Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)S
Posts
9
Comments
308
Joined
3 yr. ago

  • I found my answer then! I appreciate that 👍

  • I wonder if someone can make a Firefox extension that auto fills user profiles in various accounts with nonsense... fake address, fake bio, fake job, etc. Make it easy for users to poison data.

    And the extension could add nonsense to various posts, like here on Lemmy. Not enough to ruin the content, but enough to taint any LLM data scraping.

  • Have you taken very verbose logs to see if there’s differences that stand out?

    Nope, but that could be a next step. I was hoping for an easy fix (like I did something wrong in the settings).

    Realistically, if this can't be solved by using the built-in setting options, then it's not worth the trouble to "fix" anything. It still works, so I'm glad to have the option.

  • To match it completely I think you need to disregard the presets and set everything yourself, like CRF and everything.

    Is there a way to disable the preset completely? There's no "none" option, just "speed, balanced, and quality", but I set everything else up manually.

  • Basically saving space. The dashcam I use records in h264, and I've been saving space (50%) by converting them to h265 (through my Samsung phone). Handbrake is far more convenient for these batch conversions.

  • Same version of Handbrake (1.9.2), but I'm not even sure how to check for the ffmpeg version on my NAS, or what Handbrake is using. The system panel in the Handbrake info window doesn't list ffmpeg at all.

  • Thanks for the ideas.

    GPU does seem to be passed onto the docker container, not only because of the dramatic increase in speed, but also because the CPU isn't being maxed out like it was before adding the GPU hardware line.

    Volumes are bind mounts.

    I experimented with permissions and user groups, and noticed no difference to the behaviour in render output.

    For these videos, quality isn't super important, but file size is. I spent most of yesterday tweaking settings on my local Handbrake install to get the right balance. So, I assumed transferring the settings over would give the same results.

  • The NAS (Synology DS920+):

    • CPU: Intel Celeron J4125 4-core 2.0 GHz, burst up to 2.7 GHz
    • RAM: upgraded to 20GB
    • NVME cache: 256GB
    • HDD are all Seagate ironwolf 8GB x 3 and a 4TB.

    Laptop (Framework 13 11th Gen intel):

    • CPU: Intel Core i5-1135G7 (integrated Iris Xe Graphics 80EU)
    • RAM: 64GB
    • Storage: 2TB WD Black 770

    I would expect a difference in speed between the two, but does the hardware affect the output quality/size if the settings are identical?

  • Synology's Hyper backup can be encrypted, so yes for my use case.

  • From what I understand, they charge "by the hour", so if you use less than a month, you get charged for the number of hours instead.

    It's not an egress or added charge... from what I understand 😬

  • There's probably some SEO "hack" involved, too.

    Either way, as a user, it's infuriating.

  • So it acts like an adult child with deeply anti-humanist traits? What could go wrong?

  • So many:

    • Unsubscribe links that are magically broken.
    • Newsletter popups when you're trying to view content.
    • Websites that are paywalled coming up in search engines.
    • Broken shopping cart payment systems.
    • The fact that it's an assault on your senses to browse without an adblocker.
    • "Sign in to view content" for articles, posts, etc.
    • Banking sites that time out after a minute.
    • likewise, shopping cart sessions that expire before you're done shopping.
    • AI generated articles that don't even sound like they're referring to the right topic.
    • Recipe websites that have a 15 page story for each recipe without a way to skip to the actual recipe.
    • "Endless scrolling" that prevents you from accessing links at the bottom of the page.
    • Business websites that don't list an address.
    • "Password must be short enough to guess."
    • "How did we do? " emails.
    • Automatic newsletter sign-up.
    • Desktop websites that act as if you're on a phone (i.e swipe through grocery filters instead of scrolling through them).
    • Unreadable, archaic, and/or low-contrast fonts.
    • "This content is not available in your country."

    I could list dozens more. 😮‍💨

  • If you read how they programmed this robot, it seems that it can anticipate things like that. Also keep in mind that this is only designed to do one type of surgery.

    I'm cautiously optimist.

    I'd still expect human supervision, though.

  • Being trained on videos means it has no ability to adapt, improvise, or use knowledge during the surgery.

    Edit: However, in the context of this particular robot, it does seem that additional input was given and other training was added in order for it to expand beyond what it was taught through the videos. As the study noted, the surgeries were performed with 100% accuracy. So in this case, I personally don't have any problems.

  • Interesting, it's slightly slower for me through the web interface both with a direct connect to my network, or when proxied through the internet. Still, we're talking seconds here, and the results are so accurate!

    Immich has effectively replaced the (expensive) Windows software Excire Foto, which I was using for on-device contextual search because Synology Photos search just sucks. Excire isn't ideal to run from Linux because it has to be done through a VM, so I'm happy to self-host Immich and be able to use it even while out of the house.

  • OK, indexing finished some time yesterday and I ran a few searches like:

    "Child wearing glasses indoors"

    "Cars with no wheels"

    "Woman riding a bike"

    Results come up (immich on android) in three seconds.

    But the quality of the results do appear to be considerably better with ViT-B-16-SigLIP2__webli compared to the default model.

    I'm pretty happy. 👍

  • Nope. Amazon Prime, YouTube Premium, Netflix, etc. will all still track your behaviour, collect your data to be used against you, and still make you the product.

    Corporate greed and enshittification have ruined everything.