CES 2019: ‘8K Is Getting Real’

puredrifting said:
If your definition of "insanely good" includes quality commensurate with a P&S or your mobile phone, sure.
M43 imager at 8k with the need for incredibly small pixels to image at 8k = tons of noise at medium to high ISOs.
Click to expand...

Not with this particular imager from Sony.

It's literally GH5s/E2 sensor with its Quad Bayer unlocked to output 8K. Because the pixel size is 4 times smaller, Sony's dual Correlated Double Sampling actually achieves much better SNR due to reduced Full Well Capacity.

Conclusion in English: This 8K sensor delivers similar noise performance as GH5s and Zcam E2 (the Zlog2 mode excluded).
 
"a little miffed at my audio guy, because it was our second channel that they other crew was on and I don’t know why he wasn’t using the #1 channel transmitter to begin with."
Did he explain?
How many are allotted to a given crew?
 
Gweilo66 said:
"a little miffed at my audio guy, because it was our second channel that they other crew was on and I don’t know why he wasn’t using the #1 channel transmitter to begin with."
Did he explain?
How many are allotted to a given crew?
Click to expand...

Never really asked him about. I usually request four total channels. Two coming into the bag and then two for the hop going from the bag to my camera. It's incredibly rare if we use two channels of wireless coming into the bag, but it's a 2 channel receiver(Lectro SR) and I get both cleared just in-case we would need two wireless mics going into the bag at the same time. I guess he just grabbed that particular transmitter that was dialed in to the second channel freq. Honestly, it should not have mattered, because we were on our assigned frequencies and he was routing it to me correctly. It ended up being just one of those bad luck/chance things. In the end, it worked out, because we had (a little) time to "fix" it and the co-ordinator actually ended up moving me to a higher and cleaner part of the band for the rest of the season. And it's usually no the other network type crews you have to worry about. We all play by the rules, because we know how important it is. It's the "locals". As in the local TV station photogs. I've gotten into it with them before. They always give the excuse that it's shared gear and they just picked it up at the station before the game. Funny enough, I didn't have any issues with any locals this season. Maybe because the frequency co-ordinator came in before one of the press conferences during a preseason game and laid it all out in black and white. One local guy made some excuse and the co-ordinator looked at him and said, "I know your chief engineer", meaning don't give me that BS, you better register and stick to it.

It depends. I'd say like a local photog would probably register one, maybe two. I register four. Two coming into the bag and two for a hop to the camera. The NFL Films guys, a lot, plus they ask for a lot of spacing/bandwidth between channels and what's around them. Which eats up channels and they're not even being used. Plus you have the broadcast and things like stadium personnel But it's a first come, first served basis and depends on what block(s) you operate in and if there are available frequencies when you register. Many years ago, we were standing out on the field early and the co-ord came buy and asked if anyone was planning on using wireless, We all said yeah, so he asked what are the freq's. I had a 'rain man' moment and immediately respond with mine. I think it may have been a fixed freq. system and I just happened to know it. He looked down his list, made a note and said I was all good. The guy next to me just had this dejected look on his face and half jokingly gave me the eye, because it was the same thing he had and he couldn't use it, then. Here at my "home stadium", it's not too bad for me, because I register before the season starts and have them for the entire season. Mercedes Benz in Atlanta is an absolute wireless nightmare. I've been denied there, because by the time I sent my request in, there was nothing available in my blocks. I have a friend who lives down there and is friends with the co-ord for the stadium and says it's almost impossible now, because the operations department takes up so much with their radio systems and so much of "our" wireless spectrum has been taken away. So, stadium operations and the broadcast take precedence and whatever is left, is all there is. Not fun...
 
Last edited:
combatentropy said:
Virtual reality buzz has faded at CES 2019 --- Associated Press
Click to expand...

The 2nd generation is a massive improvement over 1st gen. The quality you can get with a recent phone and a ~$30 headset is pretty decent. The desktop systems don't look much better, and having to be tethered is a drag. Perhaps 3rd time's a charm as Apple is working on (at least) a dual 8K system which is likely mixed reality (similar to Microsoft and Leap, only a much better and more consumer friendly design). Something you can wear even while walking outside, perhaps even driving (not clear if any legal issues: will be a similar experience to a fighter-pilot HUD). Current designs are still too big, heavy, clunky, not-sexy, and not enough resolution. I'm super impressed with the GPU+CPU on the iPhone-X re: graphics/rendering quality. Whenever Apple releases their device, expect it to be decent in all aspects: self-contained, wireless, good design, very high quality graphics, ~8K per eye, etc.

Searching the app store, I haven't found many VR games: it's kind of risky and a much smaller market vs. normal games (FPS, tycoon etc.) and casual games (candy/birds etc.). I'm finishing up a relatively simple VR game (with advanced physics + physically based materials) and will see how it does. It's a 2019 remake of the very first VR game ever (probably going to use original sound effects though iOS HRTF 3D is much much better than what I did back then (just Doppler, amplitude, and phase (inter-ear delay)). My teenage relatives like VR (and niece has a headset for her phone) whereas they didn't care for 3D (and never saw 1st gen VR). Agree that VR will trend down for a while but isn't going away fully like last time. When Apple (and similar) release their 3rd gen gear, it will pulse up again, and might finally go mainstream.
 
NorBro said:
When are you going to make a camera, jcs?

I know you've thought about it :engel017:
Click to expand...

Here's what I'd build (with today's tech):

1. Sony full frame BSI sensor with 5-axis IBIS
2. Color science created by machine learning to provide accurate, calibrated color better than ARRI. Exposure invariant color science and optimal color accuracy in mixed / sub-par light quality. What you see with your eyes the camera captures.
3. Phase detect AF with AI/cinematic/humanized behavior (pleasingly smooth).
4. Single-handed form factor- sexy design: slim, fits in hand like a glove, easy to shoot one-handed. Replaceable rubberized shell (magnesium body).
5. E-Mount (Sony lenses are better than Canon for video: AI processing can mimic desired lens coloring, distortion, artifacts, etc.). AI can also mimic Cooke and other lenses with desirable character.
6. 10-12-bit RGB 444 H.264 4K120 (max) + same for H.265 (optional). Who needs 8K when the TV has AI upscaling ;) Farther in the future AI image stream deconstruction (not compression) will capture resolution-less information that can be reconstructed at any desired resolution. This is basically the same process these 8K TVs are doing in real-time: deconstruct images into an AI-descriptive form (much like a video game: shapes + textures), then render frames in 8K (+ without aliasing!). This is vastly different than upscaling and sharpening (which can alias if too aggressive and doesn't provide any actual detail improvement, only perceptual (acutance)). Sony has this tech for their 8K TVs too.

Basically an A7S III with better than ARRI color science and lots of AI driven features. My year-old iPhone-X probably has the computational power to do this. This now oldish low power marvel could also work: https://www.movidius.com/myriadx (used in many drones). There's an AI revolution coming- many products will see massive improvements in usability and quality (e.g. see latest drone flight control systems). Once AI / machine learning is applied to camera color science, no more issues with wonky color, especially for skin tones.

For business and other reasons, a camera like this won't likely be on the market for a while :)

Farther in the future: new sensors will provide the basis for computational camera systems- focus & DOF can be performed computationally (no traditional single lens). Very high quality lenses will be very tiny- improving on nature's best optics: birds of prey. A bit further will be lens-less light field sensors. Perhaps even a spray-on-surface sensor that can capture a complete light field (yeah, nanotechnology will be used to 'wire' it together). Already progress in lens-less cameras: https://phys.org/news/2017-12-lensless-camera-d-images-scanning.html
 
marvinhello said:
Not with this particular imager from Sony.

It's literally GH5s/E2 sensor with its Quad Bayer unlocked to output 8K. Because the pixel size is 4 times smaller, Sony's dual Correlated Double Sampling actually achieves much better SNR due to reduced Full Well Capacity.

Conclusion in English: This 8K sensor delivers similar noise performance as GH5s and Zcam E2 (the Zlog2 mode excluded).
Click to expand...

If so, that would be pretty cool because the GH5S has excellent low light capability, I was duly impressed when I shot with it for a few weeks. I kept saying to myself, "This is an M43 imager?!?" The side by side tests that the Panasonic engineers shot between the GH5S and the A7 were impressive. That said, I'll never stop believing that 8K is merely a marketing exercise to sell more gear to a saturated market.
 
Last edited:
I know you're joking, but just in case you're asking about neural style transfer... link
It's not useful yet for cinematography, but give it a few years and it will tell you how to light and frame your specific shot in order to mimic somebody else's style.
 
indiawilds said:
How do I create art like Roger Deakins? Will be keen to know if AI can help.
Click to expand...

Lots of research and practice. AI will help show patterns of what shot types will work for what scripts and themes (matching appropriate emotion). Training data can come from scripts, categorized shot types, as well as visual input, user ratings, live-while-viewing fMRI, box office earnings. This could be used to suggest shots that work and perhaps provide insight into emotional cues correlating single and groups of shots to specific emotions related to script themes/patterns.

See also: https://willrobotstakemyjob.com

DP / Directing is pretty safe.
 
Samuel H said:
I have absolutely no use for 8k, but... this 8k Sharp camera looks insanely good
https://www.mirrorlessrumors.com/th...-announced-a-new-8k-micro-four-thirds-camera/
Click to expand...



46704494372_e555345bdb_h.jpg

46032478584_49274e3457_h.jpg
 
Last edited:
8K cameras, hmmm....
Here's what I will do, I will skip the 8K round and wait till the 16K cameras go past the maturity stage and then pick one up for sale...... :)
 
Cary Knoop said:
8K cameras, hmmm....
Here's what I will do, I will skip the 8K round and wait till the 16K cameras go past the maturity stage and then pick one up for sale...... :)
Click to expand...

In the not-so-distant future, AI will provide resolution-less cameras. Moving-image deconstruction at capture then render at any desired resolution in post.
 
jcs said:
In the not-so-distant future, AI will provide resolution-less cameras. Moving-image deconstruction at capture then render at any desired resolution in post.
Click to expand...
I'll skip that stage and wait for the hookup to the Matrix "creative cloud" for $60 per month. :)
 
Or we could just go out into the World and look at stuff with our own eyes... You know, before Skynet takes over and our AI equipped TV's enslave us and destroy the planet.
 
You must log in or register to reply here.
Back
Top