Musicvid,
It's all in the link I already provided, written by the National Asosiation of Broadcasters and offering a perfectly viable explanation and confirmation of what I wote.
It's google books, which I cannot seem to copy / paste, so here are screenshots for you. Hope they are readable. You can also open the link I provided above and read the text there.
I'll give you a quick teaser:
"When footage acquired using 4k sensors is oversampled or downscaled to HD, the results are several perceptual effects...
...the percieved sharpness of the image as originally converted through the optical path of lens to sensor and the camera's digital signal processing is substantially improved compared to HD acquisition."
There's a lot more similar, excellent and enlightening explanations and straight to the point statements, please read the original article, I won't copy it all by hand.
So, some research instead of a rather emotional response to what doesn't seem to match your experience or knowledge would have helped. And given the fact that I've communicated with you for quite some time here and that I really do respect and value your experience and the help you often provided me, I'm not saying this in a disrespectful manner. Just don't be too fast in judging the whole issue as mere "internet hype".
It's all in the link I already provided, written by the National Asosiation of Broadcasters and offering a perfectly viable explanation and confirmation of what I wote.
It's google books, which I cannot seem to copy / paste, so here are screenshots for you. Hope they are readable. You can also open the link I provided above and read the text there.
I'll give you a quick teaser:
"When footage acquired using 4k sensors is oversampled or downscaled to HD, the results are several perceptual effects...
...the percieved sharpness of the image as originally converted through the optical path of lens to sensor and the camera's digital signal processing is substantially improved compared to HD acquisition."
There's a lot more similar, excellent and enlightening explanations and straight to the point statements, please read the original article, I won't copy it all by hand.
So, some research instead of a rather emotional response to what doesn't seem to match your experience or knowledge would have helped. And given the fact that I've communicated with you for quite some time here and that I really do respect and value your experience and the help you often provided me, I'm not saying this in a disrespectful manner. Just don't be too fast in judging the whole issue as mere "internet hype".
@Mindmatter I am judging the situation based on my experience with my cameras. Like others, I doubt downscaling GoPro 4K or iPhone 4K source material is better than HD acquired material. But, I will give my new iPhone 8 a test, when I got the time for it and report back.
OldSmoke,
I agree, surely consumer grade sensors will not really make a difference in this case.
I hope I'll be able to do a thorough test on my A7S or my pal's FS7 next week.
But I know what I saw on those HD documentaries, there was a level of definition I have just never seen before, and it was sure as hell due to the 4k shooting.
More likely it was the glass. Seen it many times in the past.
Eagerly awaiting your controlled tests with steps to replicate.
Until then...
Musicvid did you read the article?
I read the article and find the comparisons to film and good microphones very odd. I know this is a legitimate source, but I question the conclusions and the comparisons.
No I didn't read the article because your link still doesn't work. Forgiven.
But you had to ask.
I was able to glean enough from your short quotes to see that the article is speculative, and floats on the assumption that there is some sort of "useful" correlation between analog reduction and digital decimation from downsampling.
The principle of analog reduction to improve density and for destructive noise reduction was known to me in middle school, when I spent a spring and summer reading Adams together with my father and best friend. I recommend to you reading the series, as well.
Back to your article, again you've read only what you want to see, when others here have already tried to reinforce critical reading and thinking skills in these discussions.
So I stopped reading when I came across this nugget of ambiguity:
It may seem implausible to scientifically measure such benefits when downscaling from a high-resolution format to a low-resolution one.
That is correct. Yet the author then tries to justify his tenuous position by referring again to an analog model, which just doesn't hold up in the real world of rgb production, which is where HD still lives. I learned the analog model in college and at Technicolor, and I learned the digital model at Pallas and Kodak during my career. If what the author says today is true, why wouldn't we all still be shooting film and recording to magnetic tape?
mindmatter, seriously, the only thing l would be interested in seeing are original source in two sizes from the same camera at the same shoot with no internal processing or enhancements. Then I will gladly test your product, downsampled by any means of your choosing, using tools you may not be familiar with, and share my impression based on the evidence.
You can bet I will be thorough.
And quick to acquiesce if proven otherwise.
I read the article and find the comparisons to film and good microphones very odd. I know this is a legitimate source, but I question the conclusions and the comparisons.
And by virtue of this discussion, @Former user just became my newest online bff.
@Musicvid Wow, you're really missing out with your cynicism causing you to just demand more proof, as if you haven't already been pointed at plenty. You could also do some searching yourself, you know. 😉
People have been doing this for years. I certainly have. Any camera that can produce 4k (even a GoPro) will produce a better result if you downsample it to 1080p than if you shoot in 1080p with the same camera.
So, argue about glass, or specific cameras all you like. Even say the GoPro is a piece of junk (it is), but if your camera shoots both (i.e a better camera than a GoPro) , all other things being the same: the 4k>1080p result will be better. You can get almost the equivalent of 4:2:2 from downsampled 4:2:0 because the information that was previously lacking can be inferred from the combination of the 4 pixels used to provide the final 1, the result being produced by combining 2x2 pixels into that lovely, single pixel.
You're also generally working from a higher overall bitrate per frame, as say a camera shoots 1080p @ 60Mbps and shoots 4K @ 100Mbps. It would be nice if it was 4x60=240Mbps, but we works with what we have. So, you have more data per frame, spread over a larger area. Yes, this can lead to some artifacting - depending on amount of information changed per frame - because you are stretching the data further per frame. But, luckily for everyone: both of these things are greatly reduced when resampled down to 1080p and the result is always better than if you shot it in 1080p in the first place, because at that resolution: the camera actually threw that extra data away.
It just works. 😉
Benefits of shooting in 4K ...
In no particular order ...
1) Downsampling from 4K to FHD produces better FHD than if shot only in FHD, with higher compression, lower data rates. This may not apply with high data rates.
2) Allows better pan/crop/zoom capability in 4K than in FHD.
3) Gives more latitude when stabilising footage.
4) Gives larger frame grabs, 4K vs 2K images.
I saw a clear benefit with the GH4, when I compared, some time ago.
I think the GH5 comparison would be of interest as the GH5 FHD is reputed to be better than the GH4's FHD, plus GH5 data rates are higher.
There’s plenty of info. on this out there, this is Wolfcrow's take on it ...
For me the important test is not necessarily the math, rather, does a camera/system that I use gain or not from the downsampling. I’d guess that it’s probably a law of diminishing returns as the quality of the camera/system's FHD (and bit rate) gets better.
Any camera that can produce 4k (even a GoPro) will produce a better result if you downsample it to 1080p than if you shoot in 1080p with the same camera.
I fully agree. I film now over two years with two different 4k cameras, a Panasonic FZ300 and since a half year with a Yi 4k+ actioncam. The 4k footage I downscale always to 1080p with Vegas. The downscale is so crispy sharp that I still didn't buy a 4k TV or feel the need to do so.
I tested both cameras in 1080p and 2160p rec mode. The 2160p downscale picture is always far better than the directly 1080p recording.
Three months ago I combined a 2160p Yi actioncam recording with a 1080p Sony PXW-X160 XAVC recording in a 1080p project. I can tell you that you clear can distinguish between the 2160p footage and the 1080p footage and which one looks best.
is downscaled HD from 4k sharper?
is downscaled HD from 4k sharper?
Nope, you're missing the point. The images in the thread which you created rather than continue the conversation here, aren't even close to what happens. And you must have tried pretty hard to create something like you think happens and avoid seeing the result that we all actually get. Then blow it up 5x larger than normal to emphasize the differences. 😉
Who said no Anti-Aliasing is applied in camera at 1080p? The result you get is going to look like the last, smoother, one in both "original" 1080p and downsampled "4k>1080p", because: same number of pixels at 1080p.
The difference is in the recovery of lost colour information, which using Black and White for your examples ignored.
Try this instead. Take a large photo (reality) and resize it to 1080p. Then take it again and resize it to 4k. Then resize that 4k one again to 1080p. Then overlay it on the first one, zoom in and toggle visibility off on that top layer. You should see a slight difference in colour detail between the two images. More shades of green in trees and bushes etc. That's the effect. It's subtle, but definitely visible.
It's a little like the image has been sharpened, but without the fringing effects that applying an actual Sharpen Effect would produce.
Brandigan, 3Point, JN,
thanks for some help from the finally arriving cavalery 🐎
😉
I feel it has more to do with many "HD" camera's resolution are NOT real actual HD. Sharper 4k downscaled WILL be in FULL HD, not the already lower resolution, "almost HD" captured on many "HD" cameras.
I feel it has more to do with many "HD" camera's resolution are NOT real actual HD. Sharper 4k downscaled WILL be in FULL HD, not the already lower resolution, "almost HD" captured on many "HD" cameras.
You feel? It would be a pretty strange camera manufacturer who would cripple their HD output but make sure their 4K output is nice and sharp.
I feel it has more to do with many "HD" camera's resolution are NOT real actual HD. Sharper 4k downscaled WILL be in FULL HD, not the already lower resolution, "almost HD" captured on many "HD" cameras.
This is generally accepted as part, maybe not all, of the reason why the downsampled 4k is better.
I feel it has more to do with many "HD" camera's resolution are NOT real actual HD. Sharper 4k downscaled WILL be in FULL HD, not the already lower resolution, "almost HD" captured on many "HD" cameras.
This is generally accepted as part, maybe not all, of the reason why the downsampled 4k is better.
"Generally accepted?"
By whom? People who "feel" things, instead of doing some simple investigation for themselves?
Cite your sources, please. 😉
I have read many reviews of hybrid cameras and non hybrid cameras, also dslr's that have both FF 4K, sometimes with a crop. There are many examples where the 4k is better than the FHD on these cameras. Its because the manufacturers use sometimes 2 different methods to produce either the 2K or the 4K. At times the 2K method is inferior.
Cameralabs is a good resource for reviews, they go into good detail as to the differences in video quality as well as photo quality. You could start there, theres EOSHD, not a review site as such but good info on this topic. Theres Tony Northrup, Imaging Resource, DpReview, and many many more expert than I.
So don’t just take my word for it, or wwjd's, do your own research.
I feel it has more to do with many "HD" camera's resolution are NOT real actual HD. Sharper 4k downscaled WILL be in FULL HD, not the already lower resolution, "almost HD" captured on many "HD" cameras.
You feel? It would be a pretty strange camera manufacturer who would cripple their HD output but make sure their 4K output is nice and sharp.
HD is a fuzzy frontier. Canon's prosumer/DSLR "1080 HD" was determined to be REALLY about 600 lines of resolution, so when you look at HD clips, they are fine, until you look at 4K at HD. Then, the "HD" camera fail is apparent. I went so far as to take a 1920x1080 jpg photo, and compared it to 1920x1080 "HD" video. Wasn't a matter of more colors, but obvious pixel resolution. LOOKED like about half. Was it REALLY "HD" video? I say no. They didn't cripple it, it was just counted different.... maybe each individual Red Green Blue pixel counted on sensor, instead of a singular combined "RGB" output pixel? ...marketing...
Here's someone else's HD pic vs vid comparison. Makes it obvious. Sure, the FILE it puts out is 1920x1080, but.... https://www.dpreview.com/forums/post/53037916?image=0
Oh, I know camera manufacturers cheat, even or especially, GoPro, which was the first camera I personally found this worked with.
At the faster fps (96 and 100) speeds they used to suddenly start halving the vertical resolution to 540, then resampling it back up to full HD, but it's pretty obvious when you have something with lines in it, such as a building. If you are snowboarding, then you might miss it. At even faster speeds they start halving the horizontal resolution too. That was with Hero 3 and 4. The 5 and 6 don't need to cheat, having better CPUs and sensors. I assume, haven't tried them, because the 4 does enough for what I use it for.
But, at bog-standard 1080p@30, or 25fps they really are using all 1080p lines (skipping every second one?) of their 4k sensor. Then at 4k they use the whole thing, as well as when you take photos. So between the two you can tell the difference is real.
It's also really simple to use an app to control the camera and quickly change settings and take near identical clips to process and compare.
But, even if they were cheating the whole time at every resolution, downsampling from high to low outside the camera will always give better results than when it's done inside the camera, so no need to get hung up on any particular camera.
and if you are shooting at 8k, downsampling to 4k....
(skip ahead to 3:55, in this video)
But, seriously folks,
To Mind mender, I film 4k and edit on a 4k screen. Not quite following what you are really needing here, but 4k is great for final touchups of 4k material. Does it help Vegas? I don't know, but I edit 4k in "PREVIEW Half" quality anyway, for cuts and stuff, then for grading and sharpening and effects punch it up to BEST QUALITY FULL And half. That may not answer your question, but that's what works best for me.
But, even if they were cheating the whole time at every resolution, downsampling from high to low outside the camera will always give better results than when it's done inside the camera, so no need to get hung up on any particular camera.
I disagree and rather say it entirely depends on the camera and way the image is processed in the camera. One would to compare a downsampled 4K to HD with a camera that shoots native 1080, like a HXR-NX5R which I use for most of my bigger events.
And as mentioned earlier, the HD1080 from my a6300 is bad and in that case the 4K downsampled does look better but I would generalize or draw a conclusion from it. Don’t just messed up the dow sampling in the camera. The a7s on the other hand not only down samples 4K from the full frame sensor it also crops in further by easing out a smaller area to achieve good HD quality. Something that obviously can’t be done with an APC sensor which already has a crop factor of 1.5.
So again, this my work well for a particular GoPro model but only because of the way the camera processes the resulting HD image.