Ces'17

Since we talked about it here before... Warmup for the new AMD processors: Intel just launched Kaby Lake, and there are already nice reviews of the i5 7600K and i7 7700K. Summary: just like the previous generation, but new (ok, they're slightly better, but the difference is very, very small).
The new AMD Ryzen should beat these 4-core processors at well parallelized tasks (the top performer on those charts is the 6-core i7 6800K).


Boy Intel needs some competition. The last four generations have had barely enough to call it new. Corporate complacency to the fullest extent. Intel makes a great product but this lack of performance change and the $1750 10-core chip has lost a lot of my respect for them. But I will still buy their chips so what can you do... My hope is that the Skylake 10 core chip will actually be worth it and kick some butt with a decent clock speed. Not in the low 3Ghz.
 
Boy Intel needs some competition. The last four generations have had barely enough to call it new. Corporate complacency to the fullest extent. Intel makes a great product but this lack of performance change and the $1750 10-core chip has lost a lot of my respect for them. But I will still buy their chips so what can you do... My hope is that the Skylake 10 core chip will actually be worth it and kick some butt with a decent clock speed. Not in the low 3Ghz.

My hopes are with AMD right now. Let's see...
 
It looks pretty neat, but: I don't want to build it (they should use it to make final products), it needs higher res (it takes a 3-meters-wide screen to get 1080p), and I haven't heard the price but everybody says it's too expensive.
My guess: until all that comes to life and they can use this kind of tech to make bigger TVs that are not very expensive, it's just not for me, but for advertisers and the like.
 
I figure that, at the moment and for a while going forward, it will be for events, public stagings/concerts, billboards, advertising displays, etc. As costs come down, they will be able to do all of the above plus serve as movie screens/displays, thereby eliminating the need for a projector room. For a 15-20 meter wide movie theater screen, even the current resolution will more than suffice. I believe the current laser powered high end/pro projection systems like Barco can run in excess of $250,000. Small LED pixel screens run about $2,000-$8,000 grand per tile.

The 6mm screens for the outdoors are already fairly inexpensive.

https://www.alibaba.com/product-detail/Leader-supply-full-color-small-screen_60523929789.html?s=p

The higher end stuff can add up to half a mil for a full setup.

 
My eyes would have to be 25cm (10") away from that screen to notice any difference between 4k and 8k
Obviously, at $5,000, it's for editing and the up-close image analysis. I'd think, as with that high ISO/Neat noise removal clip from Luke Neumann's GH5 video, 8K would help.

But, of course, it's mostly made for the next generation pro cameras (aside of the Red Weapon, which is out already) with the 8K sensor. On the lower end pro/high grade consumer level, I expect the GH6, A6700 types to no longer oversample 5K/6K into 4K inside the camera but offer options to do this in post.

And, getting back to computers. Intel is going to slowly release its Optane (Cross Point) memory with its latest Kaby Lake chips. Supposedly, Optane + HDD will be faster than DRAM + SSD but, for the fastest with the mostest Optane + SSD will be the one to get.

http://www.computerworld.com/articl...-can-expect-in-pcs-and-when-it-will-ship.html
 
The problem with 8K is that it's 16 times the data of 2K, which is all anyone can see most of the time.

Or to put it another way, its pixels are 1/16 what they could have been if the sensor were 2K, and therefore 4 stops less dynamic range, 4 steps less ISO. Or, it could have been global shutter and still have 1 or 2 stops more dynamic range and 1 or 2 steps more ISO, than an 8K rolling-shutter sensor of the same size and circuitry.

Or to put it another way, it is 16 times as expensive to record, and takes 16 times as long, or 16 times the computer, to put together.

All that for an ultimate viewing-pleasure increase between 0.00 and 1.00%.

For the super-jittery perfectionist, I can see the desire for a sensor whose RAW, undebayered resolution is 4K. And I can kind of see the desire for a computer monitor between 2K and 4K, so you can have a big screen with a lot of stuff on the desktop, multiple windows open, etc.

But there really isn't a need for a 4K living-room TV, much less 8K.

I am appalled that manufacturers are peddling TVs greater than 2K, but I am aghast that video professionals would give them a second thought.
 
Last edited:
Euro Tier One 4K camera is 4K RGB or 8.8x3. Only F65 qualified so far and even that is 17.6 MPX. So, for better color rendering at higher resolution, more pixels are needed. Pro production and daytime footage will take care of the light, with the dynamic range a milieu of multiple exposures. Not that a modern display can handle even the GH4 maximum DR.

8K imagery is destined for these huge video walls that I had linked to a few posts earlier. Documentary, sports, product presentations will all benefit from it. An average movie at an average home will probably benefit the least but, for a high end home theater, the limit is always at or near the sky.

As to global shutter, a $2,000 GH5 already displays very little rolling shutter at 60p. 96 will make it negligible for even the discerning pros and that will come along on the next $2,000 model.
 
Well, I can see a world in which I would want a 4k TV. But it needs to be HUGE, at around 150" (which is cinema-theater-like on my living room).
Also: make those tiles twice the horizontal and vertical resolution, and they can be used to create that screen.

edit: just added the 1DX2 rolling shutter numbers on this thread. It joins the BlackMagic cameras on the list of "4k with unproblematic rolling shutter" cameras.
 
Last edited:
As to global shutter, a $2,000 GH5 already displays very little rolling shutter at 60p.

This sample video of whip-panning the GH5 is far from good enough for my taste. If it's better at 60 fps, that's good to know, but doesn't help me much, because I would prefer most of the time to shoot at 24 fps.

And don't forget that lateral movement is usually the least offender. No matter how fast the read-out, a rolling shutter will still show artifacts with super-fast things like: the flash-bulbs of still-photographer's shooting beside you, dance-club strobe lights, and lightning. These have been distractingly bad in videos I've seen.

And for the perfectionists who pursue 8K on principle, regardless of whether anyone will ever notice, I suggest you aim your perfectionism on the thought that you will never correctly capture phenomena like fans, propellers, musical-instrument-string vibrations, baseball-bat swings, golf-club swings. Even the moving lips of someone talking is irreversibly distorted.
 
This sample video of whip-panning the GH5 is far from good enough for my taste. If it's better at 60 fps, that's good to know, but doesn't help me much, because I would prefer most of the time to shoot at 24 fps.

And don't forget that lateral movement is usually the least offender. No matter how fast the read-out, a rolling shutter will still show artifacts with super-fast things like: the flash-bulbs of still-photographer's shooting beside you, dance-club strobe lights, and lightning. These have been distractingly bad in videos I've seen.

And for the perfectionists who pursue 8K on principle, regardless of whether anyone will ever notice, I suggest you aim your perfectionism on the thought that you will never correctly capture phenomena like fans, propellers, musical-instrument-string vibrations, baseball-bat swings, golf-club swings. Even the moving lips of someone talking is irreversibly distorted.

Rolling shutter is still the biggest problem with most CMOS sensors. I just don't think most manufacturers GET how much of a problem it is. I brought it up to Panasonic at the launch of the V35 at NAB '14 and the reps basically dismissed me and played it down(not like there was anything they could do about it at that point anyway). Flash banding, skew and jello have no place in professionally produced images. Global shutter is the main reason I bought an F55 over the F5.

I'll gladly trade a stop of sensitivity for global shutter.
 
Global shutter is one reason why I still have some HMC 150s that we use when we don't need higher res cameras. I am not a fan of rolling shutter but have no choice at my financial level.
 
And don't forget that lateral movement is usually the least offender. No matter how fast the read-out, a rolling shutter will still show artifacts with super-fast things like: the flash-bulbs of still-photographer's shooting beside you, dance-club strobe lights, and lightning. These have been distractingly bad in videos I've seen...
There's a quite a bit of motion - hands, saws, etc. - in this video (for the record, my connection buffered too much for the 2160 @60p, so I had to lower it to 1440@ 60p). I didn't find it, in the least, distracting compared to the well-known jello effect in cameras like A6500. Mind you, GH5 also does - or will do upon shipping - 180p at Full HD.

 
The problem with 8K is that it's 16 times the data of 2K, which is all anyone can see most of the time.

Or to put it another way, its pixels are 1/16 what they could have been if the sensor were 2K, and therefore 4 stops less dynamic range, 4 steps less ISO. Or, it could have been global shutter and still have 1 or 2 stops more dynamic range and 1 or 2 steps more ISO, than an 8K rolling-shutter sensor of the same size and circuitry.

Or to put it another way, it is 16 times as expensive to record, and takes 16 times as long, or 16 times the computer, to put together.

Put another way, 8k is pointless. The industry is going to try to push 8k on consumers just like they did 3D. And they are going to get exactly the same results. No value = no sales. Learn the frellin' lesson!

More resolution is not the answer. More dynamic range is, better tonality, smoother gradients, better handling of highlights and in particular clipping (more ARRI-like), fewer and better controlled compression artifacts, higher bit rate recording. And wait for it... global shutter.

I get that it's easier to make 8k than solve the real problems. But do they get that I want to spend my money on real improvements and not useless extra resolution? They will, because I control my money, and I'm not buying 8k. And I'm definitely not alone.
 

This was nicely shot. I still dislike the 60p look but it is what it is. The resolution is amazing for such an inexpensive camera with such a small sensor and the low light was also much better than my GH4. The skin tones were better than I ever had on my GH4 but still a bit "off" to my eye. Probably not enough to matter to clients. Skin tones are leagues better than any Sony short of the F55.
 
Put another way, 8k is pointless. The industry is going to try to push 8k on consumers just like they did 3D. And they are going to get exactly the same results. No value = no sales. Learn the frellin' lesson!...
No sales? For what? 4K TV's? 4K cameras? Both are doing quite well.

This was nicely shot. I still dislike the 60p look but it is what it is...
60p looks fine for a documentary. I wish more sport casts were in 48p/60p. It smooths out the motion.
 
Back
Top