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

[beatnik]

I qued a clip in media encoder that was 8 minutes, 4 hrs to encode! h.264, you tube 1080
Really....seriously

Make sure you have Open GL selected in encoder?

 

[Charles Harris]

I did have the open gl selected. Footage did look crisper, I think I need to get 32 gig of ram. My computer did not like doing anything else while I was encoding.Took 3 1/2 hours for a 10 minute video clip.

 

[izzydrone]

Not sure if I should start a new thread but ...If I film in 4k and export to 1080 in final cut will it still be better then filming in 1080 and exporting to 1080... Because I can't seem to be allowed to upload in 4k anywhere directly from final cut so I'm deciding if it's worth filling up my hard drive with 4k over and over again.

 

[Ken Cruise]

Not sure if I should start a new thread but ...If I film in 4k and export to 1080 in final cut will it still be better then filming in 1080 and exporting to 1080... Because I can't seem to be allowed to upload in 4k anywhere directly from final cut so I'm deciding if it's worth filling up my hard drive with 4k over and over again.

The higher the quality of the source video the higher quality every resolution below it will be.

 

[beatnik]

The higher the quality of the source video the higher quality every resolution below it will be.

But does that hold up in the video blind test. Is 4K downconverted to 1080p really better than the 1080p that the P3A can shoot in the real world without 4K TV's? This is a limited test since 4K encoded with a Phantom P3 using a highly compressed Integer Transform codec such as H.264. It's not a RED Weapon or Epic with larger sensors where the larger sensor is being used full capacity in 4K compared to 1080 mode.

I kinda like the 1080p footage on Phantom 3 better than the 4K footage when presented like my example in 1080p.

Do you think CLIP 1 or 2 is 4K?

 

[Ken Cruise]

My guess is Clip 1 is 4K

 

[Griszka]

I didn't think of that. Hmmm. If you take the 4K first into color correction in a 4K sequence - then take that one out to your final 1080p delivery. I'll test that. Shoot 4K, grade in 4K then downconvert. .. (few minutes later)

OK Here is another example with CLIP 1 and CLIP 2. You guess which is which. 4K and 1080p shot. THIS time, both with color grading. The same effect on both clips. RGB Curves, Vignette and Grad in Sky. Effects applied in a 4K sequence on 4K clip before bringing it into the 1080 sequence for comparison again the other clip. Both rendered out at 1080p30.

You should downscale first. Render clip as 10 bit (ProRes or Gopro Cineform for example) and then color grade. I will do some tests today and publish here.

 

[SimonCB]

Thanks, I have not tried down scaling 4k to 1080. Does the 1080 look better? Will that also eliminate the choppiness of the video after being down scaled?

Thanks for your reply

Rchjr

Your choppiness is likely your shutter speed. In the video world there is something called the 180 degree rule where your shutter speed should be double your frame rate.

A frame rate of 30fps should have a shutter speed of 60.

24fps = 48 shutter (50 is closest)

60fps = 120 shutter.

Its to do with motion blur. Following the 180 degree rule will give you a pleasing natural movement. Because your shutter is 400 every frame is super sharp with no motion blur. Causing an unnatural look.

Sometimes a film will choose to do it intentionally like the beach landing scenes at the start of Saving Private Ryan to put viewers on edge.

 

[beatnik]

You should downscale first. Render clip as 10 bit (ProRes or Gopro Cineform for example) and then color grade. I will do some tests today and publish here.

I didn't do this in the video link. It's the Phantom clip directly placed in a 4K sequence with "Create Sequence From Clip". Then I added color grading. This sequence was nested in a 1080p sequence in PPro and exported. Vimeo would be better looking but I don't have a paid account.

I'll give it a day or so and then reveal the answer.

 

[SimonCB]

Definitely Clip 1. Clear in the tree detail.

 

[Hessam]

Definitely Clip 1 is 4k.
Received my P3 a while ago and have done amazing things with it.

BTW my first post on this site

 

[beatnik]

Tomorrow I will re-export the same sequence in 4K for Youtube 4K so you can see which clip is 4K and which is 1080p. Of course my monitor on my Mac and PC only goes to 1080p

 

[Griszka]

Promised test for you guys.

Enjoy.

 

[sergekouper]

I believe clip 1 is better.

 

[sergekouper]

Promised test for you guys.

Enjoy.

Maybe 4K a tiny bit better on large views but about equal to inferior to direct 1080 at 400% zoom in.

 

[beatnik]

Promised test for you guys.

Enjoy.

Thanks, now if you extrapolate this test to my footage you get the answer !

Thats a great test showing the integer transform compression of h.264 in the details. In the same given bitspace, 1080 just has less to compress. Nice view BTW. Of course youtube adds it's own to the mix.

I think 4K would win hands down if you show it on a 4K monitor compared to 1080p upscaled to 4K. Or if you scale 4K up 200% in a 1080p sequence compared to 1080p footage scaled up the same which would look awful. Otherwise, the benefits are somewhat more unclear. I think the 1080 stuff is nice in your example.

 

[Havavision]

Definitely Clip 1 is 4k.
Received my P3 a while ago and have done amazing things with it.

BTW my first post on this site

Welcome

 

[jimwinstead1]

I personally chose a P3A because I thought that for the expected life of the bird that I would not really need 4k, nor do I currently have a 4k monitor. I thought about resale and the higher residual value of a P3P, but I just know that I never really sell anything. The P3A seemed fine to me.

Now that I have a P3A, I sometimes wish I had gone with a P3P for just one reason. The pesky tilted horizon similar to the tilted horizon in your videos. I did not know at the time of purchase that a tilted horizon was such an issue. An issue that is now well known but still may never be fixed.

With a 4k source, I could have corrected the tilted horizon in post and still had a stunningly crisp video remaining. With a 1080 source, any cropping to edit the tilted horizon results in a less than 1080 video with visible (to me) degradation.

If I knew then what I know now, I would have gone with a P3P.

I chose the P3P because I got tired of the retailers telling me that the P3A's weren't in, but had plenty of the P3P's. Looking back, I'm glad I did!

 

[ianwood]

My guess is clip 2 is 4K. One would assume the 4K clip would yield better results after downsampling in post but in the case of the P3, there's just too much compression to get the benefit. The A9 hardware simply doesn't have enough juice to process and capture 4K and deliver a 720p live output at the same time.

DJI needs to add a 2.7K mode similar to the H4B. It is the perfect compromise of higher resolution at a suitable bit rate. They also need to tweak their compression. In both clips, the grass and trees have excessive loss of detail in many shots.

Disregarding the optics and bigger imager and thinking just about compression, compare the grass and trees in those clips against the RX100 IV at Youtube 1080p.

Fingers crossed for a new SoC wrapped around the A15 Cortex that supports H.265 encoding with an equally improved imager and some nice optics.

 

[Griszka]

Thanks, now if you extrapolate this test to my footage you get the answer !

Thats a great test showing the integer transform compression of h.264 in the details. In the same given bitspace, 1080 just has less to compress. Nice view BTW. Of course youtube adds it's own to the mix.

I think 4K would win hands down if you show it on a 4K monitor compared to 1080p upscaled to 4K. Or if you scale 4K up 200% in a 1080p sequence compared to 1080p footage scaled up the same which would look awful. Otherwise, the benefits are somewhat more unclear. I think the 1080 stuff is nice in your example.

I was surprised by the result of this test. 4K downscaled to 1080p is definitely noisier and faster generates artifacts during grading. It seems, like you said, that it is actually, a matter of H264 codec . I'll do some other tests (grading 4K timeline and then exporting to 1080P), but for now, it looks that for color grading 1080P is better choice.