Recording in 5.4k VS. 4k - Did I screw up?

[Zeke]

Photo editing is pretty much Greek to me, I just can't seem to find clear answers on what each of the various settings do. I see some amazing videos and photos out there and I'd be happy if I could produce something half as good.

Anyway, one of the multitudes of YouTubes that I've watched advised shooting in 5.4k for better resolution. My first red flag was that for some reason in 5.4k you lose both the zoom capability and active track functions (Air 2s).

So I drove to a fairly remote area and shot 4 batteries worth of footage... and it all looks like crap. Everything appears relatively OK when the drone isn't moving, but any motion just blurs everything out. "Unsharp" is probably the best way to describe it.

Anybody have any ideas about salvaging the footage, or should I just chalk this up to lessons learned? Obviously I won't be shooting in 5.4k anymore.

Thanks in advance.

 

[jphoto]

Yeah, well, 5.4K isn't really "better resolution". Here's why: To benefit from any video format your display device (TV or monitor) has to at least match if not exceed that resolution. Anybody got a 5.4K TV? Me neither. So what would you do with the extra pixels? Scaling, which means converting from one resolution to another, is really bad when you scale down between two resolutions that aren't evenly mathematically related. So, taking a cue from what happens in the pro world, the extra image area simply get cropped off. Pro Digital Cinemal 4K is actually 4096 x 2160, but consumer 4K (more correctly, "Ultra HD") is 3840 x 2160. A theatrical projector is also fully Digital Cinema, so it shows 4K just fine, but consumer 4K displays are smaller, and that's a really difficult thing to scale well, so they simply crop off the extra. (Who thought consumer 4K shouldn't be the same as Digital Cinema 4K? Therein lies a tale!) Turns out, it's not much of a change. But the lesson learned is: shoot for your display.

Frankly, when I was first learning to fly a drone, I shot exclusively 1080p because I was a long ways from getting anything great, 1080p it saved a ton of memory space, and was far easier to deal with later. Now I shoot 4K (in the DJI Air 2S world, that's 3840 x 2160, consumer "4K" / or Ultra HD) when needed, and never mess with 5.4K, because it's just dumb. And 1080p shows fine on just about every display in the world now.

As to salvaging the footage, frankly I'd have to see the original file it to know what went wrong. Generally speaking, there's no fixing bad footage to make it good footage. If you're using editing software, most of those will take a file and convert it to whatever the project is set for. For example, if you set your editor for a 720P project, and you throw some 4K at it, the editor will convert the file for you, usually in the background, but it may or may not be a graceful conversion. If you're looking at your footage via editing software, there could be conversion going on you may not know about. There are ways to sharpen up a blurry image, but they're only band-aid level fixes, not crap-to-great converters.

The other thing to be aware of, the Air 2S can shoot two different color profiles. One is just standard, nothing special. The other is D-Log, which to say it simply, produces an intentionally low contrast and weak color image that is easier to adjust later, though "easier" is a bit relative. It means a standard color video does't have as much adjustment "room" but also doesn't always need it, where D-Log has lots more adjustment "room", but requires it every time. Adjusting color like that is called "Color Grading", and is somewhat simplified by using a bit of code called a LUT, which is simply a pre-set correction for a specific camera, saving you some (but not all) of the grading task.

It gets complex quickly. You'll find the color profile choice in the app under "Gimbal and Camera". Choose the standard setting for now, learn color grading later.

Lastly, you have to understand what "digital zoom" means. The camera only has so many pixels with which to create an image, and the lens focal length is fixed. To perform a digital zoom, two things must happen. Assuming 4K as an example, it first crops the image, taking only the center of the entire 4K sensor's data, which results in an image with smaller than 4K dimensions. Next, it blows that portion back up to 4K, but it doesn't create new image information to do that, it just "interpolates" between the valid pixels it has. It's the equivalent of looking at a printed picture under a magnifying glass. It's bigger, but softer. Digital zoom is not worth much because of that. In fact, you can do exactly the same thin in post in an editor, only under better control, and possibly better results. I've tried digital zoom, hated the results, and gave up on it, and now lust for a Mavic 3.

It's a little confusing that your stuff looks Ok until the drone moves. If you have a short file that illustrates your problem, and could load it into Dropbox or something, I'd be glad to take a look at it.

One last question: did you buy this drone new or used? If used, could there be an ND filter on the camera you might not have noticed? An ND filter, (Neutral Density) is intended to cut down light intensity. They look dark gray or black in color. The camera responds buy slowing the shutter speed to get the correct exposure. A slower shutter will result in motion blur. Sometimes this is desirable, but not always. There are some specific applications, otherwise you can take it off (if its there).

 

[sbunting]

Yeah, well, 5.4K isn't really "better resolution". Here's why: To benefit from any video format your display device (TV or monitor) has to at least match if not exceed that resolution. Anybody got a 5.4K TV? Me neither. So what would you do with the extra pixels? Scaling, which means converting from one resolution to another, is really bad when you scale down between two resolutions that aren't evenly mathematically related. So, taking a cue from what happens in the pro world, the extra image area simply get cropped off. Pro Digital Cinemal 4K is actually 4096 x 2160, but consumer 4K (more correctly, "Ultra HD") is 3840 x 2160. A theatrical projector is also fully Digital Cinema, so it shows 4K just fine, but consumer 4K displays are smaller, and that's a really difficult thing to scale well, so they simply crop off the extra. (Who thought consumer 4K shouldn't be the same as Digital Cinema 4K? Therein lies a tale!) Turns out, it's not much of a change. But the lesson learned is: shoot for your display.

Frankly, when I was first learning to fly a drone, I shot exclusively 1080p because I was a long ways from getting anything great, 1080p it saved a ton of memory space, and was far easier to deal with later. Now I shoot 4K (in the DJI Air 2S world, that's 3840 x 2160, consumer "4K" / or Ultra HD) when needed, and never mess with 5.4K, because it's just dumb. And 1080p shows fine on just about every display in the world now.

As to salvaging the footage, frankly I'd have to see the original file it to know what went wrong. Generally speaking, there's no fixing bad footage to make it good footage. If you're using editing software, most of those will take a file and convert it to whatever the project is set for. For example, if you set your editor for a 720P project, and you throw some 4K at it, the editor will convert the file for you, usually in the background, but it may or may not be a graceful conversion. If you're looking at your footage via editing software, there could be conversion going on you may not know about. There are ways to sharpen up a blurry image, but they're only band-aid level fixes, not crap-to-great converters.

The other thing to be aware of, the Air 2S can shoot two different color profiles. One is just standard, nothing special. The other is D-Log, which to say it simply, produces an intentionally low contrast and weak color image that is easier to adjust later, though "easier" is a bit relative. It means a standard color video does't have as much adjustment "room" but also doesn't always need it, where D-Log has lots more adjustment "room", but requires it every time. Adjusting color like that is called "Color Grading", and is somewhat simplified by using a bit of code called a LUT, which is simply a pre-set correction for a specific camera, saving you some (but not all) of the grading task.

It gets complex quickly. You'll find the color profile choice in the app under "Gimbal and Camera". Choose the standard setting for now, learn color grading later.

Lastly, you have to understand what "digital zoom" means. The camera only has so many pixels with which to create an image, and the lens focal length is fixed. To perform a digital zoom, two things must happen. Assuming 4K as an example, it first crops the image, taking only the center of the entire 4K sensor's data, which results in an image with smaller than 4K dimensions. Next, it blows that portion back up to 4K, but it doesn't create new image information to do that, it just "interpolates" between the valid pixels it has. It's the equivalent of looking at a printed picture under a magnifying glass. It's bigger, but softer. Digital zoom is not worth much because of that. In fact, you can do exactly the same thin in post in an editor, only under better control, and possibly better results. I've tried digital zoom, hated the results, and gave up on it, and now lust for a Mavic 3.

It's a little confusing that your stuff looks Ok until the drone moves. If you have a short file that illustrates your problem, and could load it into Dropbox or something, I'd be glad to take a look at it.

One last question: did you buy this drone new or used? If used, could there be an ND filter on the camera you might not have noticed? An ND filter, (Neutral Density) is intended to cut down light intensity. They look dark gray or black in color. The camera responds buy slowing the shutter speed to get the correct exposure. A slower shutter will result in motion blur. Sometimes this is desirable, but not always. There are some specific applications, otherwise you can take it off (if its there).

Nicely explained…

 

[DougMcC]

An excellent, thoughtful response!

 

[Dave Maine]

The issue for me is a little more complicated. 95 % of the previous post is great, but resolution reduction can be very successful with any good non-linerar editor. Reducing 5 k to any smaller resolution should loose no quality, unlike upressing.

The first problem comes from the fact that many computers will not play 5 k smoothly, and without a machine that will, it is virtually impossible to see what you are doing.

With that problem solved, reducing involves choosing the algorithm that the software uses to perform the calculation to perform the reduction. A little trial and error experimentation will be required. Note that for this to work, the reduction must have the same aspect ratio. 16:9 is the most likely ratio to use as that matches the DJI video as it comes out of the camera.

A big advantage with using 5 k images, is that the additional area allows you to crop to a lessor size without any quality loss in the result unless the crop is extreme.

 

[Arnold LeVine]

Zeke, can you attach a 10 sec clip here for analysis?

 

[vindibona1]

I loved the comment that @jphoto posted above.
In short, there is no need for overkill when your monitor or the end viewer's equipment can only represent so much resolution. That's one reason why I'm sticking with the Mavic 2 Pro rather than jumping up to a Mavic 3 or one of the Autel drones.

I also agree that there isn't a good explanation why consumer 4k is truncated on the long dimension. Another good point is, that unless you're color grading for effect or really have a high contrast situation where HDR is necessary, the standard record mode (in 90%+ of situations) the normal record mode is going to be good enough, especially for someone who has limited editing skills, and presumably less experience with color correction and color grading. In D-Log and D-Cinelike at least color/contrast correction is required. Unless you're color grading for mood and effect, it's just easier, and often better (for some), just to shoot in normal mode.

But to address the OP's issue of motion being "unsharp".... Is it possible that when shooting in 4k one has to make sure they have an SD card capable of faster write speeds? Logic would have it that if you have more data coming through per second then the card has to be capable, no?

 

[moldorf]

Yeah, well, 5.4K isn't really "better resolution". Here's why: To benefit from any video format your display device (TV or monitor) has to at least match if not exceed that resolution. Anybody got a 5.4K TV? Me neither. So what would you do with the extra pixels? Scaling, which means converting from one resolution to another, is really bad when you scale down between two resolutions that aren't evenly mathematically related. So, taking a cue from what happens in the pro world, the extra image area simply get cropped off. Pro Digital Cinemal 4K is actually 4096 x 2160, but consumer 4K (more correctly, "Ultra HD") is 3840 x 2160. A theatrical projector is also fully Digital Cinema, so it shows 4K just fine, but consumer 4K displays are smaller, and that's a really difficult thing to scale well, so they simply crop off the extra. (Who thought consumer 4K shouldn't be the same as Digital Cinema 4K? Therein lies a tale!) Turns out, it's not much of a change. But the lesson learned is: shoot for your display.

Frankly, when I was first learning to fly a drone, I shot exclusively 1080p because I was a long ways from getting anything great, 1080p it saved a ton of memory space, and was far easier to deal with later. Now I shoot 4K (in the DJI Air 2S world, that's 3840 x 2160, consumer "4K" / or Ultra HD) when needed, and never mess with 5.4K, because it's just dumb. And 1080p shows fine on just about every display in the world now.

As to salvaging the footage, frankly I'd have to see the original file it to know what went wrong. Generally speaking, there's no fixing bad footage to make it good footage. If you're using editing software, most of those will take a file and convert it to whatever the project is set for. For example, if you set your editor for a 720P project, and you throw some 4K at it, the editor will convert the file for you, usually in the background, but it may or may not be a graceful conversion. If you're looking at your footage via editing software, there could be conversion going on you may not know about. There are ways to sharpen up a blurry image, but they're only band-aid level fixes, not crap-to-great converters.

The other thing to be aware of, the Air 2S can shoot two different color profiles. One is just standard, nothing special. The other is D-Log, which to say it simply, produces an intentionally low contrast and weak color image that is easier to adjust later, though "easier" is a bit relative. It means a standard color video does't have as much adjustment "room" but also doesn't always need it, where D-Log has lots more adjustment "room", but requires it every time. Adjusting color like that is called "Color Grading", and is somewhat simplified by using a bit of code called a LUT, which is simply a pre-set correction for a specific camera, saving you some (but not all) of the grading task.

It gets complex quickly. You'll find the color profile choice in the app under "Gimbal and Camera". Choose the standard setting for now, learn color grading later.

Lastly, you have to understand what "digital zoom" means. The camera only has so many pixels with which to create an image, and the lens focal length is fixed. To perform a digital zoom, two things must happen. Assuming 4K as an example, it first crops the image, taking only the center of the entire 4K sensor's data, which results in an image with smaller than 4K dimensions. Next, it blows that portion back up to 4K, but it doesn't create new image information to do that, it just "interpolates" between the valid pixels it has. It's the equivalent of looking at a printed picture under a magnifying glass. It's bigger, but softer. Digital zoom is not worth much because of that. In fact, you can do exactly the same thin in post in an editor, only under better control, and possibly better results. I've tried digital zoom, hated the results, and gave up on it, and now lust for a Mavic 3.

It's a little confusing that your stuff looks Ok until the drone moves. If you have a short file that illustrates your problem, and could load it into Dropbox or something, I'd be glad to take a look at it.

One last question: did you buy this drone new or used? If used, could there be an ND filter on the camera you might not have noticed? An ND filter, (Neutral Density) is intended to cut down light intensity. They look dark gray or black in color. The camera responds buy slowing the shutter speed to get the correct exposure. A slower shutter will result in motion blur. Sometimes this is desirable, but not always. There are some specific applications, otherwise you can take it off (if its there).

that reply was more informative than 100 youtube videos talking about the same things. Excellent work jphoto

your next task is to make a 'DaVinci Resolve For Dummies' post....

 

[Zeke]

Yeah, well, 5.4K isn't really "better resolution". Here's why: To benefit from any video format your display device (TV or monitor) has to at least match if not exceed that resolution. Anybody got a 5.4K TV? Me neither. So what would you do with the extra pixels? Scaling, which means converting from one resolution to another, is really bad when you scale down between two resolutions that aren't evenly mathematically related. So, taking a cue from what happens in the pro world, the extra image area simply get cropped off. Pro Digital Cinemal 4K is actually 4096 x 2160, but consumer 4K (more correctly, "Ultra HD") is 3840 x 2160. A theatrical projector is also fully Digital Cinema, so it shows 4K just fine, but consumer 4K displays are smaller, and that's a really difficult thing to scale well, so they simply crop off the extra. (Who thought consumer 4K shouldn't be the same as Digital Cinema 4K? Therein lies a tale!) Turns out, it's not much of a change. But the lesson learned is: shoot for your display.

Frankly, when I was first learning to fly a drone, I shot exclusively 1080p because I was a long ways from getting anything great, 1080p it saved a ton of memory space, and was far easier to deal with later. Now I shoot 4K (in the DJI Air 2S world, that's 3840 x 2160, consumer "4K" / or Ultra HD) when needed, and never mess with 5.4K, because it's just dumb. And 1080p shows fine on just about every display in the world now.

As to salvaging the footage, frankly I'd have to see the original file it to know what went wrong. Generally speaking, there's no fixing bad footage to make it good footage. If you're using editing software, most of those will take a file and convert it to whatever the project is set for. For example, if you set your editor for a 720P project, and you throw some 4K at it, the editor will convert the file for you, usually in the background, but it may or may not be a graceful conversion. If you're looking at your footage via editing software, there could be conversion going on you may not know about. There are ways to sharpen up a blurry image, but they're only band-aid level fixes, not crap-to-great converters.

The other thing to be aware of, the Air 2S can shoot two different color profiles. One is just standard, nothing special. The other is D-Log, which to say it simply, produces an intentionally low contrast and weak color image that is easier to adjust later, though "easier" is a bit relative. It means a standard color video does't have as much adjustment "room" but also doesn't always need it, where D-Log has lots more adjustment "room", but requires it every time. Adjusting color like that is called "Color Grading", and is somewhat simplified by using a bit of code called a LUT, which is simply a pre-set correction for a specific camera, saving you some (but not all) of the grading task.

It gets complex quickly. You'll find the color profile choice in the app under "Gimbal and Camera". Choose the standard setting for now, learn color grading later.

Lastly, you have to understand what "digital zoom" means. The camera only has so many pixels with which to create an image, and the lens focal length is fixed. To perform a digital zoom, two things must happen. Assuming 4K as an example, it first crops the image, taking only the center of the entire 4K sensor's data, which results in an image with smaller than 4K dimensions. Next, it blows that portion back up to 4K, but it doesn't create new image information to do that, it just "interpolates" between the valid pixels it has. It's the equivalent of looking at a printed picture under a magnifying glass. It's bigger, but softer. Digital zoom is not worth much because of that. In fact, you can do exactly the same thin in post in an editor, only under better control, and possibly better results. I've tried digital zoom, hated the results, and gave up on it, and now lust for a Mavic 3.

It's a little confusing that your stuff looks Ok until the drone moves. If you have a short file that illustrates your problem, and could load it into Dropbox or something, I'd be glad to take a look at it.

One last question: did you buy this drone new or used? If used, could there be an ND filter on the camera you might not have noticed? An ND filter, (Neutral Density) is intended to cut down light intensity. They look dark gray or black in color. The camera responds buy slowing the shutter speed to get the correct exposure. A slower shutter will result in motion blur. Sometimes this is desirable, but not always. There are some specific applications, otherwise you can take it off (if its there).

jphoto,

A MILLION thank you's for that incredibly helpful tutorial! I hope the mods see it and pin it somewhere for future reference! And many thanks to all who contributed to this thread, lots of great information.

So the mystery might be clearing up a bit. I'm using Filmora for the video editing mainly because that's what I started with and it seemed to be the most intuitive for me. When I shoot in 4k I have no problems at all with the finished project (my limited knowledge notwithstanding) but the 5.4k for some reason is problematic. I cut a couple of clips of the "raw" drone footage and then the exact same "finished" footage from Filmora, which looks like crap. I even viewed it on multiple 32" monitors to rule out the monitor, and it looks the same no matter where it's played (crappy). I loaded them to YouTube, the links are below. I'd love to hear what you all think the problem might be, or if it's just the limitation of cheap software.

Thank you all again for the helpful responses!

Drone clip:

Filmora clip:

 

[Grandpa]

I loved the comment that @jphoto posted above.
In short, there is no need for overkill when your monitor or the end viewer's equipment can only represent so much resolution.

I would just point out that recording video at a higher resolution than necessary is helpful if you need to edit in a zoom in a clip during post.

 

[FLDave]

The both show up as 720p on YouTube, so it is unclear what you actually did from the original 5.4K. What was the frame rate?

 

[Zeke]

The both show up as 720p on YouTube, so it is unclear what you actually did from the original 5.4K. What was the frame rate?

I have no idea, whatever the default goes to. Watch the ground as the clip plays... the Filmora footage looks like it's covered with bugs moving around.

 

[Ty Pilot]

I also agree that there isn't a good explanation why consumer 4k is truncated on the long dimension.

A couple points; a discussion on resolutions is incomplete without understanding aspect ratio's, and how, and why they got to where they are today. Simply put; 16 x 9 is the most widely used format in consumer drones because the majority of devices that video from said drones will be played on is - 16 x 9. 3840 x 2160 as well as 1920 x 1080 are both 16 x 9 aspect ratios. I'll link to a great article about resolutions below.

Also, shooting in the highest resolution possible has a couple advantages - it future-proofs footage because the resolution train is not stopping and in the case of uploading, sites like Youtube, they give priority and better compression to higher resolution uploads. A typical video uploaded to Youtube will receive a compression that is more aggressive than it will to a video uploaded at 4K. However, when watching that same 4K video in a 1080 frame/player it will have better resolution than the 1080 only video. Videos with high frame rates also are give the better compression. To see this 'right click' on any youtube video and select 'stats for nerds' and notice the Codec - VP09 = , AVC1 =
See Bottom Pic.

Finally, D-log footage is not just washed out, it is 10 bit. H264 footage in most DJI drones is 8 bit, meaning it is capable of producing 256 colors per channel while 10 bit produces 1,024 colors per channel. Meaning: 8 bit video can only produce 16.7 million colors - 10 bit D-log produces 1.07 billion colors. 8 bit video is given a profile in terms of color and sharpness in it's compression, D-log does not because of the aforementioned massive amount of colors and that is why log looks flat. Think of it like a pie - 8 bit video is baked - right out of the camera, while 10 bit is like all the ingredients you could possibly use to make a pie and you bake it your self.


CLICK PICTURE FOR ARTICLE.

 

[FLDave]

I agree, I looks but, but, more importantly, what did you do to get it to that point? YouTube will default to your upload resolution up to 4K. What were you export settings in Filmora?

 

[Zeke]

I agree, I looks but, but, more importantly, what did you do to get it to that point? YouTube will default to your upload resolution up to 4K. What were you export settings in Filmora?

Here's what Filmora tells me (most of it's Greek to me).

 

[FLDave]

You exported at 720P, not 1080p, and not 4K, and not 5.4K
it's difficult to figure since we don't the all the details of the original files.
A program like IrfanView could give you details about the properties of the original files.

 

[Zeke]

I agree, I looks but, but, more importantly, what did you do to get it to that point? YouTube will default to your upload resolution up to 4K. What were you export settings in Filmora?

So fooling around a little with the project settings and I discovered that if I change the resolution of the clip to DCI 4k (whatever that is) the clip renders much more smoothly!

 

[jphoto]

So fooling around a little with the project settings and I discovered that if I change the resolution of the clip to DCI 4k (whatever that is) the clip renders much more smoothly!

No, don't do that either.

Remember what I said about "shooting for your display"? "DCI 4K" means "Digital Cinema Initiative", in other words, theatrical 4K, 4096 x 2160. There are NO consumer displays that show that without scaling, except possibly some high end projectors. When you shoot 5.4K, it will always be scaled somewhere unles you can teach Filmora to export at an odd resoultion...but then, you can't display it without scaling anyway!

Shoot at 4K, and it won't have to be scaled if you output in UHD, 3840 x 2160, which is exactly what your Air 2S shoots. No scaling done. Forget zooming in post for now, we'll come back to that if need be.

Now, what to do about that 5.4K footage. I'm not a Filmora editor, so there are a few things about how it works that I don't know. But the idea is, for your 5.4K clips to be in a 4K project without scaling, rather than match to the exact frame size, they would be zoomed in, essentially a crop so the result fills the screen, and every pixel on screen lines up perfectly with 3840x2160. So for the numbers, you want 70.17544% of your 5.4K image to fill your 4K screen, which would result in no scaling (only cropping) done to the 5.4K image. Then, output your project at 3840 x 2160. I'm not positive of which tool to use, but it looks like there is a "scale" function, which oddly, might end up doing what you want. The confusing part here is that it's not clear if the "scale" function references the original size of the clip, or the rendered clip size, or what. You'll have to play with it to figure it out. And it may not matter a lot, once you get the next bit (below) right.

From your samples, what we already know is, Filmora doesn't scale very well. Your videos went to YouTube at 720p, so everything, even the 3840x2160 4K stuff, was beat up pretty badly. And the 5.4K stuff was obviously worse. Scaling 4K to 720p should have been easy, but the results were bad. You already figured out that if you output in DCI 4K, it's much better. Now, output everything to 3840x2160, and see what you get. And send those clips to YouTube for us, if you would.

 

[jphoto]

A couple points; a discussion on resolutions is incomplete without understanding aspect ratio's, and how, and why they got to where they are today. Simply put; 16 x 9 is the most widely used format in consumer drones because the majority of devices that video from said drones will be played on is - 16 x 9. 3840 x 2160 as well as 1920 x 1080 are both 16 x 9 aspect ratios. I'll link to a great article about resolutions below.

Yup. "Shoot for your display". Worthwhile article. What a confused mess up world we live in.

Also, shooting in the highest resolution possible has a couple advantages - it future-proofs footage because the resolution train is not stopping

I tend to apply practicality to this. Yeah, the resolution keeps going up in numbers, but we've long passed the practical limits in the 4K > 8K hop when visual acuity is considered. Human eyes with 20/20 vision can resolve details that occupy as little as 1/60 of a degree of arc. Using that bit of information, you can determine the maximum distance you can be from a particular screens size with a specific resolution before you begin to miss details. An eample of a 65" diagonal screen at 1080p: any farther than 8.5' and you can no longer see the finest details of 1080p. Now we go to 4K, and the maximum distance gets much, much shorter - 4.3'. Past that, and you simply can't see the potential details of 4K anymore. 8K? Gald you asked: if you sit farther than 2' from that 65" screen, your eyes can't pick up the finest possible detail. And I don't know anyone that sits even 4' from a 65" screen.

Note I also said "potential details" and "possible detail", because the entire analysis is assuming detail comprised of radically different adjacent pixels (line pairs), which only are possible with a digitally generated image. It just isn't what actual video is, because the chunk of glass in the front of the camera messes it up, particularly so in drone cameras. A really good, really expensive 35mm lens is right on the edge of not being able to produce a true 4K resolution image at the sensor (and that's a full-frame sensor), and when you move away from the ideal aperture or off to the edge, it's more like 3.5K. So, you ain't getting 4K anyway.

So the question becomes...is it really important to shoot at the highest resolution possible to future-proof the footage? The answer, when considering visual acuity, is a firm no. The answer, when considering the footage end usage is also probably no, unless you're shooting for an IMAX release. Meanwhile, high resolution causes a whole chain of problems in production. Your choice, as always. But also, consider that resolution is not the thing to obscess over. It's not the big factor in determining image qualty anymore. Cinematographers already know we have more than enough resolution, and they've been using image-softening filters for a century to deliberately go the other way. Here's another perspective.

and in the case of uploading, sites like Youtube, they give priority and better compression to higher resolution uploads. A typical video uploaded to Youtube will receive a compression that is more aggressive than it will to a video uploaded at 4K. However, when watching that same 4K video in a 1080 frame/player it will have better resolution than the 1080 only video. Videos with high frame rates also are give the better compression. To see this 'right click' on any youtube video and select 'stats for nerds' and notice the Codec - VP09 = , AVC1 =

I'm not doubting you here, but I also couldn't find a reference that backed this up. What I did find is that the codec YouTube applies is related to the number of views also.

Finally, D-log footage is not just washed out, it is 10 bit. H264 footage in most DJI drones is 8 bit, meaning it is capable of producing 256 colors per channel while 10 bit produces 1,024 colors per channel. Meaning: 8 bit video can only produce 16.7 million colors - 10 bit D-log produces 1.07 billion colors. 8 bit video is given a profile in terms of color and sharpness in it's compression, D-log does not because of the aforementioned massive amount of colors and that is why log looks flat. Think of it like a pie - 8 bit video is baked - right out of the camera, while 10 bit is like all the ingredients you could possibly use to make a pie and you bake it your self.

I knew someone would jump on this. I was trying to really, really simplify the whole color grading thing for the novice drone video maker. This should be a separate thread.

8 bit color is adequate for someone learning, and comes out Ok most of the time. 10 bit video does two things: it allows more adjustment, and requires adjustment. You can't use it without adjustment, our displays and most paths to them are all 8 bit. And yet, there actually IS some adjustment range on that lousy 8 bit video! A surprising amount. And yeah, there's more with 10 bits.

 

[Mikieh]

I loved the comment that @jphoto posted above.
In short, there is no need for overkill when your monitor or the end viewer's equipment can only represent so much resolution. That's one reason why I'm sticking with the Mavic 2 Pro rather than jumping up to a Mavic 3 or one of the Autel drones.

I also agree that there isn't a good explanation why consumer 4k is truncated on the long dimension. Another good point is, that unless you're color grading for effect or really have a high contrast situation where HDR is necessary, the standard record mode (in 90%+ of situations) the normal record mode is going to be good enough, especially for someone who has limited editing skills, and presumably less experience with color correction and color grading. In D-Log and D-Cinelike at least color/contrast correction is required. Unless you're color grading for mood and effect, it's just easier, and often better (for some), just to shoot in normal mode.

But to address the OP's issue of motion being "unsharp".... Is it possible that when shooting in 4k one has to make sure they have an SD card capable of faster write speeds? Logic would have it that if you have more data coming through per second then the card has to be capable, no?

I ditched my pro2 for the zooming capabilities of the air2s. I loved the pro, but zooming in 4.2x onto an object or landscape in real time is very entertaining. I used my 2 pro for real estate jobs, but the Air2s has me covered with it’s one inch sensor. It took me a while to bite on the change, but I eventually did it!