. HEVC is good balance and quicksync will handle it as you’ve said. 10th gen stuff will handle 4-5 at the expense of more power (but less than like a typical gpu build).

Last statement you made is critical - usage dictates build

Just as long as you’re fine with your media server absolutely eating power all the time

Stop encoding in av1 and get a low power older intel chip around 10th gen or so with quick sync. Unless you have like 5+ users watching 4k media at the same time this will handle transcoding absolutely fine while using far less power than a dedicated gpu

Sure, that’s a better phrasing. I agree she was ultimately gaming to take over the party post Trump and it backfired on her miserably. Oh well, fuck her

Polling got to her I bet.

That’s what happens when you run as a maga ally and then suddenly grow a backbone. I’m sure she’s seeing internal polling that shows her base has abandoned her and based on her absolutely toxic rhetoric and past she’s not picking up much in terms of votes outside of the maga crowd.

So she saves face and resigns rather than facing the embarrassment of getting destroyed in a primary

Just fyi the Philips security thing does apply to the hub as well. If you care about local only access ensure your hub is on a vlan now with no internet access so it cannot capture the eventual update that will push a forced cloud connection. Philips has explicitly stated this is coming to all hue devices and that you must sign in with a “hue account”.

If you must buy them ideally buy them used so you do not directly financially support a company that is hostile to its customers

Hue is excellent from a build quality standpoint but if you value privacy or respect for consumers know that Philips is forcing consumers into a cloud model and changing the terms of service after the fact, for some people well over a decade after purchase.

In general your IOT shit should be isolated to a virtual lan that does not have internet access. As of right now if you buy hue hardware it will do this but eventually this will not work as it will require you to sign in to “Philips security” to activate and utilize lighting (you will need to maintain an active cloud connection.

The bulbs do have zigbee so you can always bypass their bridge of course but this is more complicated to setup and in my experience leads to performance that as not as solid as using the bridge. Also, the ethics of financially supporting a company that will change the TOS after sale to fuck you so they can worsen the product solely to harvest data for sale. Yuck

ugh and then self hosting people being like “just use tailscale if you want to expose your stuff to the web”. You know, the “cloud based” solution that requires you to use apple/google/microsoft for authentication. What a great idea for someone that is looking to get away from all this dumb bullshit

I agree with this. I am not self hosting for some end of days apocalypse (except I do have a whole house generator and am looking into off grid electricity, though I really missed the boat on that one haha).

It’s because literally all of these services are parasites run by scumbags. I don’t want anything to do with them. I use Jellyfin/komga because I don’t want my watch/read history sold.

I seek out local only iot devices and isolate them to a physical switch that does not have internet access, only a physical link to the pfsense box where there is an isolated vlan on their own subnet with a hard rule in the firewall to block any outbound traffic. I can still view things like cameras and home assistant remotely via headscale or even homebridge (with some added risk) but not nearly as much risk as the plethora of security issues created by lazy iot firmware developers who rarely, if ever, update their shit. But further than the massive security risks created by 500 iot devices are the massive amounts of profiling data.

I run my own apps like nextcloud and local llms because I don’t trust companies like openai and Google not to suddenly shutter products or drastically change the tos once I’ve integrated them into my workflow (plus tons of data harvesting).

I refuse to use apps and run local proxies to view paywalled news articles because I either don’t trust websites that offer this service or they capitulate to everyone that sends a cease and desist (see: 12ft.io).

Etc etc etc

It’s inescapable in the modern age and I’m so goddamn sick of it. I was talking to a family member the other day and they were signing their literal fucking infant up for daycare. The daycare is owned by private equity, because of fucking course it is. Group homes for the disabled, emergency rooms, drug and alcohol rehabs, etc have been bought up by those scum in increasing numbers over the past decade with the anticipated results (people dying, worsening outcomes, etc) so why wouldn’t they sink their disgusting claws into child and infant care?

The daycare (bright horizons, fuck them) will only send updates to parents via a proprietary app, of fucking course. Updates include pictures of their child, diaper changes, feedings, play, social time, developmental milestones, etc. both parents work so they’re sending the kid there 40 hours, it will be a lot of data - daily updates. The TOS is disgusting. Of course the company owns all the data, of course they can and will share it with third parties, of course they collect as much personal info as possible, and of course they do the bare minimum in terms of cybersecurity. Anything to drive up profits, even if it means exploiting the children you’re supposed to be serving and educating.

The worst part is that when I point this out, when I point out that COPPA exists and they can insist that an account not be made because their child is under 13 and demand an alternative, I look like the pain the ass. they know that ultimately the daycare can be like “fuck you, the app is policy, do it or no pictures”. everyone just wants it to be easy and wants pictures of their kid. And we’re so indoctrinated by tech at this point to just click “agree” and not think about it. It just fucking sucks. No one wants to be the martyr and people like me that demand to be respected are seen as assholes. Meanwhile the company that’s charging them $1600/month for childcare that’s spitting in their face to earn extra on the side with their literal infants harvested data gets a pass because “that’s what companies do I guess”.

They definitely just made an account on the app. In 10-20 years their child will probably have higher health insurance premiums because predictive modeling will have data starting from day 1 and it shows a higher risk of colorectal cancers or whatever.

I cannot stress how badly I hope for a drastic change in the next 5-10 years. This is unsustainable. The villains are winning.

Goes in waves and based on vendor. My build is based on 18tb drives and 6 months ago when a drive in my array failed my go to vendor (serverpartdeals) had no stock in that capacity. Like 0 units. Had to use my backup vendor (go hard drive) who in my experience is slightly more expensive.

Looking now spd has 18tb back in stock with several models to choose from. 10tb too but much less options (literally 2). Prices are wack nowadays - in 2023 I could get a refurb 18tb exos drive for $180 and now it’s $270-290. Fucking wild and I’m glad I built up my array when I did. It’s generally the cheapest decent option in that capacity; goharddrive has some cheaper options (Toshiba for $230, unaware of statistics for that drive, MDD for $260, rebranded exos drives so a slightly better deal but still way over 2023 prices).

That all said my array is 15 refurb drives and has been going for like 7 years now. Drive failure rate is low, 3 failures in that time. 2 drives was replaced by seagate with a brand new drive bc the spd drive still had mfr warranty left. The 3rd was an HGST that didn’t have mfr warranty but did have spd warranty and they replaced it immediately with an equivalent drive (they didn’t have more HGST at the time). That said I do worry seeing times where they are literally wiped out of stock; if I need a drive replacement I need it asap and if they have nothing on hand does that mean they hold none back for warranty replacements? Sure hope not.

Struggle to recommend them now though because the prices aren’t worthwhile. A new 18tb exos is 280-290 on amazon. Why the fuck would you buy a refurb when a new drive is only $20 more or the same price???

In 30-50 years we will maybe say this about stunted emotional development due to endocrine disruption from microplastics, or maybe social-behavioral disruption thanks to dopamine gaming from social media especially for youth, constant digital stimulation, pfas, ultra fine air particulate pollution, or whatever.

The point is the harm is happening still but we don’t care because much like when tobacco was becoming clear as a public health crisis. Or using your example like how it was fairly clear leaded gas had neurotoxic effects in like the 1920s almost immediately after they started using it because it kept destroying the brains of the gas factory workers, then they spent like 30 years studying it and basically finding atmospheric lead pollution increases as well as blood level concentration increases from engine exhaust.

But in both cases regulators drug their feet for decades while people literally died because god forbid they harm profits of companies to protect the public health and safety. We have some very clear ideas of what needs to be acted on now but we don’t because god forbid google be valued at 2.7 trillion dollars instead of 2.98 trillion

https://github.com/dayanch96/YTLite

Ytlite now, uyou stopped updating awhile ago. There’s also a fork of ytlite (I think ytlite+?) that has some plugins rolled in by default, though you can just roll them in yourself when you build the app. The plus version may be better if you get them prebuilt from an appstore (sketch, imo) instead of sideloading an app you build yourself

I used a pi for HA for years with no issue. Just use a good quality legitimate sd card and make regular backups of your setup

I use a cheap NUC now but that is about processing overhead mainly due to rtsp streams. If you have a setup like mine (10+ cameras, 4k streams) you’ll find it starts to get unusable and chokes after a certain number of streams (iirc like 5-6). This is exacerbated if you bridge the streams to homekit or stuff like that (though not as much if you do something like scrypted external to the pi obviously but then it’s like what do you run that on?)

Also https://youtu.be/aNghg1Y-WIc which isn’t perfect but for an American comedian is pretty fucking stellar

“Anything else anywhere else”

Clarify - do you mean remote access? Are you using your phone for local access (eg localhost:8096 or 192.168.0.1:8096 or some kind of mDNS like arch.local:8096)? None of that will work outside of your home without additional configuration nor should you just openly expose it to the greater internet without doing much more research. Short version though - easy mode: tailscale or wireguard, harder version headscale but can be fully self hosted.

If this isn’t a remote access issue and it’s just some devices aren’t working locally (eg your phone and browser can connect but tv can’t) I would bet it’s a networking issue with firewall/router. What do logs say? This is one that can be trickier because logs on jellyfin side won’t necessarily have any useful info (they might if client is trying and failing to connect, worth a look) and logs on the other side might be a nightmare or even impossible to access (eg a smart tv and you don’t have a developer account for apple/android/webos)

Until users turn them back on

AFAIK there’s no setting to outright disable subs. Maybe there is one somewhere, or you could strip them from media, but that’s terrible because subs in is always the correct choice. What can help is running the extract subtitles plugin though that won’t help for format.

Subtitles force transcoding a lot of the time depending on playback situation and media. A lot of compatibility issues there - downloaded movies often will have the PGS subs from a bluray, anime can often have ass/ssa, ibx subs, etc plus they all tend to be embedded in the media (and in anime’s case often with extra fonts and attachments).

all of these (basically anything but external srt) can interrupt direct playback depending on users Jellyfin client. Some have better sub support than others, android vs apple vs some esoteric client like webos all have their own niche weirdness. You need to search what the best option is for each users scenario. Like for my users that have apple devices checking this box in settings>playback allows direct streaming of a lot of content:

But for “the problem show” what do logs say? How is it encoded? Quick sync can’t transcode everything. The older your processor the less it can do and niche formats it definitely can’t do at all. Like unless you have a real new cpu (13th or 14th gen) it’s not doing av1 or vp9 content and that’s still getting offloaded to cpu for transcoding. And if it’s some ogg vorbis thing or whatever it won’t work. And do logs verify igpu is being utilized for transcoding/mapped correctly? It can be a bit of a pain in the ass (unless they improved the process, it’s been a while since I’ve had to do it)

What is the client situation?

The easiest fix is to find a client that can direct play all of your files and take transcoding out of the equation. Ugoos am6b+ as an example but if you don’t need Dolby vision there are cheaper options that are easier to configure with native jellyfin clients (instead of coreelec/kodi). Or if you need av1. But this needs to be done per user and costs money

Alternatively what is your hardware? Do you have intel quick sync video? If so do you have hardware transcoding setup? Like if you have Jellyfin setup in a docker are you passing through the igpu to the container? And if you’ve done that have you set up the hardware transcoding in Jellyfin? What gen cpu and what kind of files?

If you have the transcoding happening on the CPU and not the iGPU (assuming you don’t have a discrete gpu in a mini pc and frankly with quick sync you don’t need one unless your cpu is ancient, save the power usage) transcoding will crawl. But if you have quick sync video it should be fine with <4-5 users

More like we got tired of our kids swimming in per and polyfluoroalkyl substances, arsenic, mercury, cadmium, lead, chromium, benzene, toluene, xylene, ddt, glyphosate, organophosphates, atrazine, polychlorinated biphenyls, nitrates, phosphates, plastic garbage, etc

Almost like time and time again corporations have shown that if you do not explicitly force them to safely dispose of hazardous waste they will simply dump it in public waterways where it will eventually seep into the soil and even if it doesn’t it will persist for potentially thousands of years

This is not inexpensive ($500-1300+ easily depending on number of and quality of cameras) but it’s what I use and can recommend

Also no cloud with reolink is an asterisk situation. Out of the box with the reolink app for several of their cameras it will contact reolink servers for things like notifications. However you can config with app and then never use again. If you have some of the fancier cameras like the e2 pro or whatever the notification processing is thankfully not server sided but occurs on the camera so you can roll it into home assistant and do everything locally

To OP keep in mind this gets computationally expensive as well depending on your goals. If you just want 1-3 720p streams in the home assistant app then a pi is probably fine, especially if it’s like a pi 5. But if you want to encrypt those streams and forward them to homebridge so that less tech savvy users in your home can just view the streams on the home app in their iphone, you have like 10+ streams that are 4k, etc you’ll find the pi will choke and it may make sense to offload to a more powerful server or NVR (but if you want the homebridge thing even the nvr won’t save you)

Yes exactly, thus “liberal darling”