Love to be subjected to repeated Claim your Trump Account emails from the IRS because I had kids over a decade ago.
Search
210 results for “scott”
-
Dad being kept overnight in hospital. Always that constant panic at the back of my mind that this is the *last time*.
Will this. be how it all ends? -
The secondary impact of a massive buyout (that puts a lot of voluntarily displaced workers in job market) is to extend the job search for long-term unemployed and recent college graduates still looking for their first job.
None of these three groups of people are counted in official #unemployment figures.
2/x
-
1959 Imperial LeBaron Southampton head-on at the Knight Riders car show, Longmont. Four headlights in chrome pods, legal in the US only since 1957... Monday's 1948 Buick has one per side. Owner bought it used in 1967 and still has it.
-
Dad back in hospital, I'm feeling AWFUL so need to write. Think I'll work more on Moseltlyucid.Relevance, my general purpose RAG(ish) adaptive, predictive recommendation engine. Also apply for jobs that would make me write stupid crud apps.
If (as I feel right now) it's the last thing I ever release it'll be GREAT. No UX...not a 'general use' tool but a framework for a universal recommender engine. -
The currently-complete tests are all 1080P. I'm comparing various H.264, H.265, and AV1 compression options with FFMPEG 9.0.2. Generally, I'm searching for the CRF (~quality) value that gives me a VMAF score of 95, and then tracking how much CPU time (and bitrate) is needed, then iterating over a ton of flags and videos.
Basically, I'm looking for the best compression options for each video to get a consistently good-to-great quality encoding with minimal visual artifacts.
The big surprise so far is the compression difference between the same videos using a ProRes original vs a H.265 (Main10 4:2:2) original.
I have two test videos for this, from two different cameras (Panasonic S1 II and Blackmagic Pyxis 12k), both shot in raw. I then exported the exact same 30s clip as 4K ProRes and as 4K H.265 (Main10 4:2:2) files. The two files for each clip are visually ~identical, but there's a 20-40x difference in file size.
Then I did 3 different compression runs:
1. Source=ProRes, scored vs ProRes.
2. Source=H.265, scored vs H.265.
3. Source=H.265, scored vs ProRes.VMAF is designed to catch compression artifacts by comparing the output video the uncompressed original, which makes case #3 kind of interesting. I'm doing this because I'm currently exporting most of my "final" videos out of Resolve as H.265, because I don't want to lug around *enormous* ProRes files. But I don't know what this is actually costing me, and this gives me a concrete measurement.
So far, using 1080P (downscaled from 4K originals, which *should* minimize H.265 artifacts quite a bit), I end up needing around a *30%* higher bitrate for H.265 sources to match the ProRes originals.
To put this in concrete terms, for one sample video, my best AV1 output sizes for VMAF=95 were:
- ProRes/ProRes: 3.48M
- H.265/H.264: 3.30M
- H.265/ProRes: 4.68M (+34%)Encoding into H.265 and H.264 gave similar ratios. I kind of expected H.265 to gain some sort of advantage here, but it didn't.
As a follow-up, I'll probably be investigating additional output options for Resolve -- surely there's *something* usable between the huge ProRes files and the much smaller H.265s.
-
Did I say 5 different source videos? It's up to *19* now, with a mix of action cam footage (from several different cameras), mostly-static landscape video, long-lens wildlife video, and a couple low-quality home videos.
I also have a couple videos in both HDR and SDR, plus 2 exported from Resolve in ProRes vs H.265, to see what that does to the final compressed version.
There's no kill like overkill...
-
Smoosh
-
Reminder: we failed to 'Save the Cheeleader'
-
CW: Left Wordle Spoiler / Hint
Johnny B. Goode
#LeftWordle #WordPuzzle #Hint #Spoiler #LeftWordleSpoiler #LeftWordleHint
-
Looking For Recommendations
Hey Fediverse, what are some of your favorite #peertube #channels or #creators to follow? My Youtube Premium is ending soon so this seems like a good opportunity to spend more time on the platform.
I enjoy #cozy content, #fibercrafts , #travel, #microcontrollers, #software engineering, #diy , #transit, #bikes and probably too many other things to list.
-
Old football leagues:
Creetown F.C. played in the Southern Counties League
https://en.wikipedia.org/wiki/Creetown_F.C%2E -
-
-
-
-
-
An ad for Firefox?!
#marketing #KCvsSEA #SNF #NFL -
-
This is America. Don't pretend we're better than this.
-
My #phone won't focus in the moon.
#Baltimore #photos -
Crazy week of work completing the SHOW ME WHAT YOU GOT series of articles. Now have the post-frenzy depression for a few days.
I need to work to function; it's sad but without work I spin out. -
We all died during covid and this is hell.
-
Pink flowers at the Denver Botanic Gardens, unidentified. Five petals apiece, each notched into two lobes, four blooms off one stalk. f/8 to keep all four sharp across slightly different planes, ISO 6400 in deep shade.
-
'Suwalki Gap' - I predict that'll soon become a word we hear a LOT in the news as Russia pushes it's 'test if Trump will support NATO allies' plan.
-
It's interesting to see the intellectual controtions and downright criminal activity as the big US AI companies try to defend their turf against open models until they cash out in their IPOs.
Things are accelerating in ways nobody is prepared for. PHP2Shell was thankfully lesser than we all panicked about but expolits are about to experience exponential growth.
The proitection racket that onlly frontier AI models can defend against this 'open model' proliferatio IS the next product. AI driven security arms race...right at the IPO where they have to justify >$1tn valuations. -
Seeing a BUNCH of Qwen 3.8 obliterated models being advertised for use in exploit development. This is getting SUPER dangerous… I suspect you don't need an Astra / Mythos to find exploits. Just THOUSANDS of little agentic fleets probing constantly.
What I built Stylo.Bot to defend against really… These are my 'level 5' autonomous attackers I was worried about.The PHP2Shell exploit is the first… I suspect ANY website not under active patching will be vulnerable as these exploits increase in both complexity and impact in an exponential manner.
Won't be PHP2Shell… WILL be some popular open-source framework getting ripped open… the same frameworks currently drowning under PR hell.
Interesting times indeed! -
Of course they are - in the same way any audience segmentation works. People discovering MOST companies have detailed profiles of their users is crazy...
Basically *as soon as you start selling ads you need to sell the advertisers your user profiles - so they know who you have to advertise to*. If you're not paying for a product you ARE the product.. -
Part 4 in the increasingly ridiculous series… it’s kinda sorta about recommendation systems but really about me trying to explain fairly complex ML concepts in a practical structure.
Part 4 goes into how you plan 'interventions' in a recommendation system, how you characterise and evolve THOSE alongside customer/system characterisations by finding the value of those interventions. Covers concepts like context-bandits, causal inference and adaptive operational control in HOPEFULLY an easy way for a C# dev to follow.
SHOW ME WHAT YOU GOT (Part 4): Mind Control – The Subtle Art of Manipulation.
https://mostlylucid.net/blog/show-me-what-you-got-part4-mind-control
-
No I didn't write 6 articles last week. I wrote 24 over the past 18 months and I'm panicking that when I run out of money I'll never be able to publish them. It's a fire-sale.