Devices > Video Cards, Monitors, Televisions, and Projectors
4K TV's
glynor:
--- Quote from: 6233638 on November 14, 2013, 03:05:51 pm ---Well that's not strictly true - Red shoots raw video, and I'm fairly sure DPX files are uncompressed.
--- End quote ---
No. I've used a TON of RED footage. Basically all of the B-Roll I've used for work over the past three years has been shot on a RED Scarlett in 4K mode.
All of RED's cameras shoot in REDCODE RAW (R3D) format, which is a wavelet compression scheme that does between 3:1 and 18:1 compression depending on the setting and the complexity of the scene. It is a very nice codec, but it is most-definitely lossy, and it isn't too difficult to film examples that expose the compression errors even at the highest-quality settings (as usual, sudden lighting changes, slow wide pans, and moire patterns expose the faults in the codec).
The ARRI Alexa can shoot in ARRIRAW format, which is truly uncompressed (it records the raw sensor data), but many productions actually use it in ProRes 2K 4:4:4 mode (because you have to record to an external device to use the RAW mode, and real-world production makes this difficult in many circumstances). Topping it off, the effective sensor resolution on the Alexa used in 16:9 mode is 2880×1620, so it is scaling in hardware when it "records" 4k.
--- Quote from: 6233638 on November 14, 2013, 03:05:51 pm ---That said, I know someone that does mastering for DVDs & Blu-rays, and honestly if it is properly encoded (40GB X.264 for the Blu-rays) you're approaching visually lossless.
--- End quote ---
I wasn't arguing that it matters, only that there is almost certainly no such thing as truly lossless for video. Even if you did it at the consumer-end, there has almost certainly been multiple generations of lossy encoding between you and the people and the lights.
Hell, you could even argue that film is "compressed" (mechanically) by the cinemascope optics they use to squeeze widescreen content onto 4:3 negatives.
Also, you know two people. ;)
--- Quote from: 6233638 on November 14, 2013, 03:05:51 pm ---Actually, it would be better to move to 4K than 1080p with 4:4:4 chroma. 4K 4:2:0 contains 1080p chroma.
--- End quote ---
It is nowhere near that simple. 4:2:2 at 4k would have:
Luma Channel: 3840x2160
Chroma A: 1920x1080
Chroma B: 1920x1080
For each frame.
4:2:0 does an interleave, so Frame A has:
Luma Channel: 3840x2160
Chroma A: 1920x1080
Chroma B: 0
and frame B has:
Luma Channel: 3840x2160
Chroma A: 0
Chroma B: 1920x1080
But, of course, you need BOTH Chroma A & B (and Luma) to make up a full YUV pixel, so the effective per-frame color resolution is 1/4 of the luma resolution at 4:2:0.
And, the chroma channels are subsampled of course, so it is effectively blurring 4 pixels together using an algorithm to make one composite pixel (because the sensor doesn't work that way, so it is done either in camera or in post). Worse, if the source and post workflow isn't all completely 4:2:2 (much more common than you might think), then the output often ends up being the equivalent of 4:1:1 or 4:1:0, even though the wrapper and codec thinks it is 4:2:2 (which then they compress to H264 or MPEG-2).
Again, for MOST things, it doesn't matter. For digital effects and greenscreen, though, it most certainly does matter.
--- Quote from: 6233638 on November 14, 2013, 03:05:51 pm ---As resolution goes up, the size of compression artifacts should be smaller as well.
--- End quote ---
That's handwaving. The math gets harder too, and the ASICs they put in consumer hardware are generally junk barely able to decode H264 at 1080p24 properly. I'll believe it when I see it in production. It is one thing to show a difference on a $500k theater projection system, and very much another on a consumer TV at home with a variety of (mostly crappy) DSPs in the way.
I'd much, much, much rather see them spend time actually optimizing things properly for 1080p before trying to go for 4K in the home. But, they want to try to sell HDTVs to people again, so they're going to keep trying things. 3D fizzled. Ultrawidescreen was DOA. So now they're pushing 4K. We'll see. I wouldn't hold my breath. The consumer replacement cycle is around 10 years, and I think we're about 2-3 years into the period when the wave of upgrades from SD crested.
So, whatever they're pushing in 7-8 years might have a chance. But it has to be "as better as" the SD > HD conversion is, not marginal.
6233638:
--- Quote from: glynor on November 14, 2013, 04:29:20 pm ---4:2:0 does an interleave
--- End quote ---
Are you sure about that? I thought 4:2:0 was quarter-res chroma, and 4:2:2 was half res chroma. (full vertical resolution, half horizontal resolution)
--- Quote from: glynor on November 14, 2013, 04:29:20 pm ---I'd much, much, much rather see them spend time actually optimizing things properly for 1080p before trying to go for 4K in the home.
--- End quote ---
What do you think needs changed from what we have today? A good Blu-ray encode seems about as good as 1080p is likely to get on the consumer side of things.
--- Quote from: glynor on November 14, 2013, 04:29:20 pm ---Ultrawidescreen was DOA.
--- End quote ---
I'm really hoping that this returns, because I desperately want an Ultrawide 4K OLED TV.
--- Quote from: glynor on November 14, 2013, 04:29:20 pm ---But it has to be "as better as" the SD > HD conversion is, not marginal.
--- End quote ---
I'm not sure that's actually possible now. I think most people are satisfied enough with how thin their TVs are, and things like contrast or color accuracy are of no interest to them. Maybe wide gamuts or high framerates will work, but a lot of people seem to be against high framerate content, and that's only been 48fps so far.
glynor:
--- Quote from: 6233638 on November 14, 2013, 04:44:54 pm ---Are you sure about that? I thought 4:2:0 was quarter-res chroma, and 4:2:2 was half res chroma. (full vertical resolution, half horizontal resolution)
--- End quote ---
I oversimplified a bit*, because it actually works at the block level and not at the size of full frames (the sampling pattern doesn't "know or care" about the full size of the image). The difference between 4:1:1 and 4:2:0 is the shape of the sample (in 4:2:0 it is a 2x2 grid and in 4:1:1 it is a 1-pixel-tall line). The net effect of 4:2:0 and 4:1:1 is that for every 4 pixels, only one is truly color accurate. The difference is in the dispersion of those one pixels and how separated they are from the averaged pixels.
4:2:0 is marginally better than 4:1:1 (in some circumstances), but 4:2:2 is much better than that. Here's a great resource that describes the different color sampling patterns in detail:
http://www.dvxuser.com/articles/colorspace/
PS. I don't have time now to comment on anything else, but I will try to later.
* Okay, I explained it very poorly. The frames aren't interleaved at all, because each sample generates both a Chroma A and Chroma B channel "pixel". But the sampling pattern does work out to effectively 1/4 res for 4:2:0, and if you think of it in "lines" that's how it works out (interleaved, but by line, not by frame). Read the article I linked above and look at the sample patterns. It isn't long, and it isn't quite as good as the examples I've seen in production classes (since those color patterns are completely non-real-world), but it is pretty good for a web resource.
glynor:
I forgot, this page from RED is also very good. They (also) don't go into the technical details of how the samples progress across the frame (which is sometimes important, especially for on-screen graphics) but it has nice animations that show how the different common patterns relate to each other.
Sparks67:
--- Quote from: 6233638 on November 14, 2013, 02:29:09 am ---Rear projection televisions have always been quite unreliable, and many people were sold them without even knowing that you have to replace the lamp in them every few years. Sony's SXRDs did have a massive recall though, due to defective optical blocks causing green blobs to show up over the image.
--- End quote ---
Mine works fine, but I am still on the first lamp. So it should die soon. Of course, I said that 2 years ago. Sony also had problems with LED TV's as well. Sony Changers, etc. The problem with Sony is more quality control.
--- Quote from: 6233638 on November 14, 2013, 02:29:09 am ---Well, a lot of people sit too far from their TVs, or choose displays which are too small. But the distance at which 4K provides a benefit is further than many people expect. On many AV sites there's a bogus chart I see posted all the time, which says you need to sit much closer than you actually do.
--- End quote ---
The chart is based off several years of studying the human eye. So, it is not bogus. Are you referring to this chart? http://carltonbale.com/does-4k-resolution-matter/ Well, if you plug in 70" then you get 4 feet, but I think his calculations is a bit wrong. Sharp recommends 6 feet on their new 70 inch 4k TV.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version