Is 4K better than 1080p when playback is at 1080p

Which is the 4K downconverted to 1080p CLIP 1 or CLIP 2?


  • Total voters
    27
  • Poll closed .
This test tends to prove the inverse, 4K downscaled seems to better when zooming in.
To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.
 
This test tends to prove the inverse, 4K downscaled seems to better when zooming in.
To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Am I crazy? The 100% (normal) view is clearly sharper, cleaner with more detail for the native 1080, but as you zoom in, the 4K starts to look better. By the time you get to 500% zoom, the 4K is quite clearly superior.

Also not sure how applicable GoPro footage is to the DJI 4K camera.

Overall, the results of both studies suggest:
  • If you are going to only color correct in post, then film in 1080. The entire process will be faster and your out put will be at least as good, probably better.
  • If you are going to crop in post (e.g. stabilization, zoom), then film in 4K. You will need the extra size to keep a full size 1080 video.
 
If you're planning to crop/pan your videos during the editing process, and have a reasonably-powerful computer, you'll have more content to work with if you shoot in 4K, without losing resolution down to 1080. Otherwise, if the end result will be 1080 or 720 on YouTube or Vimeo, it's probably better to shoot in 1080 which will save space on the SD, and be much easier on your computer's processing power when rendering the final version. As far as which looks "best", if it's viewed on a standard monitor or mobile device, rather than a good HD screen, it will only occupy a small portion of your eyes' field of view anyway, probably in poor lighting conditions. So again, choose the options that are most efficient and least time-consuming for your entire production process!
 
The Results are in. The Actual 4K Video is CLIP 2. So it was evenly divided as to the guessing on Youtube. Maybe more people would have guessed if viewing it in it's native res. As promised here are a few clips with the proper labels.
  1. CLIP 1 is 4K
    13 vote(s)
    48.1%
  2. CLIP 2 is 4K
    12 vote(s)
    44.4%
  3. Can't Tell the Diff
    2 vote(s)
    7.4%
To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

And A 4K upload showing the real difference is in Size compared to a 1080p video.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.
 
Well that's totally embarrassing for me. I was fooled by the sharpness and details in trees... What software did you use to convert the footage? Adobe Media Converter?

I had my first flight recorded in 1080p but wasn't satisfied with the quality and after that I have always recorded in 4k and I'm happy with that. (regardless of the extra steps needed to convert the footage from 4k to 1080)

this weekend I need to replicate this test myself and if everything goes as with your results, then this will change my whole workflow of recording and editing....etchttp://www.phantompilots.com/members/beatnik.29256/
 
The difference between the GoPro results and the P3P results is interesting considering they are using exactly the same hardware (except for the lens). The GoPro results are what I would expect. The 4K version has more resolution and detail than the 1080p version albeit at the sacrifice of some latitude. In the P3P comparison, the 4K version and 1080p version are very similar.

This really highlights to me that DJI needs to do more work on their compression algorithms. They're using a sensor that is perfectly capable of more detail than is being recorded.
 
The difference between the GoPro results and the P3P results is interesting considering they are using exactly the same hardware (except for the lens). The GoPro results are what I would expect. The 4K version has more resolution and detail than the 1080p version albeit at the sacrifice of some latitude. In the P3P comparison, the 4K version and 1080p version are very similar.

This really highlights to me that DJI needs to do more work on their compression algorithms. They're using a sensor that is perfectly capable of more detail than is being recorded.

I don't know. I thought the GoPro results were a little surprising in that the picture is sharper in the native 1080P at 100%. The 4K downscale advantages don't show until you zoom in. I would have guess the downscaled image would the advantage across the board.

Your point on camera software is very accurate. One look at cell phone results shows that the software may be as important as the sensor. Many phones have the same sensor, but give very different IQ due to the software.
 
Yes, it is a little softer. And the exposure is not as good. It all depends on the downsampling method so admittedly not the best apples to apples comparison. Sometimes working at the highest resolution is not the best choice.

I actually don't like the GoPro 4K very much. I prefer the 2.7K which is right in the middle. It's the perfect mode for the sensor. I really really want DJI to implement a 2.7K mode at 65mbps.
 
I'm using Adobe Media Encoder to encode for , but Premiere Pro CC 2015 to downconvert. Perhaps a better method would be to take the camera files directly to Media Encoder to downconvert to Pro Res HQ. or 4444 in 1080p before editing. Thats a test worth doing. But all in all from the very start I saw that fuzziness in trees and grass. (The hardest for the Macroblock types of codecs to compress at low bit rate - at least for 4K frames.). I think the two look very similar at 1080, the 4K has a little bit more fuzziness. But for correcting the ever present horizon tilt, I guess 4K wins to zoom in and rotate in post.
 
One significant advantage to shooting video in 4K over 1080p is that 100% quality jpeg still frames taken from the video are 4 MB in size, instead of just 1 MB. That's 4x more information to work with in Lightroom! 4K video is more like a RAW file, while 1080p video is more like a baked jpeg. :cool: After creating a jpeg from the RAW file, they may look the same, but you have so much more dynamic range and information to work with in the 4K video. Jpeg only shooters may appreciate the simplicity of shooting in 1080p, but RAW shooters will want the flexibility and forgiveness of shooting in 4K, even if it's more work in post. Plus, it looks stunningly better!
 
Last edited:
One significant advantage to shooting video in 4K over 1080p is that 100% quality jpeg still frames taken from the video are 4 MB in size,... Plus, it looks stunningly better!

Yeah i did not think of that - another reason to shoot 4K if you need stills later. Rather than stopping record to shoot full size stills.
 
Yeah i did not think of that - another reason to shoot 4K if you need stills later. Rather than stopping record to shoot full size stills.
Plus your 4K video recording is continuous, rather than being fragmented by intervening still shots. You get 25 images per second to choose from! Simply awesome! Pick the sharpest frames, and then optimize them in Lightroom. You also have a continuous record for DJI, in case something goes horribly wrong, to help make your case for a warranty claim.
 
Just doing some initial video testing on the P3P. Here is a video that I shot yesterday that shows the same relative scene with 4K and 1080p30. Both using 100 ISO, 400 Shutter, Normal color, and custom sharpening set to -3.

See if you can tell which clip is the 4K and which is 1080p CLIP 1 or CLIP 2? Which looks better playing back at 1080p. Make sure you set youtube to playback fullscreen and at 1080. Otherwise the test is rather pointless LOL.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.
I was just saying in another thread about why it is always best to start with the best image (highest resolution) so that you can do whatever you want with it.

Yes, if done correctly, 1080P down-razzed from UHD or 4K will look better than something shot at 1080P. Depending on the camera, how much and what is done in post matters but there is a theorem that I can't remember and just spent about 15 minutes searching for that basically states that the best you can start with leads to the best finished product. It's a general algebraic theorem that has real life application to almost everything. I'll find out by tomorrow and post it.

Either way, yes, definitely, if you have the space and the computer power, you should always start with the best.

Example: When scanning a film (when they used to shoot everything on film), in order to send it to digital projectors when the change was happening from analogue to digital, there was a process that was very important called "Digital Intermediaries" and what happened in that process is there was a master print, scanned by the master film (the film out of the camera) that was used to make all the other slave prints that are/were sent around the world to theaters.

That scanning process started at around 2k and literally seconds each frame and by the time I was done with DI, there were scanning up to 10K at like less than a second per frame. Very few scanned at 10K but the Northern Light company that made the scanners touted that they could scan at 10K and a few places bought them (they were literally millions of dollars) like Lip Sync, The Mill, and those places in London and I think the first to have DI capabilities in LA were the now defunct Rhythm and Hues and Digital Domain. My point is that there was and never has been a 10K projector but they wanted their master slave to be a high resolution scan so that the prints would be better looking for that reason. That alone should answer your question. Why would the scan at even 6K when they were sending the prints out at usually 2K?

The answer is the reason I gave you above and it's true for everything.

Hope this helped.
 
Great comparison video!

I had similar results with my P3P (LOG color. 4096×2160, 3840×2160 vs 1920×1080 video input with 1920x1080 output from FCP 10.2.3 at 10 Mb/s H.264. I then compared TIF images extracted from the video):

At 100% there wasn't any real difference. Only at 400% zoom the 4096/3840 × 2160 were different from 1920x1080 input but even then I wasn't quite sure whether I prefer the more pixelated but sharper 1920×1080 vs more detailed but pixel-smoothed 4096/3840 × 2160.

I guess the downscaling algorithm plays a role on the results. So with my default FCP settings 4K to 1920x1080 produced smoother borders to objects like buildings. I usually have P3P Sharpness -2, Contrast -3, Saturation -2 -- maybe that is one cause for 1920x1080 being sharper while FCP downscaling smoothed the borders. In my test all those settings were set at 0.

I also compared some real-world video clips shot at 4096×2160 with 4096×2160 vs 1920×1080 Final Cut H.264 output. The 4096×2160 video is somewhat more crisp, but on the other hand, it somewhat stutters even on very recent PC HW and 6MP screen on VLC so I prefer 1920×1080 for presentation.

I very often fix a tilted horizon ~1° and shooting 4K definitely allows me to do that well. Extracting high-res still images is also a bonus because the max "real" still image resolution 4000x3000 is essentially the same as in 4K video.
 
Last edited:
Yes, it is a little softer. And the exposure is not as good. It all depends on the downsampling method so admittedly not the best apples to apples comparison. Sometimes working at the highest resolution is not the best choice.

I actually don't like the GoPro 4K very much. I prefer the 2.7K which is right in the middle. It's the perfect mode for the sensor. I really really want DJI to implement a 2.7K mode at 65mbps.

THIS!

I am reading this thread and wondering how there is even a discussion about it. First, all the videos should have been up-rezzed to ProRes 422 BEFORE being exported again as you lose a generation. Once you take h264 and recompress it, its garbage at any resolution.

So that, and then what are your methods for your downconversion? There are so many variables that to not even list them makes this comparison inconclusive about anything really. The workflow should be like I said and then if done right, you should DEFINITELY get better imagery from your 4k video than your 1080P video on a downconversion. This is objective. You simply have more space and color. All things being equal (meaning the same camera), if you have the space, ALWAYS shoot in the same aspect ratio and fps you plan to use but at the highest resolution you can. **** I wish I could remember that theorem. I will have it tomorrow.

Ian, as far as your example with the GoPro camera, I don't have that model and so I can't comment on it but if you are getting better results with a 2.7K because it has a faster bitrate, thats a different story. Bitrate is the more important than resolution.
 
I don't know. I thought the GoPro results were a little surprising in that the picture is sharper in the native 1080P at 100%. The 4K downscale advantages don't show until you zoom in. I would have guess the downscaled image would the advantage across the board.

Your point on camera software is very accurate. One look at cell phone results shows that the software may be as important as the sensor. Many phones have the same sensor, but give very different IQ due to the software.
Software is hugely important on phone cams. Look at some of the pro apps for the iPhone 6s. I am amazed at how little fanfare the 6s has gotten as a formidable camera player in the 4K market. I think it's better than any GoPro or anything on a prosumer drone that I've seen by a long shot and you need the good software to get the best out of it.
 

Recent Posts

Members online

No members online now.

Forum statistics

Threads
143,066
Messages
1,467,352
Members
104,933
Latest member
mactechnic