A short thread: 1/ Traditional video compression moves pixels more efficiently. The human brain doesn’t store pixels at all — it stores meaning. Faces. Motion. Objects. Relationships.

That's an incredible milestone, congrats to @alextoussss on the first air-to-air moth intercept. Precision autonomous flight without cameras is genuinely impressive work. If you ever add vision to the stack, we'd love to help. At Mforja we do video distillation shrinking footage dramatically while keeping the detail, cutting storage, bandwidth, and compute costs for teams running vision at scale. 86.93% smaller video with the same quality. Rooting for the mosquito-free backyard future 🦟

Nicolle Merrill's "Can AI Keep a Secret?" is a smart, no-nonsense take on AI data privacy. Credit to her + Office Hours for the original: youtu.be/JcMj8BEjC6A We distilled it: same video, same runtime, 61% smaller (4.86MB → 1.88MB). Not a codec — we remove waste your codec leaves behind. Compare both yourself 👇 More: youtube.com/@MForja | linkedin.com/company/mforja/ | x.com/mforjatech

Night patrol coalition footage. The noise is high, but the waste is higher. MForja is the tactical strip. 25.8 MB → 3.5 MB. 86.6% reduction. Compare the original on the right to the MForja version on the left. No difference.

Jordan Walker hit six home runs in a row. MForja is that kind of precision. @SportsCenter Your videos have waste and higher CDN costs are the result. MForja removes the noise. 29.4 MB → 16.8 MB (42.8% reduction). Compare the videos to see there is no difference.

Distilled this 2013 USAF RPA squadron video: 21.41 MB → 1.78 MB (91.68% smaller). PSNR 28.69 dB / SSIM 0.9417. MForja isn't a codec — we cut waste before encoding, then your codec does the rest. Compare both videos yourself 👇 Source: dvidshub.net/video/288204 (credit: SrA Anthony Gamez, USAFCENT) More: youtube.com/@MForja | x.com/mforjatech

F-35B weapons load + short takeoff off HMS Prince of Wales — sharp footage from DT-3 testing in the Western Atlantic. Credit: Michael Land / Pax River F-35 ITF, courtesy Royal Navy. We distilled it: 12.26MB → 2.95MB (75.94% smaller), PSNR 33.81dB, SSIM 0.954. Not a codec — we cut the waste before your codec touches it. Compare original vs. distilled yourself 👇 dvidshub.net/video/900470/f-… More: youtube.com/@MForja | x.com/mforjatech #F35B #RoyalNavy

That F-35A B-roll from Daniel Eby / 56th Fighter Wing is sharp footage. We distilled it: 46.9 MB → 4.8 MB (89.8% smaller), same libx265 codec — no swap, no fallback. We're not a codec. We remove the waste before it. Compare it yourself 👇 Source: dvidshub.net/video/977326/f-… Credit: Daniel Eby / 56th FW / DVIDS (Public Domain) More: youtube.com/@MForja | x.com/mforjatech

Inside a 62nd ERS ground control station in Afghanistan: RPA pilots and sensor operators working the screens that keep troops on the ground safe. Real crew, real coordination, around the clock. Credit: SrA Anthony Gamez / AFCENT, via DVIDS. Ran a 2-min segment through MForja — not a codec, doesn't replace one (libx265 in, libx265 out). Just strips redundant data post-encode. 17.8 MB → 1.7 MB. 90.3% smaller. Compare source vs. output yourself below More: youtube.com/@MForja | linkedin.com/company/mforja/ | Follow: x.com/mforjatech #RPA #ISR #DefenseTech #Aerospace

Coalition ISR training from African Lion 2025: U.S. Army 119th Wing + Ghana Armed Forces running UAS ops, live remote control, and a simulated intercept scenario. Solid multi-domain interoperability on display. Credit: Staff Sgt. Jon Cortez / SETAF-AF, via DVIDS. We distilled this ISR clip with MForja — not a codec, doesn't replace one (libx265 in, libx265 out). Just strips redundant data post-encode. 93.3 MB → 14.0 MB. 85% smaller. Compare source vs. output yourself: Source: [insert link] Distilled: [insert link] More: youtube.com/@MForja | linkedin.com/company/mforja/ | Follow: x.com/mforjatech

"ISR Capabilities": Marines demo real-time surveillance & recon with small unmanned aircraft systems. Great footage credit Cpl. Jonathan Hodges, II MEF Info Group (DVIDS, public domain). We distilled the file: 34.4MB → 2.9MB (-91.6%), same libx265 codec, no shortcuts. Compare source vs. output yourself: dvidshub.net/video/761109/is… More: youtube.com/@MForja

RQ-4 Global Hawk B-roll — launch, landing, taxi, maintenance. Solid footage of a high-altitude ISR platform in action. We ran 1 min through MForja "distillation": 9.0 MB → 3.8 MB (-57.4%). Not a codec — we cut the waste before your encoder even starts. Less storage. Less bandwidth. Less compute/energy. Your codec (AV1 here) runs leaner too. Compare it yourself — source below. 🎥 dvidshub.net/video/153124/rq… Credit: SrA Andrew Buchanan, 9th Recon Wing, Beale AFB (DVIDS, public domain) More distillation demos: YT: youtube.com/@MForja LI: linkedin.com/company/mforja

Hey fellow engineer, did you know that footage like THE GOLDEN GIRLS improv scenes can be highly compressible due to static backgrounds? MForja exploits this, reducing the original 26.6 MB file to 3.1 MB without quality loss (PSNR 38.86, SSIM 0.9696). Check it out and see how much you could save on CDN and storage costs. @todd_spence, thanks for sharing this golden moment from THE GOLDEN GIRLS.

Coffee break! 🚨 Did you know that even in video compression, there's room for a 'perfect shot'? Stuff Worth Seeing's coffee clip compressed by 67.9% with MForja - no quality loss. The real magic happens when you apply this to your entire library... or your daily uploads. Imagine sipping on those savings! 🌐 MForja.com

This Week in Startups just got a whole lot smaller. We're talking about the file size. MForja compressed this clip by 87.0% from 131.2 MB to 17.0 MB without sacrificing quality. Credit: @twistartups 🌐 MForja.com 📺 YouTube: youtube.com/@MForja 💼 LinkedIn: linkedin.com/company/mforja 🐦 X: x.com/MForjaTech (@mforjatech) #VideoDistillation #MForja

Every day your streaming platform delivers sports clips like this @Nike Erling Haaland ad without optimizing them, you're losing money. Real calculated dollars: $50 per million views at current rates. MForja recovers that difference - no quality loss, same codec. What's the real cost of not doing this? MForja.com, YouTube: youtube.com/@MForja, LinkedIn: linkedin.com/company/mforja

Most people don't know that traffic cam footage is one of the most compressible content types out there. The static backgrounds and predictable motion make it a perfect candidate for MForja's pre-codec video conditioning. We took the Tennessee bridge accident footage from Traffic Cam Watch and reduced its file size by 81.2%, from 2.4 MB to 449.4 KB, with no quality loss. That's not just a number; that's a significant reduction in CDN and storage costs for anyone hosting this kind of content. Learn more about how MForja can help: mforja.com Credit: Traffic Cam Watch youtube.com/watch?v=SgkPaf0J… #MForja #VideoCompression #TrafficCam

You just watched a dramatic car crash. Your CDN paid for every frame of that crash, including the parts you didn't even notice — like the static background. MForja noticed those patterns too and removed the redundant data, achieving a 79.6% reduction in file size without touching the codec or altering perceptual quality (PSNR 37.35, SSIM 0.9578). That's not just efficient; it's a game-changer for video distribution. #MForja #VideoCompression #Efficiency

Crash footage doesn't have to crash your CDN bill. Traffic Cam Watch's dramatic video was compressed by 72.4% with MForja - no quality loss, same codec. The static background makes all the difference. @TrafficCamWatch, we've got a cost-saving solution for you. Check it out: mforja.com

Snow driving tips just got a whole lot smaller. Traffic Savvy Club's tutorial compressed from 1.5 MB to 511.2 KB with MForja - that's a 65.1% reduction! What would you do with the extra bandwidth? #MForja #VideoCompression #SnowDrivingTips